Winn L. Rosch - Hardware Bible

WINN L.
ROSCH
HARDWARE BIBLE,
SIXTH EDITION
ii
Winn L. Rosch Hardware Bible,

Sixth Edition
Associate Publisher
Copyright 2003 by Que Publishing
Acquisitions Editor
All rights reserved. No part of this book shall be reproduced, stored

in a retrieval system, or transmitted by any means, electronic,
mechanical, photocopying, recording, or otherwise, without written
permission from the publisher. No patent liability is assumed with
respect to the use of the information contained herein. Although
every precaution has been taken in the preparation of this book, the
publisher and author assume no responsibility for errors or omissions. Nor is any liability assumed for damages resulting from the
use of the information contained herein.
Angelina Ward
International Standard Book Number: 0-7897-2859-1

Library of Congress Catalog Card Number: 2002095010
Printed in the United States of America
First Printing: March 2003
06
05
04
03
Trademarks
All terms mentioned in this book that are known to be trademarks or
service marks have been appropriately capitalized. Que Publishing
cannot attest to the accuracy of this information. Use of a term in
this book should not be regarded as affecting the validity of any
trademark or service mark.
Greg Wiegand
Development Editor
Howard Jones
Managing Editor
Charlotte Clapp
Project Editors
Carol Bowers
Tonya Simpson
Copy Editor
Bart Reed
Indexer
Ginny Bess
Proofreader
Kevin Ober
Technical Editors
Rima Regas
Tyler Regas
Team Coordinator
Warning and Disclaimer

Every effort has been made to make this book as complete and as
accurate as possible, but no warranty or fitness is implied. The information provided is on an as is basis. The author and the publisher
shall have neither liability nor responsibility to any person or entity
with respect to any loss or damages arising from the information
contained in this book.
Sharry Lee Gregory
Interior Designer
Anne Jones
Cover Designer
Anne Jones
Page Layout
Stacey Richwine-DeRome
iii
Contents at a Glance
Foreword
Part 1
xv
19
Introduction
1 Computers
Part 5
Human Interface
Principles:
Connecting to You
693
20
Keyboards
725
Functions and Components 17
21
Pointing Devices
753
Software
22
Scanners
771
23
Digital Cameras
787
Part 2
33
Processing
Digital Electronics
61
24
Display Systems
813
Microprocesssors
83
25
Audio Systems
883
Chipsets
153
26
Printers
923
BIOS
175
27
Fax
957
Part 3
Part 6:
Communications
Infrastructure
Channels and Coding
225
28
Principles: Hardware
969
Expansion Buses
261
29
Motherboards
983
10
Interfaces
299
30
Expansion Boards
997
11
Ports
353
31
Power
1011
12 Local Area Networking
417
32
Cases
1061
13
Telecommunications
449
Index
1077
14
Internet
479
Part 4
Storage
15 Principles: Freezing Time
511
16
Memory
547
17
Magnetic Storage
583
18
Optical Storage
(CD and DVD)
643
iv
Table of Contents
1 Introduction
1 Computers
Computers in History ..........................5

Computers in Todays World................7
Computer Intelligence..........................9
So What Is a Computer? ....................10
Electronics ................................11
Digital ......................................12
Logic ........................................13
Programmability ......................14
Calculating and
Decision-Making....................14
2 Functions and Components
17
Processing............................................19
Microprocessor ........................20
Chipset ......................................20
BIOS..........................................21
Communicating ..................................21
Expansion Buses........................21
Interfaces ..................................22
Input/Output Ports ..................22
Local Area Networks................23
Telecommunications ................23
Internet ....................................23
Remembering......................................24
Memory ....................................24
Hard Disk Drives......................24
Floppy Disk Drives ..................25
Optical Drives ..........................25
Tape Drives ..............................26
Listening and Monitoring ..................26
Keyboard ..................................27
Pointing Devices ......................27
Scanners ....................................27
Digital Cameras ........................28
Display Systems ........................28
Audio ........................................29
Printers......................................29
Fax ............................................29
Infrastructure ......................................30
Motherboards ..........................30
Expansion Boards ....................30
Power ........................................30
Cases..........................................31
3 Software
33
Definitions ..........................................34
Applications ........................................35
Suites ........................................37
Utilities......................................37
Applets ......................................38
Operating Systems ..............................39
Windows 9X ............................40
Windows NT............................41
Unix ..........................................41
Linux ........................................42
Programming Languages....................43
Machine Language ..................44
Assembly Language ..................45
High-Level Languages ............45
Hardware Interfaces............................49
Control Hierarchy ....................50
Application Program
Interface..................................51
Device Drivers ..........................52
BIOS..........................................53
Device Interfaces ......................54
vi
WINN L. ROSCH HARDWARE BIBLE, SIXTH EDITION
2 Processing
4 Digital Electronics
59
61
Electronic Design ..............................62

Digital Logic Circuitry ............62
Electronics ................................64
Moores Law..............................69
Printed Circuits ........................70
Logic Design ......................................74
Boolean Logic ..........................74
Logic Gates ..............................75
Memory ....................................76
Instructions ..............................76
Registers....................................77
Programs ..................................79
Interrupts ..................................80
Clocked Logic ..........................80
5 Microprocessors
83
Background ........................................84
Input/Output Unit....................85
Control Unit ............................86
Arithmetic/Logic Unit ............86
Floating-Point Unit..................87
Instruction Sets ........................88
Microcode ................................88
RISC..........................................90
Micro-Ops ................................91
Single Instruction,
Multiple Data ........................92
Very Long
Instruction Words ..................93
Performance-Enhancing
Architectures ....................................93
Clock Speed ..............................94
Pipelining ..................................95
Branch Prediction ....................96
Speculative Execution ..............96
Superscalar Architectures ........96
Out-of-Order Execution ..........97
Register Renaming ..................98

Explicitly Parallel
Instruction Computing ..........98
Front-Side Bus..........................99
Translation Look-Aside
Buffers ..................................100
Caching ..................................100
Electrical Characteristics ..................103
Thermal Constraints ..............103
Operating Voltages ................104
Power Management................106
Packaging ................................109
History ..............................................114
Initial Development................115
Personal Computer
Influence ..............................116
Age of Refinement..................118
Mature Design ........................121
Current Products ..............................128
Intel Microprocessors ............128
Advanced Micro Devices
Microprocessors ..................143
Transmeta Microprocessors ..149
6 Chipsets
153
Background ......................................154
Functions ..........................................155
Memory Control ..............................156
Addressing ..............................157
Refreshing ..............................158
Error Handling ......................158
System Control ................................159
Timing Circuits ......................159
Oscillators ..............................160
System Clocks ........................161
Timers ....................................164
Interrupt Control....................165
Direct Memory Access ..........169
Power Management................171
CONTENTS
Peripheral Control............................172
Hard Disk Interface................172
Floppy Disk Controller ..........173
Keyboard Controller ..............173
Input/Output Ports ................173
7 BIOS
175
Background ......................................176
Firmware ................................176
Functions ................................177
Initialization Functions ....................179
Power-On Self Test ................180
BIOS Extensions ....................182
Initial Program Load ..............183
Plug-and-Play ........................184
Interface Functions ..........................188
Compatibility ..........................190
Software Interrupts ................191
Parameter Passing ..................195
Entry Points ............................196
Linking to Hardware ..............197
Storage Functions ............................199
Legacy Data ............................199
System Configuration ............203
Tuning for Performance ........214
Upgrades ..........................................220
3 Communications 223
8 Channels and Coding
225
Signal Characteristics........................226
Channels ................................226
Bandwidth ..............................227
Modulation..............................229
Parallel Channels ....................236
Ports ........................................237
Buses........................................239
Serial Channels ......................246
Bit-Stuffing ............................250
Group-Coding ........................251
9 Expansion Buses
261
Background ......................................262
Peripheral Component
Interconnect....................................264
Architectural Overview ..........264
Data Integrity Signals ............267
Serialized Interrupts ..............268
Bus-Mastering and
Arbitration ............................270
Low-Voltage Evolution ..........270
Slot Limits ..............................270
Hot-Plugging..........................271
Setup........................................272
PCI-X......................................272
PCI Express ............................274
Standards and
Coordination ........................276
InfiniBand..........................................276
History ....................................277
Communications ....................278
Structure ................................279
Standards and
Coordination ........................281
HyperTransport ................................281
History ....................................282
Performance............................282
Structure ................................283
Standards and
Coordination ........................285
PCMCIA ..........................................285
PC Card ..................................287
CardBus ..................................293
Zoomed Video ........................295
Standards and
Coordination ........................295
10 Interfaces
299
AT Attachment..................................300
Performance............................301
Addressing and Capacity ........304
Logical Interface ....................312
vii
viii
Signals and Operation ............313

Compatibility ..........................317
Wiring ....................................318
Sanctioning Organization ......320
Serial ATA ........................................320
SCSI Parallel Interface ....................322
Standards ................................322
Clocking ..................................324
Bus Width ..............................326
Capacity and Addressing ........326
Signaling ................................327
Wiring ....................................328
Cables......................................336
Connectors..............................337
Mixing SCSI Standards ..........341
Operation and Arbitration......343
Sanctioning Organization ......344
Fibre Channel ..................................345
Floppy Disk Interface ......................346
Controller ..............................347
Operation ................................347
Cabling....................................348
11 Ports
353
Universal Serial Bus ..........................354

Background ............................355
Connectors..............................357
Cable ......................................358
Protocol ..................................359
Standard ..................................361
FireWire ............................................361
Overview ................................362
Signaling ................................363
Configuration..........................364
Arbitration ..............................364
Connectors..............................365
Cabling....................................366
IrDA ..................................................367
History ....................................367
Overview ................................368
Bluetooth ..........................................371
History ....................................372
Overview ................................373
Software Side ..........................374
Hardware Side ........................375
Components............................378
Standards and
Coordination ........................378
RS-232C Serial Ports........................378
Electrical Operation ..............379
Connectors..............................380
Signals ....................................382
Cables......................................385
UARTs ....................................388
Register Function ..................389
Parallel Ports ....................................390
History ....................................391
Connectors..............................395
Adapters ..................................398
Cable ......................................398
Electrical Operation ..............399
Control....................................410
12 Local Area Networking
417
Concepts............................................418
OSI Model ..............................419
Internet Model........................422
Windows Model ....................422
Architecture ......................................424
Topologies ..............................424
Hierarchies..............................426
Standards ..........................................428
Ethernet ..................................428
Wireless ..................................431
HomePNA ..............................436
HomePlug ..............................438
CONTENTS
Hardware ..........................................441
Network Interface Cards........442
Hubs and Access Points..........443
Routers ....................................445
Cabling....................................445
13 Telecommunications
449
Analog Services ................................450

Background ............................450
Modem Technology................451
Analog Standards ....................458
Modem Hardware ..................462
Digital Services ................................466
Telephone Services ................467
Cable Services ........................473
Satellite Services ....................476
14 Internet
479
History ..............................................481
Addressing ........................................482
IP Addresses............................482
Subnet Masks ..........................483
Network Classes ....................485
CIDR ......................................486
IPv6 ........................................487
What Address to Use..............487
Addresses You Cannot Use ....488
DHCP ....................................488
Domain Name System......................489
How DNS Works ..................499
Root Name Servers ................501
WINS......................................502
Routing..............................................502
Web Pages ........................................503
Service Providers ..............................504
Performance Limits ................504
Security ..................................505
4 Storage
509
15 Principles: Freezing Time
511
Defining Memory ............................511

Volatility..................................513
Measurement ..........................514
Solid-State Memory..........................519
Read-Write Memory ..............520
Read-Only Memory................523
Errors ................................................523
Causes......................................523
Detection and Prevention ......526
Mass Storage Systems ......................528
Access ......................................528
Data Organization ..................529
Magnetic Storage ..............................532
Magnetism ..............................533
Magnetic Materials ................534
Magnetic Storage....................534
Digital Magnetic Systems ......535
Saturation................................536
Coercivity................................537
Retentivity ..............................538
Optical Storage ................................539
Virtual Memory ................................541
Demand Paging ......................541
Swap Files ..............................542
Data Compression ............................543
16 Memory
547
Requirements ....................................548
Minimums ..............................548
System Resource Memory......549
Maximum Usable Memory ....550
Logical Organization ........................552
Real Mode Memory................552
Protected Mode Memory ......553
Frame Buffer Memory............554
Unified Memory
Architecture..........................554
Shadow Memory ....................554
ix
Technologies......................................555
Read-Only Memory................555
Mask ROM ............................556
Read-Write Memory ..............559
Packaging ..........................................567
Dual Inline Memory
Modules ................................568
Rambus Inline Memory
Modules ................................570
Small Outline DIMMs ..........571
Installation ........................................572
Memory Speed........................573
Mixing Modules......................573
Contact Material ....................574
Orientation..............................575
Memory Cards ..................................575
CompactFlash ........................576
SmartMedia ............................577
Memory Stick ........................578
Multimedia Cards ..................579
Secure Digital ........................579
xD Picture Card......................580
17 Magnetic Storage
583
Hard Disks ........................................584

History ....................................584
Mechanism ..............................585
Medium ..................................595
Geometry ................................598
Physical Format ......................600
File System..............................601
Performance Issues ................606
Drive Arrays............................609
Floppy Disks ....................................616
History ....................................617
Format ....................................622
SuperDisks ..............................625
Mechanisms ............................628
Tape ..................................................630
Technologies ..........................631
Formats ..................................634
18 Optical Storage (CD and DVD)
643
History ..............................................643
Medium ..................................645
Materials..................................648
Read-Only ..............................649
Recordable ..............................649
Eraseable/Rewritable ..............652
Physical Format ................................653
Drives ................................................654
Speed ......................................655
Access Time ............................658
Controls ..................................658
Compatibility ..........................659
Standards ..........................................659
CD-DA ..................................660
Super Audio CD ....................661
CD-ROM................................663
CD-Recordable ......................667
CD-Rewritable ......................677
DVD-ROM ............................678
DVD-Video ............................679
DVD-Audio ......................................682
Write-Once DVD Formats....684
Rewritable DVD Formats ......685
Blu-ray ....................................688
5 Human Interface 691

19 Principles: Connecting to You
693
Teletypes............................................693
Pointing Devices ....................695
Graphic Input Devices ..........696
Output ..............................................697
Teletype to Terminal ..............697
BIOS Support ........................698
Character Technologies..........699
Two-Dimensional Graphics ..705
Color Coding..........................710
Three-Dimensional
Graphics ..............................715
CONTENTS
20 Keyboards
725
Technologies......................................726
Contact....................................727
Non-Contact ..........................728
Touch ................................................729
Legends ............................................731
Layouts ..............................................732
QWERTY ..............................732
Dvorak-Dealey........................733
Control and Function Keys....735
Ergonomic Designs ................737
Electrical Function............................738
Keyboard Port ........................739
Software Commands ..............742
Host Interface ........................744
Compatibility ..........................746
Connections ............................747
USB Port ................................750
Wireless ............................................750
21 Pointing Devices
753
Mice ..................................................753
Mechanical Mice ....................754
Optical Mice ..........................755
Buttons ....................................757
Wheels ....................................758
Resolution ..............................758
Trackballs ..........................................759
Switches ..................................759
Ball Size ..................................760
Handedness ............................760
Resolution ..............................761
TouchPads ........................................761
TrackPoint ........................................763
Game Controllers ............................764
Interfaces ..........................................765
Mouse Port ............................765
USB Port ................................766
Serial Mice ..............................766
Wireless ..................................766
Protocols and Drivers ......................767
22 Scanners
771
Types..................................................772
Drum Scanners ......................772
Flatbed Scanners ....................773
Hand Scanners........................773
Video Scanners ......................775
Photo Scanners ......................775
Slide Scanners ........................775
Features ............................................776
Scanning Speed ......................776
Dynamic Range ......................777
D-max......................................778
Resolution ..............................779
Transparency Adapters ..........780
Optical Character
Recognition ..........................780
Sheet Feeders..........................781
Electrical Interfacing ..............782
Application Interfacing ....................782
23 Digital Cameras
787
The Big Three ..................................788

Still Cameras ..........................788
Video Cameras........................788
Web Cameras..........................789
Imaging Sensors ................................789
Size ..........................................790
Resolution ..............................790
Sensitivity ................................791
Lenses ................................................791
Aperture ..................................791
Focal Length ..........................793
Zoom ................................................794
Optical Zooming ....................794
Digital Zooming ....................795
Zoom Range ..........................795
Supplemental Lenses ..............795
Shutters..............................................796
Shutter Speeds ........................796
Shutter Delay..........................797
xi
xii
Exposure............................................798
Sensing ....................................798
Control....................................799
Automatic................................800
Flash ..................................................801
Focusing ............................................802
Fixed Focus ............................803
Manual Focus..........................803
Auto-Focus..............................803
White Balance ..................................804
Image Storage ..................................805
Viewfinders........................................806
View Panels ......................................807
Single Image ..........................807
Multiple Image ......................807
Slideshow ................................807
Zoom ......................................807
Connections ......................................808
Analog Video Output........................809
Power ................................................809
Batteries ..................................809
AC Power................................810
24 Display Systems
813
Video Technology ............................814

Scanning..................................814
Synchronizing Signals ............815
Retrace ....................................816
Blanking ..................................817
Front and Back Porches ........817
Vertical Interval ......................817
Video Integration ..............................818
Standards ................................818
Video Overlay ........................821
Video Capture ........................822
Image Compression ..........................822
Filters and Codecs ..................823
JPEG ......................................823
MPEG ....................................824
Computer Display Standards ..........826

Display Adapters ..............................827
Accelerator Chips ..................828
Video Controllers ..................831
RAMDACs..............................832
Memory ..................................832
Displays ............................................833
Background ............................833
Cathode Ray Technology ......834
Flat-Panel Display
Technology ..........................852
Electronics ..............................858
Interfaces ................................869
Monitor Interfacing and
Connectors ..........................874
25 Audio Systems
883
Sound ................................................884
Analog Audio ....................................885
Frequency................................885
Amplitude................................885
Decibels ..................................886
Distortion................................887
Digital Audio ....................................888
Sampling Rate ........................888
Resolution ..............................890
Bandwidth ..............................890
Transmission ..........................891
Compression ..........................891
Capturing Sounds ..................901
Hardware ..........................................904
Basic Sound System ................904
Sound Systems ........................908
Digital Audio Output
Systems ................................913
Transducers ............................915
26 Printers
923
Print Engines ....................................924

Impact Dot-Matrix ................925
Inkjets......................................927
CONTENTS
Phase-Change Inkjets ............931

Laser........................................931
Direct Thermal ......................935
Thermal Transfer....................936
Dye-Diffusion ........................937
Characteristics ..................................938
Speed ......................................938
Quality ....................................941
Color ......................................943
Paper-Handling ......................946
Consumables ....................................951
Cartridges................................952
Refilling ..................................953
Paper ......................................953
Interfacing ........................................954
27 Fax
957
Technology ........................................958
Hardware ..........................................959
Modems ..................................960
Multifunction Printers............960
Group Standards ..............................961
Analog ....................................961
Group 3 ..................................961
Group 4 ..................................964
Interface Classes................................964
6 Infrastructure
967
28 Principles: Hardware
969
Design Approaches ..........................969

Bus-Oriented Computers ......970
Single-Board Computers........971
Compromise Designs ............972
Blade Servers ..........................973
Nomenclature ..................................973
Expansion Boards..............................974
Nomenclature ........................974
Construction ..........................976
Power ................................................979
Cases..................................................979
29 Motherboards
983
Physical Requirements......................984
Bus Standard ..........................984
Slot Number ..........................985
Slot Spacing ............................985
Slot Layout..............................986
Standards ..........................................987
ATX ........................................988
Mini-ATX ..............................990
microATX ..............................990
FlexATX ..................................991
NLX ........................................992
WTX ......................................994
30 Expansion Boards
997
PCI Boards ........................................998

Dimensions ............................998
Connectors..............................999
Accelerated Graphics Port..............1000
Dimensions ..........................1000
Connector ............................1002
Riser Boards ....................................1002
Audio Modem Riser ............1004
Communications
Network Riser ....................1004
Advanced Communications
Riser....................................1005
PCMCIA Cards ..............................1006
Connector ............................1009
31 Power
1011
Power Supplies................................1012
Desktop Power Supplies ......1013
Portable Computer Power ..1021
Batteries ..........................................1024
Electrical Characteristics......1024
Chemical Systems ................1028
Standards ..............................1037
Battery Charging ..................1038
Smart Battery
Specifications......................1040
Battery Safety........................1041
xiii
Power Management ........................1042

Advanced Power
Management ......................1042
Advanced Configuration
and Power Interface ..........1045
Energy Star ..........................1048
Power Protection ............................1048
Power Line Irregularities ....1048
Over-Voltage Protection ......1051
Blackout Protection ..............1053
Phone and Network Line
Protection ..........................1058
32 Cases
1061
Mechanical Construction................1061
Standards ..............................1062
Size ........................................1063
Internal Access ......................1066
Bays ......................................1067
Slots ......................................1069
Cooling............................................1069
Passive Convection ..............1070
Active Cooling ......................1070
Fan Failure ............................1071
Radiation ........................................1072
Radio Frequency
Interference ........................1072
Minimizing Interference ......1073
Health Concerns ..................1074
Index
1077
Foreword
What a time for a book about computers! Machines are faster than ever, with some systems clocking in at more than three gigahertz. Yet, amazingly, they cost less than ever
before, too. You can buy a new personal computer for less than $300at least if you turn
over rocks for the best prices and look for a free operating system (which is sort of like
buying a car and curb-surfing for tires).
But theres another side to the story. Engineers shudder in corners hoping their jobs will
exist next week. High-tech manufacturers jump as high as they can in a not-always-successful effort to keep their nostrils above the rising waters of ruin. How can you make
sense of this out-of-kilter world?
One way is by reading the book you hold in your hand. Its designed to help you understand
all the technologies that underlie the modern personal computer and all its peripherals that
help make it useful. Although you wont be able to design a computer when youre finished
(that is, unless you already knew how to when you started), you will know what it takes to
design one, how computers work, and how you can work better with themand get them
to do more for you. It doesnt matter if you regard a screwdriver as a tool, an instrument of
torture, or a new curiosity youve never before encountered. Even if you have no technical
background whatsoever, this book will guide you to a level of understanding that will let you
buy, use, and upgrade your computer with confidence. At the same time, youll find all the
technical information that you need to do a school report or business presentation on nearly
any current computer topic.
If this book has any weakness, its that I dont know what the future holds. If I did, I might
still have a retirement planbut thats another story. What I knowand what this book
relatesis whats happening in the present.
To put it plainly, were in a time of change. No one seriously doubts that the entire computer
industry is changing radically. It has matured and shifted from growth mode to a sustaining
mode. Both businesses and homes are stuffed nearly to capacity with personal computers.
On the other hand, computers have become an essential part of our lives. Rather than
exotic and revered machines serving as centerpieces to hobbies and business plans, the
computer is something you use every day for chores as mundane as looking up an address,
typing a school report, or printing a picture of the kids. As much as computers are used
for business, theyre also used for funfor making movies or playing games that test the
limits of both your imagination and current technology.
But thats not the end of it. New technologies continue to sweep through the industry. Both
still and video cameras have gone digital. Wireless networking is sweeping through both
businesses and homes, along with some novel wired systems that use your existing telephone
or electrical wiring. Direct digital connections to the Internetstuff with names like digital
subscriber lines and cable modemsare now in the mainstream. The digital versatile disc is
in nearly everyones home, and your computer can now burn your own DVDs. The universal serial bus connection system has matured, and its latest incarnation, version 2.0, will
make it the choice for most computer peripherals. Speeds of everything, from those USB
ports to microprocessors and disk drives, is going in only one direction: up.
This new edition of the Hardware Bible covers all these new developments to help you
keep up with the changes in the personal computer industry. Indeed, the industry has been
dominated by such changesdramatic movements from one technology to another, from
one generation to the next. Each has marked a major stage in its evolution, and each has
brought a new, invigorating era of innovationand a new edition of this book.
One such change was the move between computer buses. When the last edition of the
Hardware Bible appeared, the computer industry was finally moving to throw off one of its
last links with its origins, now more than two decades old: Industry Standard Architecture
(ISA). The innovation of Peripheral Component Interconnect (PCI), with its improved performance and ease of configuration, made mainstream computers possible. Without it, for
example, you could never edit video on a desktop computer; burning your own CDs would
be chancy; and computer games would be stuck somewhere between Pong and PacMan.
We are at the very beginning of another period of change. Youll see the rising forces
inside this bookIntels new Itanium processor and AMDs response, the Opteron. The
Itanium breaks with the Intel architecture that makes Windows possible. It will require a
new Windows, and new software. The Opteron hedges its bets and gives the potential for
new 64-bit performance but with backward-looking compatibility that has been so essential at helping us across cusps.
In either case, the transition to 64-bit computing will have an effect that you can see. Edit
a snapshot, and big changes will happen fast, almost instantly. You wont grow old making
a home movie. Easy animation will be something anyone can do as easily as making a
business card from a template.
This new edition of the Hardware Bible looks forward to the next generation of computing
while retaining its past roots. The good news is that the only thing Ive eliminated from this
Sixth Edition is redundancy. Nothing has fallen on the cutting room floor. Achieving the
contradictory ends of including everything while trimming length is possible thanks to the
Web. Ive tried to exploit both print and the Internet to put each to work at what it does best.
This book is designed in conjunction with the Hardware Bible Web site at www.hardwarebible.
com. At the Web site, youll find additional materialin fact, more than is in the physical
bookto enhance the printed edition. In particular, youll find updates too recent to be
printed, as well as discussions of older technologies that are no longer relevant to todays computers, more detailed explanations of the operation of interfaces, and interesting things to
know that didnt quite mesh with the narrative of this book.
The selection process was based on making this book something you can readily read.
After all, what good is a book if you dont want to read it? To prevent the Web site from
interfering with your reading experience, were not going to flag every tiny instance where
you can find more details on the Web siteyoud get the reading equivalent of driving
down a rutted road too fast with doubtful shock absorbers. In other words, this book is
designed to be a standalone book; you can read it without reference to the Web.
Go to the Web, and it becomes better. You get more. But you can select what you want,
when you want. The Web site is keyed to individual chapters, but an overall slick navigation system will help you target what you want. Later in this foreword Ive included a listing of the supplementary files posted at the time the printed book was published (more
than 140 files), along with the chapters they are meant to accompany. As new technologies
develop in the computer industry, the Web site will add them.
To preserve the reference value of the Hardware Bible, the Web site also includes a reference to all the popular industry connectors and the signals assigned to each pin. I called it
Connectorama. Youll also find many of the tables deemed too ungainly or of too little
general interest to include in the printed version of the text.
What a time for a book about computers, indeed! In this age of change and exciting new
technologies, its the perfect time for a book. Ive tried to make this edition of the
Hardware Bible the perfect book for the times.
WLR, 12 December 2002
Electronic Addenda
The following supplemental files, grouped by the chapters they are meant to accompany,
can be found on the official Hardware Bible Web site at www.hardwarebible.com.
Chapter 1
Chapter 6
PC History
Early Commercial Chipsets
Chapter 2
Chapter 7
Legacy Labeling Standards
BIOS Identification
Chapter 4
BIOS Performance Penalties
Microprocessor Operating Modes
BIOS Suppliers
Chapter 5
Error Codes
Microprocessors of the Past
Plug and Play Step-by-Step
Non-PC Microprocessors
Chapter 8
Chapter 12
Data Coding
Others Networking Standards
Data Packet History
Setting Up a Small Network
Chapter 9
Using Network Software
Distinguishing Interfaces
Chapter 13
Expansion Bus History
Legacy Telecommunications
Industry Standard Architecture In-depth
Modem Installation
Legacy Buses
Modem Standards
Specialty Buses
Chapter 15
Chapter 10
Alternate Drive Technologies
Access Bus
Chapter 16
ATA Life Story
30-pin SIMMs
Early SCSI Host Adapter Addressing

Limitations
Cache-on-a-Stick
Fibre Channel
Legacy Memory Issues
Chapter 17
Less Popular Disk Interfaces
5.25-inch Floppies
SCSI Operating System Support
Capacity Limits
SCSI Terminations
Choosing and Using a Backup System
Chapter 11
Digital Linear Tape
Bluetooth Details
Disk Caching
General-Purpose Interface Bus
Disk Capacity Limits
How Hardware Flow Control Works
DriveSpace File Compression
IrDA Details
Dual-Actuator Hard Disks
Parallel Port Performance Issues
Electronics and Interfaces
Parallel Port Plug-and-Play
Floptical Disks
RS-232C Buffer Control
Floppy Disk Care
USB and Old Operating Systems
Floppy Disk Control Electronics
Floppy Disk Setup
Chapter 22
Hard Disk Addressing
Grayscale Scanners
Hard Disk Installation
Using a Scanner
Hard Disk Reliability
Chapter 23
Hard Disk Warranty and Support

Linear Tape Open
Magneto Optical Storage
Old Tape Systems
QIC Min-Cartridge History
Sector Interleave and Cylinder Skewing
Stillborn High-Capacity Floppy Disk
Formats
Chapter 18
Exposure Range
Floppy Disk Camera Storage
Physical Issues
Chapter 24
Display Adapter History
Alternate Flat-Screen Technologies
CRT Phosphor Persistence
Field Emission Displays
Frame Buffer Addresses
CD and DVD Labeling
Image Compression
CD Caddies
Internal Interfaces
CD Changers
International Standards
Other CD Systems
Legacy Video Systems
Chapter 19
Monitor Practical Considerations
Special Input Devices
Monochrome Screen Phosphors
Chapter 20
Obsolete Digital Connection Systems
Keyboard Convenience Functions
Stereoscopic Displays
Chapter 21
Video BIOS
Installing Mouse Drivers

Joysticks and Paddles
Mouse Customization
Proprietary Mice
Working Without a Mouse
Video Board Operation

Video Board Types
Video Inter-Card Interfaces
Video Interfacing
Video Mode Flag
Chapter 25
ACPI States
Audio Installation
APM Operation
Compatibility Issues for Older Games
Backup Waveform
Generating Tones with the Timer
Other Surge Protection Devices
Impedance
Overvoltage Protection Devices
Other Audio Compression Algorithms
Power Supply Replacement
Sound Synthesis
Undervoltage Protection Devices
Soundboard Control Functions
UPS Waveforms
X-MIDI
Chapter 32
Chapter 26
Device Bay
Continuous Form Paper
Drive Installation
Daisywheels and Tulips
Health and Safety
Personal Printing
Regulations
Plotters
Printer Control
Chapter 27
Fax Codes
Chapter 29
Early Commercial Motherboard Products
Early Motherboard Standards
LPX Motherboards
Motherboard Mounting
WTX Motherboards
Chapter 30
CompactPCI
InfiniBand Boards
Legacy ISA Boards
Chapter 31
About the Author

Winn L. Rosch has written about personal computers since 1981 and has penned nearly
1,500 articles about thema mixture of reviews, how-to guides, and background pieces
explaining new technologiespublished in newspapers, magazines, and on the Web. He
holds three patents on computer technologies and has won two awards for his writing. He
has served as a contributing editor to more than half a dozen computer magazines (including PC Magazine) and as a technology reporter (now columnist) for the Plain Dealer in
Cleveland, Ohios largest newspaper. His books and articles have been reprinted in several
languages (among them French, German, Greek, Italian, Japanese, Portuguese, and
Spanish).
Besides writing, Rosch is an attorney licensed to practice in Ohio and holds a Juris Doctor
degree. A member of the Ohio State Bar Association, he has served on the associations
computer law committee. In other lifetimes, Rosch has worked as a photojournalist, electronic journalist, and broadcast engineer. When deadlines dont loom too close, Rosch has
been known to ambush people to collect their images on photographic film, some of
whose faces result in oil paintings.
Although Rosch has abandoned his efforts at creating a perpetual motion machine (it kept
him up at night), he is now putting the finishing touches on his latest creation, the worlds
first perpetual stillness machine.
Dedication
For Tess, our first edition together.
We Want to Hear from You!

As the reader of this book, you are our most important critic and commentator. We value
your opinion and want to know what were doing right, what we could do better, what
areas youd like to see us publish in, and any other words of wisdom youre willing to pass
our way.
As an associate publisher for Que, I welcome your comments. You can email or write me
directly to let me know what you did or didnt like about this bookas well as what we
can do to make our books better.
Please note that I cannot help you with technical problems related to the topic of this
book. We do have a User Services group, however, where I will forward specific technical
questions related to the book.
When you write, please be sure to include this books title and author as well as your
name, email address, and phone number. I will carefully review your comments and share
them with the author and editors who worked on the book.
Email:
[email protected]
Mail:
Greg Wiegand
Que Publishing
800 East 96th Street
Indianapolis, IN 46240 USA
For more information about this book or another Que title, visit our Web site at
www.quepublishing.com. Type the ISBN (excluding hyphens) or the title of a book in the
Search field to find the page youre looking for.
PART I
Introduction
Before venturing off on some quest, most people find it handy to know what they are looking
for. Imagine King Arthurs knights off looking for the Holy Grail and returning with an old
fishing net, a lipstick-stained Starbucks cup, an empty wine bottle, and so onwith each
knight claiming he had found the one true holy relic from the Last Supper. Without guidance
as to what a grail really is, each could easily justify his find and claim success in his quest.
Certainly such a free-form search would have changed the face of literature, but anyone
depending on the magical powers of the Grail would likely be disappointed.
Although youre more likely to succeed in a quest to learn about computers than Arthurs
knights were in finding the Grail, knowing what youre after will truly help you know when
youve got it.
The title of this book, taken in its most general sense, might imply that its all about nuts and
bolts, but were going to deal with a more specific kind of hardwarethe type that goes into
computer systems and the devices that computer systems go into. Of course, we will talk
about real nuts and boltsboth the ones you use to put computers together and the figurative kind. However, our aim here is understanding the digital hardware that defines our lives
and the age in which we live.
The first step on our quest will be defining what a computer really is, what it does, and how
it relates to us. Imagine King Arthur holding up a placard and pointing to it saying, This is
what the Grail you seek looks like. According to legend, he actually had it easierthe Grail
appeared to all the knights of the Round Table in a vision, so they all got a peek at the real
thing before they donned their armor. Odds are you own or have access to a computer
already, so youre ahead of the knights. Not only have you already seen one, youve also likely
touched one.
If you have a computer handy, try it. Touch your computer. Get a tingly feeling? If you did,
theres probably something wrong with the grounding of your electrical supply. Otherwise,
you probably feel like youve followed someones guidance and have done something that,
after thinking about it, is pretty stupid. You probably dont feel you know anything more
about computers than you did before you made this touch. Thats because, unlike the Grail,
the humble computer isnt going to fill you with mystical knowledge. Therefore, you know
right away that the computer is nothing mystical, nothing you have to hold in awe. More
than that, you need never touch a computer and feel stupid again. Youve already done that.
Lets begin our quest and peer inside to see what a computer really isintelligently.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Computers
Pinning down the meaning of a word is devilishly difficult, particularly
when its used by a sideshow huckster, lawyer, or some other professional
liar. Definitions vary not only with who is speaking, but also when.
Language evolves and meanings shift. Words often go to extremes, shifting meanings even to oppositescold becomes hot, bad becomes good,
and rap becomes popular.
Computer is one of those shifty words. Exactly what makes up a computer
depends on who you are and when youre talking about. Today a computer is something you can hold in your hand or at least lift with your
hands. Thirty years ago, you would have needed a hand from a friend or
two to roll a computer machine around. Sixty years ago, a computer
could have given you a hand.
Computers in History
Strictly speaking, a computer is something that computes, which is not a
particularly informative definition. In the vagueness of the term, however, youll find an interesting bit of history. The word computer does not
necessarily mean an electronic machine or a machine at all. If you were a
researcher a hundred years ago and you wanted to take a break from
heavy-duty math work, such as creating a tide table, you might have
taken your computer out to lunch. Scientists engaged in difficult mathematics often employed a bevy of computersmen and women with pencils, papers, and green eye-shades who computed the numbers they
needed.
Up until the end of World War II, a computer was a person who computed. She might use
a pencil (a pen if she were particularly confident of her results), a slide rule, or even a
mechanical calculator. Poke a few numbers in, pull a crank, and the calculator machine
printed an answer in purple ink on paper tapeat least if the questions involved simple
arithmetic, such as addition. If this person did a lot of calculations, the black ink of the
numbers soon faded to a pale gray, and he grew calluses on his fingertips and cranking
hand.
The early machines for mathematics were once all known as calculators, no matter how
elaborateand they could be quite elaborate. Charles Babbage, a 19th-century English
country gentleman with a bold idea and too much time on his hands, conceived the idea
of a machine that would replace the human computers used to calculate values in navigational tables. Babbage foresaw his mechanical computer-replacement as having three
advantages over number-crunchers who wielded pencil and paper: The machine would
eliminate mistakes, it would be faster, and it would be cheaper. He was right about all but
the last, and for that reason he never saw the most intricate machines he designed actually
built. Moreover, he never called his unbuilt machines computers. His names for them
were the Difference Engine and the Analytical Engine. Even though Babbages machines are
considered the forerunners of todays computerssometimes even considered the first
computers by people who believe they know such thingsthey really werent known as
computers in Babbages time. The word was still reserved for the humans who actually
did the work.
The word computer was first applied to machines after electricity replaced blood as the
working medium inside them. In the early part of the 20th century, researchers struggled
with the same sort of problems as those in Babbages time, and they solved them the same
way. In the 10 years from 1937 to 1947, scientists created the first devices that are classed
as true computers, starting with an electrically powered mechanical machine and ending
with an all-electronic device powered by an immense number of vacuum tubes, which
required an equally immense amount of good fortune for them to all work long enough to
carry out a calculation. Nobody called them computers just yet, however.
The first of these machinesa mechanical computer of which Babbage would have been
proudwas the IBM-financed Automatic Sequence Controlled Calculator, which is often
called Harvard Mark I. The five-ton design included 750,000 parts, including switches,
relays, and rotating shafts and clutches. It stretched out for 50 feet and was eight feet tall.
It sounded, according to an observer of the time, like a roomful of ladies knitting.
Many of the fundamentals of todays computers first took form in the partly electronic,
partly mechanical machine devised by John Vincent Atanasoff at Iowa State College (now
University). His ideas and a prototype built with the aid of graduate student Clifford
Berry have become a legend known as the Atanasoff Berry Computer (with the acronym
CHAPTER 1 COMPUTERS
Computers in Todays World
ABC), the first electronic digital computeralthough it was never contemporaneously

called a computer. Iowa State called the device the worlds fastest calculator as late as
1942.
In Britain, crypto-analysts developed a vacuum-tube (valve in Britain) device they called
Colossus that some people now call the first electronic computerusually British folk
who dont want you to forget that the English can be clever, too. But the rest of the world
never called Colossus a computeror anything elsebecause it was top secret until the
end of the century.
The present usage of the word computer goes back only to June 5, 1943, when ENIAC
(the most complex vacuum tube-based device ever made) was first proposed as a collaboration between the United States Army and the University of Pennsylvania. The original
agreement on that date first used the description that became its name, as well as the
name for all subsequent machines: the Electronic Numerical Integrator and Computer.
Three years and $486,804.22 later, the machine made its first computation at the university. The 30-ton behemoth, and its offspring, captured the imagination of the world, and
the term computer shifted from flesh-and-blood human calculators to machines. In
Hollywood, such thinking machines grew even bigger and took over the world, at least in
1950s science fiction movies. In business, ENIACs offspring, the Univac, took over
billing for utilities and gave a new name to bookkeeping foul-ups and bureaucratic incompetence: computer error. Also, scientists tried to figure out how to squeeze a room-sized
computer into a space capsule, into which they could barely shoehorn a space-suited
human being.
The scientists pretty much figured things outthey created the microprocessor, which led
to the age of microcircuitsbut not until after a few scientific diversions, including sending men to the moon. Oddly enough, although modern microelectronic circuitry is credited as an offshoot of the space program (shrinking things down and making them lighter
was important to an industry in which lifting each ounce cost thousands of dollars), in the
history of technology, the moon landing (1969) comes two years before the invention of
the microprocessor (1971).
Once the microprocessor hit, however, tinkerers figured how to make small computers
cheap enough that everyone could afford one. Computers became personal.
Computers in Todays World

The glitz is gone. Computers no longer rule Hollywood. Robots and aliens (and alien
robots in particular) are now the big villains set to rule the world. Businesses no longer
seek the fastest computers but instead strive for slownessthat is, slowing down the time
between buying and replacing machines.
Computers are now as disposable as old newspaper. We all have oneor will soon. But
their advantages hark back to Babbages claims for his unrealized Analytical Engine. They
are accurate, error-free, fast, and cheapparticularly when you compare using a computer
for calculations with doing the same work by hand. Got an aspirin?
The one claim that we wont award to the modern computer is being smarter than human
calculators. Computers arent smarter than you are, even if one can balance a checkbook
and you cant. Certainly the computer has a better head for numbers than you do. After
all, computers are specifically designed to work with numbers. For people, numbers are
at bestan afterthought, at least if youre not an accountant. People have bigger worries,
such as finding food, shelter, and sex. Computers have us to take care of those details for
them (maybe not the sex) so that they can concentrate on calculating.
Computers are good at calculatingand theyre fast. Even todays cheapest personal computers can figure the product of two 40-bit numbers billions of times in the fraction of a
second it takes one of us human beings even to realize we have two 40-bit numbers to
multiply.
Scientists and engineers like to make comparisons between the intelligence of their computers (usually the fastest computer ever built, which changes month to month) and the
thinking ability of animate creatures, typically something like an insect brain. Most scientists know its all balderdash, but they make these claims because it gets them big headlines. No mere bug can multiply two 40-bit numbersor even wants to.
Computers are more accurate because they are designed that way. Using digital logic,
their thinking automatically wipes out any noise that can confuse their calculations. By
elaborating on the math with error-checking, they can quickly detect and prevent most
mistakes. They dont think at all like you and I do. They have to be told exactly what to
do, led by the hand through the process of finding an answer by a set of instructions we
call a program.
Computers also have better memories than people. Again, they are designed that way.
One of our human advantages is that we can adapt and shift the way we deal with things
thanks to our malleable memories. Remember Moms cooking? Computers have longterm memories designed for just the opposite purposeto remember everything, down to
the last detail, without error and without limit. Even Deep Blue, the computer that finally
beat a flesh-and-blood chess Grand Master, would quickly succumb to the elements if left
outside. Although able to calculate billions of chess moves in seconds, it lacks the human
sense to come in out of the rain. And the feet.
What we call computer intelligence is something far different from what we call intelligence in humans. Thats actually good, because even experts cant agree on what human
intelligence is, or even how humans think. Things are much more straightforward for
CHAPTER 1 COMPUTERS
Computer Intelligence
computers. We know how they think as well as how to measure how well they do their
jobs.
Computer Intelligence
Biomechanical engineers are changing the way people think about insect intelligence. In
an effort to develop machines that walk like insects, they have discovered that form
doesnt dictate function, and function doesnt dictate form. The two are intimately linked.
Although some researchers have worked hard to develop machines that walk like insects
by adding more and more intelligence, others have gone back to the drawing board with a
more mechanical approach. By mimicking the structure of the insect body, they have been
able to make mechanical insects that duplicate the gait of your typical six-legged bug.
These mechanical insects can even change their gait and crawl over obstacles, all using no
more intelligence than an electric motor. In other words, biomechanical engineers have
discovered that the intelligence is in the design, not the insect.
Comparing the thinking power of the 12 neurons in an insects brain with the power of a
computer becomes a question of design. The insect is designed for eating garbage and
buzzing around your head. Although you might feed your computer a diet of garbage, its
really designed for executing programs that calculate answers for you, thus letting you
communicate and otherwise waste your time. Any comparison between bug and computer
is necessarily unfair. It must inevitably put one or the other on alien turf. The computer is
easily outclassed by anything that walks, but walking bugs wont help you solve crossword
puzzles.
With that perspective, you can begin to see the problems with comparing computer and
human intelligence. Although they sometimes perform the same tasksfor example, playing chessthey have wildly different designs. A human can forage in the jungle (African
or Manhattan), whereas a computer can only sit around and rust. A computer can turn
one pixel on your monitor green at 7:15 a.m. on August 15, 2012, whereas you might not
even bother waking up for the event.
Computers earn our awe because they can calculate faster and remember better. Even the
most complex computer applicationsvideo games, photo editors, and digital animators
all reduce down to these two capabilities.
Calculating involves making decisions and following instructions. People are able to and
often do both, so comparing computers to people is only natural. However, whereas it
might take you 10 minutes to decide between chocolate and vanilla, a computer makes a
decision in about a billionth of a second. If you had to make all the decisions your computer does to load a copy of Microsoft Windows, multicellular organisms could evolve,
rise up from the sea, mutate into dinosaurs and human beings, and then grow wings and
10
fly to nearby planets before you finished. So it stands to reason that computers are
smarter, right?
As with insects, the playing field is not quite level. Your decision isnt as simple as it looks.
Its not a single issue: Your stomach is clamoring for chocolate, but your brain knows that
with just one taste, a zit will grow on your nose and make you look like a rhinoceros. And
there are those other flavors to tempt you, too. You might just have several billion conflicts to resolve before you can select a cone, plain or sugar, and fill it full ofdid you
want ice cream or frozen yogurt, by the way?
No doubt some of us follow instructions better than others. Computers, on the other
hand, cant help but follow their instructions. They are much better at it, patiently going
through the list step by step. But again, that doesnt make them smarter than you. On the
contrary, if you follow instructions to the letter, youre apt to end up in trouble. Say
youve found the map to Blackbeards treasure, and you stand, shovel in hand, at the lip of
a precipice. If the map says to take 10 paces forward, youre smart enough to know better.
A computer confronted with the same sort of problem would plummet from its own digital precipice. It doesnt look ahead (well, some new machines have a limited ability to
sneak a peek at coming instructions), and it doesnt know enough to stop short of trouble.
Computers have excellent memory. But unlike a person, a computers memory isnt relational. Try to remember the name of an old friend, and you might race through every
name you know in alphabetical order. Or you might remember he had big ears, and
instantly the name Ross might pop into your mind. Youve related your friends name
with a distinguishing feature.
By itself, a computers memory is more like a carved stone tabletpermanent, unchanging, and exacting. A computer has to know precisely what its looking for in order to find
it. For example, a computer can find a record of your 2002 adjusted gross income quite
easily, but it cant come up with Rosss name based on the fact that he has big ears. On the
other hand, you, as a mere human, have about zero chance of remembering the 2002
adjusted gross incomes of 10,000 peoplesomething even a 10-year-old personal computer can tackle adroitly.
The point is not to call either you or your computer dumb but rather to make you see the
differences in the way you both work. The computers capabilities compliment your own.
Thats what makes them so wonderfully useful. They cant compete with people in the
brain function department, except in the imaginations of science fiction writers.
So What Is a Computer?
In todays world and using todays technology, a computer is an electronic device that uses
digital logic under the direction of a program for carrying out calculations and making
CHAPTER 1 COMPUTERS
decisions based on those calculations. By this essential definition, a computer has four elements to its construction: electronics, digital logic, programming, and calculating/decision-making.
Thats a mouthful of a definition, and one thats pretty hard to swallow, at least in one
piece. To help you understand what all this means, lets take a closer look at each of the
elements of this definition.
Electronics
The technology that makes the circuits of a computer work is called electronics. Dig into
this word, and youll find that it comes from electrons, the lightweight particles carrying a
negative charge that comprise one of the fundamental constituents of atoms. The flow of
electrons through metals is what we normally think of as electricity, a word taken from the
Greek word for amber, elektrum. Static electricity, the stuff that creates sparks when you
touch a metal doorknob after shuffling across a carpet on a dry winter day, was once most
readily produced by rubbing amber with a silk cloth. So, the stuff that makes computers
work is named after whats essentially petrified pine tree sap.
Electronics is a technology that alters the flow of charges through electrical circuits. In
the more than two centuries since Benjamin Franklin started toying with kites and keys,
scientists have learned how to use electricity to do all sorts of useful things, with two of
the most important being to operate motors and burn light bulbs. Motors can use electricity to move and change real-world objects. Lights not only help you see but also see what
an electrical circuit is doing.
For the record, this electricity stuff is essential for building a computer. Babbage showed
how it could be done with cams, levers, gears, and an old crank (which might have been
Babbage himself). If you, like Babbage, have too much time on your hands, you could
build a computer for yourself that runs hydraulically or with steam. In fact, scientists hope
to build computers that toy with quantum states.
However, electricity and electronics have several big advantages for running a computer
over nearly everything else (except that quantum stuff, and thats why scientists are interested in it). Electricity moves quickly, at nearly the speed of light. Electrical devices are
easy to interface with (or connect to) the real world. Think of those motors and electric
lights. They operate in the real world, and an electrical computer can change the currents
that run these motors and lights. Moreover, engineers have mastered the fabrication of
electrical circuits of all sizes, down to those so small you cant see them, even if you can
see the results of their work on your computer screen. And above all, electrical circuits are
familiar with off-the-shelf parts readily available, so you can easily tinker with electrical
devices and build them economically. And thats the bottom line. Just as celebrities are
famous primarily for being famous, electronics are used a lot because they are used a lot.
11
12
Digital
Most of the circuits inside a computer use a special subset of electronic technology called
digital electronics. The most important characteristic of the digital signals in these circuits is
that they usually have only two states, which are most often described as on and off (or
one and zero). Some special digital systems have more than two states, but they are more
than you need to understand right now.
Usually the states are defined as the difference between two voltage levels, typically zero
and some standard voltage, or between a positive and negative voltage value. The important part about the states of digital electronics is not what they are but what is between
themnothing. Certainly you can find a whole world of numbers between zero and
onefractions come to mindbut with digital technology the important fact is not
whether there could be something between the digital states, but that anything other than
the two digital states gets completely ignored. In essence, digital technology says if something is not a zero it is a one. It cannot be anything else.
Think about it, defining the world this way could make sense. For example, an object is
either a horse or it is not a horse. Take a close look. It has hooves, four legs, a mane, and a
tail, so you call it a horse. If it has six legs and a horny shell and, by the way, you just
stepped on it, it is probably not a horse. Yes, we could get on sketchy ground with things
such as horseflies, but nit-picking like that is just noise, which is exactly what the two digital states ignore.
This noise-free either/or design is what makes digital technology so important. Noise is
normally a problem with electrical signals. The little bit of electricity in the air leaks into
the electrical signals flowing through wires. The unwanted signal becomes noise, something that interferes with the signal you want to use. With enough noise, the signal
becomes unusable. Think of trying to converse over the telephone with a television playing in the background. At some point, turning the TV up too much makes it impossible
to hold a phone conversation. The noise level is simply too high.
Digital signals, however, allow circuits to ignore the noise. For computers, thats wonderful, because every little bit of added noise could confuse results. Adding two plus two
would equal four plus some noiseperhaps just a little, but a little might be the difference
between being solvent and having your checking account overdrawn. Noise-free digital
technology helps ensure the accuracy of computer calculations.
But sometimes things can get tricky. Say you encounter a beast with hooves, four legs, a
mane, a tail, and black-and-white stripes. Thats a horse of a different colora creature
that most zoologists would call a zebra and spend hours telling you why its not a horse.
The lesson here (besides being careful about befriending didactic zoologists) is that how
you define the difference between the two digital states is critical. You have to draw the
CHAPTER 1 COMPUTERS
line somewhere. Once you dothat is, once you decide whether a zebra is a horse or
notit fits the two-state binary logic system.
Logic
Logic is what we use to make sense of digital technology. Logic is a way of solving problems, so you can consider it a way of thinking. For computers, logically describes exactly
how they think. Computers use a special system of logic that defines rules for making
decisions based, roughly, on the same sort of deductive reasoning used by Sherlock
Holmes, some great Greek philosophers, and even you (although you might not be aware
of itand might not always use it).
Traditional logic uses combinations of statements to reach a conclusion. Here is an example of logical reasoning:
1.
Dragons eat people.
2.
I am a people.
3.
There is a dragon waiting outside my door.
Conclusion: If I go outside, I will be eaten.

This sort of reasoning works even when you believe more in superstition than science, as
the example shows.
Computer circuitry is designed to follow the rules of a formal logic system and will always
follow the rules exactly. Thats one reason why people sometimes believe computers are
infallible. But computers make no judgments about whether the statements they operate
on are true or false. As long as they get logically consistent statements (that is, the statements dont contradict one another), they will reach the correct conclusions those statements imply.
Computers are not like editors, who question the content of the information they are
given. To a computer, proposition 1, Dragons eat people, is accepted unquestioningly.
Computers dont consider whether a race of vegetarian dragons might be running about.
Computers dont judge the information they process because they dont really process
information. They process symbols represented by electrical signals. People translate
information into symbols that computers can process. The process of translation can be
long and difficult. You do some of it by typing, and you know how hard that istranslating thoughts into words, then words into keystrokes.
Computers work logically on the symbols. For the computer, these symbols take electronic form. After all, electrical signals are the only things that they can deal with. Some
symbols indicate the dragon, for example. They are the data. Other symbols indicate what
13
14
to do with the datathe logical operations to carry out. All are represented electronically
inside the computer.
Engineers figured out ways of shifting much of the translation work from you to the computer. Consequently, most of the processing power of a computer is used to translate
information from one form to another, from something compatible with human beings
into an electronic form that can be processed by the logic of the computer. Yes, someone
has to write logic to make the translationand thats what computer programming is all
about.
Programmability
A computer is programmable, which means that it follows a program. A program is a set of
instructions that tells the computer how to carry out a specific task. In that way, a program is like a recipe for chocolate chip cookies (a metaphor well visit again) that tells,
step by step, how to mix ingredients together and then burn the cookies.
Programmability is important because it determines what a computer does. Change the
program the computer follows, and it will start performing a new, different task. The
function of the computer is consequently determined by the program.
Thats how computers differ from nearly all machines that came before them. Other
machines are designed for a specific purpose: A car carries you from here to there; an
electric drill makes holes in boards and whatever else gets in the way; a toaster makes toast
from bread. But a computer? It can be a film editor, a dictation machine, a sound
recorder, or a simple calculator. The program tells it what to do.
Calculating and Decision-Making

The real work that goes on inside your computer bears no resemblance to what you ask it
to do. Programs translate human-oriented tasks into what the computer actually does.
And what the computer does is amazingly simple: It reacts to and changes patterns of bits,
data, and logic symbols.
Most of the time the bit-patterns the computer deals with represent numbers. In fact, any
pattern of binary bitsthe digital ones and zeroes that are the computers foddertranslates directly into a binary number. The computer manipulates these bit-patterns or
binary numbers to come up with its answers. In effect, it is calculating with binary numbers.
The computers calculations involve addition, subtraction, multiplication, division, and a
number of other operations you wouldnt consider arithmetic, such as logic operations
(and, or, and not) and strange tasks such as moving bits in a binary code left and right.
CHAPTER 1 COMPUTERS
More importantly, the computer can compare two binary codes and do something based
on that comparison. In effect, it decides whether one code is bigger (or smaller, or whatever) than another code and acts on that decision.
Work your way up through this discussion, and youll see that those codes arent necessarily numbers. They could be translations of human concerns and problemsor just translations of letters of the alphabet. In any case, the computer can decide what to do with the
results, even if it doesnt understand the ideas behind the symbols it manipulates.
The whole of the operation of the computer is simply one decision after another, one
operation after another, as instructed by the computers program, which is a translation of
human ideas into a logic system that uses binary code that can be carried out by electronic
signals.
So what is a computer? You have one answerone that more than anything else tells you
that there is no easy answer to what a computer is. In fact, it takes an entire book to
explain most of the details, the ins and outs of a computer. This book.
15
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Functions and Components

Going through a long-winded description of what a computer is doesnt
seem to make sense when you can see what a computer is, sitting right in
front of you. Its a box, a keyboard, and a screen with a headache
attached.
Rather than looking at a strictly formal definition of a computer, its
probably more useful to learn what it is with a more practical approach:
Examine the machine in front of you. After all, you can see the parts and
how they fit together.
If you have a desktop computer or a tower, youve got a big box with an
empty slot or two on the front and a bunch of jacks and connectors on
the back. A notebook scatters the slots and jacks all around its edges, but
in either style of case you get essentially the same repertory. And, if you
look closely in either case, youll see that all the parts are held together
with a few screws. With a flashlight in hand, you might be able to squint
through the slots and see some of the internal parts. Get daring and twist
out the screws, and youll get a much better view. But there will still be
something you cannot seewhat those parts do.
The functions of the components of a computer are not at all obvious
based on what they look like. Although some machines make the functions of their parts quite obviousthe wheels of a car pretty much shout
their functionwith a computer, the function of each part is hidden in a
snarled mass of digital electronic circuitry and accompanying programs.
Take, for example, arguably the most important part of any modern
computer, the microprocessor. Its inner workings must be protected
inside a hermetically sealed case because air would destroy the delicate
18
crystalline circuits inside. Even if the microprocessor were encased in glass, however, you
could never see the electricity flowing through it or the logic operations it performs in
creating your computers thoughts. About the only electricity you can normally see is
lightning, and you probably hope youll never see that inside your computer.
Even though all the good stuff inside a computer is essentially invisible, examining the
individual parts that deal with the stuff you cant see is key to understanding how the
whole thing works. This microscope-to-the-parts strategy makes sense for economic as
well as explanatory reasons. You buy a computer as a collection of parts, even if you pay
one price and get one big box. One reason is that not all computers are the same, a fact
that should be obvious as soon as you pull out your credit card. You can buy a brand new
(though technologically old) computer for barely more than $100 today, or you can pay
fifty times more. Even though you might not notice a difference between them when you
run Microsoft Wordscary as it sounds, thats often true for reasons youll understand
before you finish this bookmanufacturers and retailers can easily justify the difference.
One machine can handle some tasks (obviously other than running Word) more adroitly
than the other. The underlying reason for this difference is a matter of the component
parts from which the two machines are made.
Electronic devices, whether computers, digital cameras, or portable radios, are all built
from tiny electronic parts such as resistors, capacitors, transistors, and integrated circuits.
Each of these changes the flow of electronics in some small, simple way, and figuring out
how to connect them together to accomplish some tiny task is how electrical engineers
earn their salaries. But combine these tiny assemblies together at the next level (another
task for engineers), and the result is a module or computer component with a definite,
defined task. Each is a subassembly like the various parts of a car or refrigerator. A car has
a motor, wheels, steering system, doors, and windows. A refrigerator has a motor, a compressor, cooling coils, a box, doors, and insulation. Similarly, every computer is built from
an array of components, such as a microprocessor, power supply, and flashing lights.
Each of these individual components has a well-defined function. For example, in your
car, the motor, transmission, axle, and wheels make the car move along the highway, providing the motive function. In the refrigerator, the motor, compressor, and coils make up
the cooling function.
Of course, the car has other functions in addition to its motive function. For example, the
body, windows, and seats provide a passenger-carrying function, much as a refrigerator has a
food-holding function. Although at first thought such secondary functions might seem incidental to the overall concept of the car or refrigerator, these functions are actually essential parts that help define what a car or refrigerator is. After all, you wouldnt have much
of a car if it couldnt go anywhere or hold people. Nor would a refrigerator be as useful if
it couldnt keep cool or hold food.
CHAPTER 2 FUNCTIONS AND COMPONENTS
Processing
Similarly, the computer has several functions that define what it is and what it does.
Although some of the functions might seem incidental to the concept of a computer as a
thinking (or calculating) machine, all are essential for making a modern computer the useful device that it is.
The typical modern personal computer has four major functions. These include thinking,
communicating, remembering, and listening and monitoring, all tied together by an electronic and mechanical infrastructure. Of course, this division is somewhat arbitrary. Some
people might, for example, combine thinking and remembering, and others (say people
who sell security software) might add additional functions, such as security. But this fiveway split has one big advantage over any other: Its how Ive chosen to arrange this book.
Each of these functions requires one or more hardware components to carry it out. For
example, thinking is not merely a matter for microprocessors. It requires memory in
which to execute programs, a chipset to link the microprocessors circuits to the rest of the
computer, and some semi-permanent software in the form of the BIOS to bring everything to life. Memory involves not only the chips that we think of as computer memory
but also devices for longer-term memory, which include hard disks and other mass storage
devices that keep a computer from forgetting even after you switch its power off.
Well start our computer tour by taking a look at these functions and subdividing them
into the components your computer uses to carry them out.
Processing
The function that defines the computer is its ability to calculate and make decisions.
Without the ability to make decisions, a computer could not follow a program more complex than a simple sequence. If your computer lacked the ability to calculate, you might as
well have a footstool next to your desk or a dumbbell in your briefcase. These abilities
give the computer its electronic thinking poweror more importantly, help the computer
enhance your own thinking power.
Several components make up the thinking function of a computer. The most important is
the microprocessor, but by itself a microprocessor couldnt function. It would be like a brain
without a spine and blood supply. For the computer, such support functions are handled
by a chipset. In addition, the computer needs the equivalent of instinct in animals and
humans, the primitive behaviors that help it survive even without learning. For the computer, the equivalent of instinct is the BIOS, a set of factory-installed, essentially
unchangeable programs that give the computer its basic functions and, some say,
personality.
19
20
Microprocessor
The most important of the electronic components on the motherboard is the microprocessor. It does the actual thinking inside the computer. The power of a computer
how fast it can accomplish a given job, such as resizing a digital photodepends on the
model of microprocessor inside the computer as well as how fast that microprocessor
operates (the speed is measured in the familiar megahertz or gigahertz). The kind of
microprocessor also determines what software language it understands. For example,
Windows computers and Macintosh computers use microprocessors from different families that understand different software languages.
As fits its role, the microprocessor usually is the largest single integrated circuit in a computer. It makes more connections, so it has the biggest socket and usually holds the dominant position on the main circuit board. It is the centerpiece of every computer. In fact,
the microprocessor is the most complicated device yet devised by human beings, so complex that earlier designs couldnt fit all the silicon microcircuitry into a single chip. Many
older microprocessors (such as the Pentium II series) were modules that combined several
smaller integrated circuit chips into a big assembly that included a main microprocessor, a
coprocessor, a cache controller, and cache memory. Today, however, everything for an
advanced microprocessor such as the Pentium 4 fits on a single silicon chip about oneinch square.
Chipset
The chipset of a computer provides vital support functions to its microprocessor. The
chipset creates signals that are the lifeblood of the microprocessor, such as the clock or
oscillator that sets the pace of its logic operations. In addition, the chipset links the microprocessor to the rest of the computer, both the memory and external functions, through
input/output ports. The chipset also provides the vital link to your computers expansion
bus that enables you to add new capabilities to its repertory. The chipset is so important
that in most computers it affects the performance and operation of the system as much as
does its microprocessor. In fact, for some knowledgeable buyers, the choice of chipset is a
major purchasing criterion that distinguishes one computer from another.
At one time, a chipset was a collection of dozens of individual electronic components. In
todays computers, however, manufacturers have combined all the traditional functions of
this essential support circuitry into a few large integrated circuits. In computers, in fact,
the entire chipset has been squeezed into a single package. Typically the integrated circuit
or circuits that make up the chipset are squares of black epoxy sitting on the main computer circuit board, usually the largest individual electronic components there, except for
the microprocessor.
Communicating
BIOS
Just as animals rely on instincts to survive in the real world before they can learn from
their experiences, a computer has a built-in program that tells it what to do before you
load any software. This program is called the Basic Input/Output System because it tells the
computers microprocessor how to get input from the outside world and send output
there. The BIOS defines how a computer acts and behaves before you load software. In
modern computers, the BIOS has several additional functions, all essential to making the
computers get started and work.
Unlike the microprocessor and chipset, the BIOS is mostly ephemeral: It is a program, a
list of software codes. It takes physical form because it permanently resides in a special
kind of memory chip, one that retains its memory without the need for electricity. This
way, the BIOS program is always remembered, ready to be used as soon as the computer
gets switched on. The chip holding the BIOS typically is a large flash memory chip. Its
most distinguishing feature is its label, however. Because it holds software, the BIOS chip
is usually emblazoned with a copyright notice just like other software products.
Communicating
The real useful work that computers do involves not just you but also the outside world.
Your computer must be able to communicate to put its intelligence to work. When your
computer communicates with other systems far away, the process is often called telecommunications. When your computer connects with other computers over a network, engineers call the communication capability connectivity. When your computer plugs into
printers and other nearby peripherals, engineers say your computer is doing what its supposed totheres no fancy word for it. No matter. Thanks to the communication capabilities of your computer, it can link to any of a number of hardware peripherals through its
network jacks and input/output ports. Better still, through modems and the Internet, it
can connect with nearly any computer in the world.
Expansion Buses
Computers need to communicate with any of a number of peripherals, some of which
reside inside the computers case. The primary link to these internal components is the
expansion bus.
As the name implies, the expansion bus of a computer allows you to expand its capabilities
by sliding in accessory boards (cleverly termed expansion boards). For this to work, the
expansion bus sets up an internal communication channel inside your computer. Expansion
slots are spaces inside the computer that provide special sockets or connectors to plug in
the capabilities and functions locked in the circuitry of the expansion boards.
21
22
In a desktop computer, the expansion bus usually is a row of three or more long connectors on the main circuit board near the back of the computers case. Depending on the
overall design of the computer, one or more of these slots will be filled with expansion
boards in the basic factory configuration. In a notebook computer, expansion slots are different, meant to accept modules the size of credit cards that deliver the same functions as
expansion boards.
Interfaces
Interfaces provide a communication channel that lets your computer exchange information with a variety of devices, primarily storage systems (discussed later in the chapter).
The interface translates the signals inside your computer into a form thats more suited to
traveling outside the confines of its main circuit boards. Youve probably heard people
speak about the most familiar interfaces, such as ATA (also called IDE) and SCSI,
acronyms that describe connections used by hard and optical disk drives.
The interface takes the form of a connector. The ATA interface is usually built in to the
main circuit board of all modern computers. A cable links this connector to one on a disk
drive. The SCSI interface usually resides on a separate circuit board that fits into the
expansion bus of your computer.
Input/Output Ports
Your computer links to its peripherals through its input and output ports. Every computer
needs some way of acquiring information and putting it to work. Input/output ports are
the primary routes for this information exchange.
In the past, the standard equipment of most computers was simple and almost preordainedone serial port and one parallel port, typically as part of their motherboard circuitry. Modern standards are phasing out these ports, so well consider them (for purposes
of this book) legacy ports. Today, new and wonderful port standards are proliferating faster
than dandelions in a new lawn. Hard-wired serial connections are moving to the new
Universal Serial Bus (USB), whereas the Infrared Data Association (IrDA) system and
oddly named Bluetooth provide wireless links. Digital video connections use FireWire,
also called IEEE 1394. Even the simple parallel port has become an external expansion
bus capable of linking dozens of devices to a single jack.
The ports are the jacks or connectors youll find on the back of most desktop computers
or scattered around the edges of notebook machines. They come in various sizes and
shapes, meant to match special connectors unambiguously.
Communicating
Local Area Networks

Any time you link two or more computers together, youve created a network. Keep the
machines all in one placeone home, one business, one site in todays jargonand you
have a local area network (LAN). Spread them across the country, world, or universe with
telephone, cable, or satellite links and you get a wide area network (WAN). A network is
both a wiring system and a software system. The wires connect computers together; the
software is the language for passing messages around.
Most networks use some kind of wire to connect computers, although wireles networks are
becoming popular, especially in homes where their short range is no problem and their
lack of wires a great advantage.
Telecommunications
To extend the reach of your computer beyond your home or office, you usually must rely
on the international telephone system to provide the connection. Because short-sighted
engineers a hundred years ago never considered that youd want to connect your computer to your telephone, they built the phone system to use an entirely different kind of
signal than your computer uses. Consequently, when you want to connect with other
computers and information sources such as the Internet through the international telephone system, you need a modem to adapt your computers data to a form compatible with
the telephone systems.
In a quest for faster transfers than the ancient technology of the classic telephone circuit
can provide, however, data communications are shifting to newer systems, including digital telephone services (such as DSL), high-speed cable connections, and direct digital links
with satellites. Each of these requires its own variety of connecting devicenot strictly
speaking a modem but called that for consistencys sake. Which you need depends on
the speed you want and the connections available to you.
Internet
The Internet is properly described as a network of networks. In concept, it links all the
computers in the world together so that they can share information (but more often
games and pornography). The World Wide Web is essentially the commercial side of the
Internet. Once you link up with the Web, your computer is no longer merely the box on
your desk. It becomes part of a single, massive international computer system with a single
goal: transferring money out of your bank account. Even so, it retains all the features and
abilities you expect from a computeran Internet connection only makes it even more
powerful.
The Internet is more an idea than a physical form. Picture it as a spider web anchored to
each and every computer in the world.
23
24
Remembering
No matter how smart you are, you wouldnt know anything if you couldnt remember.
Thoughts and ideas would go in one neuron and out another, forever lost in the entropy
of the universe. You know things because you can call back thoughts and ideas, to work on
them again or just talk about them. A computer, too, needs some way of retaining its
thoughts. Like you, it needs both short-term memory for holding ideas while it works on
them and long-term memory to store all the facts and old ideas that, by chance, it might
need another day.
The short-term memory of computers is often called simply memory. The long-term
memory is often termed mass storage and involves several technologies. Hard disk drives
hold the ideas you put into the computer, both commercial programs and your own original data. Floppy disks and optical drives (CDs and DVDs for the most part) store the ideas of
others than you want your computer to access. Tape drives provide a safety net, keeping a
duplicate copy of your computers most important ideas.
Memory
Just as you need your hands and workbench to hold tools and raw materials to make
things, your computers microprocessor needs a place to hold the data it works on and the
tools to do its work. Memory, which is often described by the more specific term RAM
(which means Random Access Memory), serves as the microprocessors workbench. The
amount and architecture of the memory of a system determines how it can be programmed and, to some extent, the level of complexity of the problems it can work on.
Modern software often requires that you install a specific minimum of memorya minimum measured in megabytesto execute properly. With modern operating systems, more
memory often equates to faster overall system performance.
In todays computers, memory usually comes in subassemblies called memory modules that
plug into special sockets on the main circuit board of your computer. Most computers
have three or more of these sockets in a group, one or more of which is filled with a
memory module as standard equipment.
Hard Disk Drives

Long-term memory is where you store thoughts and ideas that, although you dont need
them immediately, you need to knowstuff like your shoe size, which doesnt come up in
everyday conversation (at least not among normal adults) but becomes a handy factoid
when musing through a mail-order catalog. Your computers hard disk holds such factoids
along with all the other stuff it needs to know but not at this exact momentsuch as the
instructions for programs youre not using, financial records you hope the auditor will
ignore, term papers you someday hope to publish as best-selling fiction, and even extra
Remembering
designs for wallpaper for your computer screen. Your computers hard disk lets you call on
any of those stored facts on a microseconds notice.
Most hard disks take the form of a sealed box, typically silver and black with a green circuit board dotted with tiny electrical components on the bottom. A hard disk connects to
the main circuit board of your computer through that interface we talked about earlier via
a short, wide, and flat set of wires, called a ribbon cable because it looks like a ribbon
(although an admittedly strange gray ribbed ribbon).
Floppy Disk Drives

Once upon a time, some computers lacked hard disk drives and instead used floppy disk
drives (once upon a time being in the unbelievable past, a little before Cinderella
cleaned out fireplaces). Inexpensive, exchangeable, and technically unchallenging, the
floppy disk served as a data interchange system for years because it was based on wellproven technologies and was mass produced by the millions.
Today, the floppy disk drive is found on nearly every computer, but functionally its about
as useful to your computer as an appendix. (Granted, an appendix wont fit into a drive
slot, but you get the idea.) In todays world the floppy disk is slow and holds too little
information, so it is gradually disappearing from computers, replaced in function by any
of a variety of optical drives.
The computers that still have floppy disk drives usually wear them proudly, out in front.
On desktop systems, its usually the uppermost drive in the case. From the outside you see
only a long, black slot about three-and-a-half inches wide with a rectangular button near
one corner. Notebook computers put their floppy disk drives, if they have one, wherever
they fitusually hidden in the side of the computer under the keyboard.
Optical Drives
Getting new memories into your computer is the primary job served by optical drives.
They store programs and data in a convenient formsmall discsthats standardized so
that you can exchange the discs (and memories) between computers. (Note that due to a
strange quirk of design and origin, magnetic drives use disks whereas optical drives use
discs.)
Optical discs currently come in two flavors: Compact Discs and Digital Versatile Discs.
These are the CDs and DVDs you slide into your home entertainment system. In either
form, discs are cheap and easy to copy, so software publishers have made the CD-ROM
their preferred means of getting their products to you, with DVDs slowly replacing CDROMs because of their higher capacity. You can create your own optical disks with a suitable drive to store either data for your computer or music and video for your
entertainment systems.
25
26
A CD-ROM or DVD drive is usually the other drive on the front of your desktop computer, bigger than the floppy and featuring a volume control and headphone jack (should
you want to slide a music CD into the drive). Youll know what it is as soon as you press
the eject button and the tray rolls out. On a notebook computer, manufacturers stuff their
optical drives wherever they fit, usually on the side near the back of the computer.
Tape Drives
Tape is for backup, pure and simple. It provides an inexpensive place to put your data just
in casejust in case some light-fingered freelancer decides to separate your computer
from your desktop; just in case the fire department hoses to death everything in your
office that the fire and smoke failed to destroy; just in case you empty your Recycle Bin
moments before discovering you accidentally deleted all your exculpatory tax records; just
in case an errant asteroid ambles through your roof. Having an extra copy of your important data helps you recover from such disasters as well as those that are even less likely to
happen.
Tape drives are optional on personal computers. They add enough cost that people would
rather risk their data. On larger computers (the servers used in business), tape drives are
more common because the cost of restoring data is so high, probably thousands of times
the cost of a drive, that their use is justified.
Listening and Monitoring

If you want your computer to do anything useful, you have to be able to tell it what to do.
Although some computers actually do listen to you speak using voice-recognition technology, most systems depend on traditional control systems, the keyboard, and the mouse to
put you in command. In addition, many applications require that you fill your computer
with datakeystrokes, images, and sounds. Your computer acquires this information by
electronically listening to any of several input devicesscanners, digital cameras, sound
board samplers, and so on.
Your window into the mind of your computer that lets you monitor what it does is its display system, itself a combination of a graphics adapter or video board and a monitor or
flat-panel display. The display system gives your computer the means to tell you what it is
thinking and to show you your data in the form that you best understandbe it numbers,
words, or pictures. The two halves of the display system work hand in hand. The graphics
adapter uses the digital signals inside your computer to build an electronic map of what
the final image should look like, storing the data for every dot on your monitor in memory. Electronics generate the image that appears on your monitor screen.
Keyboard
The keyboard remains the most efficient way to enter text into applications, faster than
even the most advanced voice-recognition systems, which let you talk to your computer.
Scientists believe that keyboarding is a more direct route from your brain. Speaking
requires more work, and the mechanics of sounding your voice take more time. But the
keyboard lets you do more than mere typing. It is the primary command-and-control center for your computer.
The keyboard is probably the most identifiable part of any computer. You cant miss an
array of over 100 buttons meant to be individually activated by your fingers. They come
built into notebook computers, tethered on short cords to most desktop systems, and
warped and bent in magnificent curves meant to soothe tired hands.
Pointing Devices
Although computers once recognized only textfor which a keyboard sufficedtodays
computers work with pictures (what computer-people call graphics). The keyboard is notoriously poor at picking out pieces of pictures, but the mousemore correctly termed a
pointing device to include mouse-derived devices such as trackballs and the proprietary
devices used by notebook computersreadily relays graphic instructions to your computer. Not only can the pointing device point, but it also lets you draw, sketch, and paint.
The most familiar pointing device is the hand-size mouse that you move around your
desktop to cause a mouse pointer to move analogously on the screen. But some people prefer upside-down mice called trackballs for rolling the pointer around. Notebook computers
use specialized pointing devices called pointing sticks (IBMs TrackPoint is the premier
example) or touchpads, which let you move around the desktop by scratching your finger
across a plastic pad.
Scanners
The range of input devices naturally grows to embrace whatever kind of information you
want to get into your computer. Say you have on-paper images you want to capture
drawings your kids made in first grade, your likeness on a poster from the post office, portraits of presidents on small sheets of paperso that you can get into your graphics
program to clean them up. Engineers developed the scanner to do just that. A scanner dissects the image into bits of graphic information like little dots that your computer can
manipulate and store digitally.
Although engineers have developed an entire zoo of scanners to suit different kinds of
original material (for example, photos or documents), most of todays scanners prefer to
look at flat sheets. They take the form of flat boxes with glass plates on top, on which you
lay printed images that you want to capture to your computer.
27
28
Digital Cameras
Scanners work when you have an image already fixed on paper, but when you want to get
something you see in the real world into computer-compatible form, you need a digital
camera. As with a scanner, a digital camera reduces an imagein this case, a scene you see
in the real worldinto tiny dots that your computer can store as digital data. The digital
camera is a completely self-contained system for capturing and saving real-world images.
It works with and without a computer, but the full power of the digital camera is realized
best when you can edit its images with software on your computer.
A digital camera looks and works much like old-time film cameras with a lens in front and
a viewfinder to help you frame the scenes you want to save. Some digital cameras are as
simple as old box cameraspress a button and let the camera do the rest. Others have
more controls than a small space shuttle, letting you unleash your full creative powers.
Display Systems
Your computers thoughts, like your own, are invisible. Making your thoughts visible to
others is what language and art are all about. The computers equivalent is the graphics
adapter. This component takes the digital thoughts of your computer and turns them into
an electronic image, essentially a picture saved in the computers electronic memory.
Once a graphics adapter has an image in this formbe it a page of text or a picture of an
oasisa monitor (sometimes called a display) connected to your computer makes the actual
image that you can see. Monitors are essentially television sets built to military specifications. They produce sharper, higher-quality images than any television (including a new
digital, high-resolution models) can hope to show.
Almost any monitor will let you work with your desktop computer (the monitor is built in
to notebook machines), but the quality of the monitor attached to your computer determines the quality of the image you see and, often, of the work you do. Although no monitor can make anything look better than whats in the signals from your graphics adapter, a
bad monitor can make them look much worse and limit both the range of colors and the
resolution (or sharpness) of the images.
In most desktop computers, the graphics adapter is an expansion board, often the only one
that comes as factory equipment, although notebook computers and many desktop systems have their graphics adapter functionality built in to the circuitry of the main circuit
boards. Modern monitors for desktop computers come in two types: Cathode-ray tube
(CRT) monitors use picture tubes like old-fashioned television sets and are big boxes that
sit on your desk and, often, crowd everything off. Flat-panel monitors use liquid-crystal
displays (LCDs) that run cooler and take up less room than CRTs (but cost substantially
more). Either type is tethered to a desktop computer by a short cable. Notebook computers universally have built-in flat-panel displays.
Audio
A pretty picture is not enough in todays world. You need a total experience, one that
involves both pictures and sound. After all, silent films went out of date in 1927.
Consequently, every computer now has a sound system. In business computers, its often
nothing more than a beeper to warn you when youve pressed the wrong key. In most of
todays computers, however, the sound system is on par with those in theaters (only not as
loud). They can generate any sound you can hear and make it appear to be anywhere in
the room.
Speakers are the most obvious part of any sound system, if just because youve got to find
a place for two or three more boxes on your office desk. The real sound-generating capability of a computer comes from its sound board, however, which either slides into an
expansion slot in the computer or is built into its main circuitry.
Printers
The electronic thoughts of a computer are notoriously evanescent. Pull the plug and your
work disappears. Moreover, monitors are frustratingly difficult to pass around and post
through the mail when you want to show off your latest digital art creation. Hard copy (the
printout on paper) solves this problem. A printer is the device of choice for making hard
copy from your computers thoughts.
Printers come in many varieties. The most common are laser printers and inkjet printers.
Either is a box with two paper trays that connects to your computer with a short cord.
Lasers make fast, cheap, black-and-white pages. Inkjets are low-cost printers that excel at
printing color but do so at a higher price per page because of their use of expensive inks.
Fax
Fax combines both the telecommunications capability of a modem with the hard-copy
output of a printer. In effect, a fax machine becomes a remote printer for your computera system that allows your computer to print remotely to anywhere in the world
you can get a telephone connection.
Today, fax is an invisible part of nearly every modem. If you can get online through a dialup connection with your computer, you can fax. In most cases, you need nothing more
than the modem already in your computer. The only addition required is invisible software.
29
30
Infrastructure
Infrastructure is what holds your computer together. It is the overall form of your computer, the substrate that it is built upon, and the essential supplies required to make it run.
The substrate of the modern computer is the motherboard (or system board or main
board) coupled with the expansion boards plugged into it. The case provides the basic
form of the computer. And the one essential service required by any modern computer is
power, the electricity used in the computers logic operations.
Motherboards
The centerpiece of the system unit is the motherboard. All the other circuitry of the system unit is usually part of the motherboard or plugs directly into it. The electronic components on the motherboard carry out most of the function of the machinerunning
programs, making calculations, and even arranging the bits that will display on the screen.
Because the motherboard defines each computers functions and capabilities and because
every computer is different, it only stands to reason that every motherboard is different,
too. Not exactly. Many different computers have the same motherboard designs inside.
And oftentimes a single computer model might have any of several different motherboards, depending on when it came down the production line (and what motherboard the
manufacturer got the best deal on).
The motherboard is the main circuit board in nearly every computerdesktop or laptop.
Usually it lines the bottom of the case like thick, green shelf-paper decorated with lines
and blobs that could be a Martian roadmap.
Expansion Boards
Expansion boards are the physical form of the expansion devices that connect to your
computer through its expansion slots. They are simply circuit boards that are smaller than
the motherboard.
However, they are not just any circuit boards. Expansion boards must follow strict standards as to size, signals, power consumption, and software interface to ensure that they
will work inside any given computer. Although at one time nearly all computers used
Industry Standard Architecture (ISA) for their expansion boards, the modern replacement
is Peripheral Component Interconnect (PCI). Some desktop computers dont allow the
use of expansion boards at all. Notebook machines use PC cards instead.
Power
Every computer requires a continuous supply of carefully conditioned direct current at
low voltage, well below that available at your wall outlet. Although batteries can provide
Infrastructure
this power for portable computers, desktop units require power supplies to convert house
current into computer current. Similarly, power supplies charge and substitute for
portable computer batteries.
Bringing power to your computer also imports danger, however, and your computer
requires protection from glitches on the power line. Surge suppressors and backup power systems help to ensure that your computer gets the proper electrical diet.
The power supply in most desktop computers is locked for your protection in a steel box
inside your computers case. It keeps your hands out of harms way if you dig into your
computer to add an expansion board or other accessory. Notebook computers put much of
the function of the power supply into their external bricks that plug directly into a wall
outlet. The output of the brick is at a voltage thats safe should you accidentally encounter
it (and safe for your notebook computer, as well).
Cases
The case is what most people think of as the entire computer, ignoring all the essential
parts attached by cables and often the individual components inside. The main box goes
by a special name, system unit, because it is the heart of the system. Some people call it the
CPU (central processing unit), although that term is also (and confusingly) used for the
microprocessor inside. The system unit houses the main circuitry of the computer and
provides spaces called bays for internal disk drives as well as a place for the jacks (or outlets) that link the computer to the rest of its accouterments, including the keyboard, monitor, and peripherals. A notebook computer combines all these external components into
one unit, but its usually called simply the computer rather than system unit or CPU.
The case is more than a box. Its also a protective shell that keeps the dangers of the real
world away from the delicate electronics of the computer. Whats more, it protects the
world from interference from the signals of the computer. The case is also part of a cooling system (which may include one or more fans) that keeps the electronics inside cool for
longer life.
The case of a modern computer can take many formsfrom small-footprint desktop
computers to maxi-tower computers, from compact sub-notebook computers (or even
handheld computers) to huge, almost unportable desktop-replacement notebook computers. In desktop systems, the size more than anything else determines how many accessories you can add to your systemand how much of your home or office you have to
give up to technology. Weight, more than size, is the limiting physical factor for notebook
computers. How much you want to carry often determines how many features you get.
31
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Software
You probably think you already know what software is. Its that CD a
friend gives you with the latest game to run on your computer. Were life
only so simple (and copyright laws so liberal) that you could live happily
ever after with such beliefs.
Life, alas, is not so simple, and neither is software. The disc isnt software. Software is nothing you can hold in your hand. Rather, its the
hardware that stores the software. The real software is made from the
same stuff as dreams, as evanescent as an inspiration, as elusive as the
meaning of the current copyright laws. Software is nothing but a set of
ideas, ideas that (one hopes) express a way to do something. Those ideas
are written in a code the same way our words and sentences code our
thoughts. The codes take the form of symbols, which may appear on
paper or be represented as pulses in electrical circuits. No matter. The
code is only the representation of the ideas, and the ideas are really the
software.
So much for the philosophy of software. Not many people think about
software that way. Most approach it more practically. Software is something your computer needs. You might say your computer needs software
to run its programs, but thats like saying you need food to eat dinner.
The food is dinner. The software is the programs.
Ouch. No matter how bad that sounds, thats the way most people think.
As a practical matter, theyre getting close to the truth, but software is
more than programs even at a practical level. Software is more than the
box you buy and little silver disc that comes inside it. Software is not a
singular thing. It is an entire technology that embraces not only what
34
you see when you run your computer but also a multitude of invisible happenings hiding
beneath the surface. A modern computer runs several software programs simultaneously,
even when you think youre using just one, or even when you dont think anything is running at all. These programs operate at different levels, each one taking care of its own
specific job, invisibly linking to the others to give you the illusion youre working with a
single, smooth-running machine.
What shows on your monitor is only the most obvious of all this softwarethe
applications, the programs such as Microsoft Office and Grand Theft Auto 3 that you actually
buy and load onto your computer, the ones that boldly emblazon their names on your
screen every time you launch them. But theres more. Related to applications are the utilities you use to keep your computer in top running order, protect yourself from disasters,
and automate repetitive chores. Down deeper is the operating system, which links your
applications and utilities together to the actual hardware of your computer. At or below
the operating system level, you use programming languages to tell your computer what to
do. You can write applications, utilities, or even your own operating system with the right
programming language.
Definitions
Software earns its name for what it is not. It is not hardware. Whatever is not hard is soft,
and thus the derivation of the name. Hardware came firstsimply because the hard reality of machines and tools existed long before anyone thought of computers or of the concept of programming. Hardware happily resides on page 551 of the 1965 dictionary I keep
in my officeyou know, nuts, bolts, and stuff like that. Software is nowhere to be found.
(Why I keep a 1965 dictionary in my office is another matter entirely.)
Software comprises abstract ideas. In computers, the term embraces not only the application programs you buy and the other kinds of software listed earlier but also the information or data used by those programs.
Programs are the more useful parts of software because programs do the actual work. The
program tells your computer what to dohow to act, how to use the data it has, how to
react to your commands, how to be a computer at all. Data just slow things down.
Although important, a program is actually a simple thing. Broken down to its constituent
parts, a computer program is nothing but a list of commands to tell your hardware what
to do. Like a recipe, the program is a step-by-step procedure that, to the uninitiated,
seems to be written in a secret code. In fact, the program is written in a code called the
programming language, and the resultant list of instructions is usually called code by its programmers.
CHAPTER 3 SOFTWARE
Applications
Applications
The programs you buy in the box off your dealers shelves, in person or through the Web,
the ones you run to do actual work on your computer, are its applications. The word is
actually short for application software. These are programs with a purpose, programs you
apply to get something done. They are the dominant beasts of computing, the top of the
food chain, the colorful boxes in the store, the software you actually pay for. Everything
else in your computer system, hardware and software alike, exists merely to make your
applications work. Your applications determine what you need in your computer simply
because they wont runor run wellif you dont supply them with what they want.
Todays typical application comes on one or more CDs and is comprised of megabytes,
possibly hundreds of megabytes, of digital stuff that you dutifully copy to your hard disk
during the installation process. Hidden inside these megabytes is the actual function of
the program, the part of the code that does what you buy the software forbe it to translate keystrokes into documents, calculate your bank balance, brighten your photographs,
or turn MP3 files into music. The part of the program that actually works on the data you
want to process is called an algorithm, the mathematical formula of the task converted into
program code. An algorithm is just a way for doing something written down as instructions so you can do it again.
The hard-core computing work performed by major applicationsthe work of the algorithms inside themis typically both simple and repetitive. For example, a tough statistical analysis may involve only a few lines of calculations, although the simple calculations
will often be repeated again and again. Changing the color of a photo is no more than a
simple algorithm executed over and over for each dot in the image.
Thats why computers exist at all. They are simply good at repeatedly carrying out the
simple mathematical operations of the algorithms without complaining.
If you were to tear apart a program to see how it workswhat computer scientists call disassembling the programyoud make a shocking discovery. The algorithm makes up little
of the code of a program. Most of the multimegabyte bulk you buy is meant to hide the
algorithm from you, like the fillers and flavoring added to some potent but noxious medicine.
Before the days of graphical operating systems, as exemplified by Microsofts ubiquitous
Windows family, the bulk of the code of most software applications was devoted to making the rigorous requirements of the computer hardware more palatable to your human
ideas, aspirations, and whims. The part of the software that serves as the bridge between
your human understanding and the computers needs is called the user interface. It can be
anything from a typewritten question mark that demands you type some response to a
multicolor graphic menu luring your mouse to point and click.
35
36
Windows simplifies the programmers task by providing most of the user interface functions from your applications. Now, most of the bulk of an application is devoted to mating
not with your computer but with Windows. The effect is the same. It just takes more
megabytes to get there.
No matter whether your application must build its own user interface or rely on the one
provided by Windows, the most important job of most modern software is simply translation. The program converts your commands, instructions, and desires into a form
digestible by your operating system and computer. In particular, the user interface translates the words you type and the motion of your arm pointing your mouse into computer
code.
This translation function, like the Windows user interface, is consistent across most applications. All programs work with the same kind of human input and produce the same kind
of computer codes. The big differences between modern applications are the algorithms
central to the tasks to be carried out. Application software often is divided into several
broad classes based on these tasks. Table 3.1 lists the traditional division of functions or
major classifications of computer application software.
TA B L E 3 . 1
Basic Types of Computer Application Software
Class of Software
Function
Web browsers
Entering and interacting with Web sites and servers to buy,

sell, research your favorite interests, and just while away
evenings, afternoons, and the rest of your life
E-mail programs
Sending and receiving instant messages from friends, associates, and solicitors in your neighborhood and around the
world
Word processors
Getting your words write for print or electronic publishing
Spreadsheets
The accountants ledger made automatic to calculate arrays of

numbers
Databases
A filing system with instant access and the ability to automatically sort itself
Drawing and painting programs
Creating and editing images such as blueprints and cartoon

cels that can be filed and edited with electronic ease
Multimedia software
Playing MP3 and WAV files for music, showing AVI or MOV
video files, playing games, or displaying images and sound like
a movie theatre under the control of an absolute dictator (you)
CHAPTER 3 SOFTWARE
Applications
The lines between many of these applications are blurry. For example, many people find
that spreadsheets serve all their database needs, and most spreadsheets now incorporate
their own graphics for charting results.
Suites
Several software publishers completely confound the distinctions by combining most of
these application functions into a single package that includes database, graphics, spreadsheet, and word processing functionalities. These combinations are termed application
suites. Ideally, they offer several advantages. Because many functions (and particularly the
user interface) are shared between applications, large portions of code need not be duplicated, as would be the case with standalone applications. Because the programs work
together, they better know and understand one anothers resource requirements, which
means you should encounter fewer conflicts and memory shortfalls. Because they are all
packaged together, you stand to get a better price from the publisher.
Today theres another name for the application suite: Microsoft Office. Although at one time
at least three major suites competed for space on your desktop, the other offerings have,
for the most part, faded away. You can still buy Corel WordPerfect Office or Lotus
SmartSuite, although almost no one does. (In fact, most copies of Office are sold to businesses. Most individuals get it bundled along with a new computer.) Office has become
popular because it offers a single-box solution that fills the needs of most people, handling
more tasks with more depth than they ordinarily need. The current version, Office XP,
dominates the market because Microsoft sells it to computer makers at a favorable price
(and, some say, with more than a little coercion).
Utilities
Even when youre working toward a specific goal, you often have to make some side trips.
Although they seem unrelated to where youre going, they are as much a necessary part of
the journey as any other. You may run a billion-dollar pickle-packing empire from your
office, but you might never get your business negotiations done were it not for the regular
housekeeping that keeps the place clean enough for visiting dignitaries to walk around
without slipping on pickle juice on the floor.
The situation is the same with software. Although you need applications to get your work
done, you need to take care of basic housekeeping functions to keep your system running
in top condition and working most efficiently. The programs that handle the necessary
auxiliary functions are called utility software.
From the name alone you know that utilities do something useful, which in itself sets
them apart from much of the software on todays market. Of course, the usefulness of any
tool depends on the job you have to doa pastry chef has little need for the hammer that
37
38
so well serves the carpenter or computer technicianand most utilities are crafted for
some similar, specific need. For example, common computer utilities keep your disk organized and running at top speed, prevent disasters by detecting disk problems and viruses,
and save your sanity should you accidentally erase a file.
The most important of these functions are included with todays computer operating systems, either integrated into the operating system itself or as individual programs that are
part of the operating system package. Others you buy separately, at least until Microsoft
buys out the company that offers them.
Common utilities include backup, disk defragmenting, font management, file compression, and schedulingall of which were once individual programs from different publishers but now come packaged in Windows. Antivirus and version-tracking programs are
utilities available from separately from Windows.
Modern utilities are essentially individual programs that load like ordinary applications
when you call on them. The only difference between them and other applications is what
they do. Utilities are meant to maintain your system rather than come up with answers or
generate output.
Applets
An applet is a small software application thats usually dedicated to a single simple purpose.
It may function as a standalone program you launch like an ordinary application, or it may
run from within another application. Typically the applet performs a housekeeping function much like a utility, but the function of an applet is devoted to supporting an overriding application rather than your computer in general. That said, some system utilities may
take the form of applets, too. Applets are mostly distinguished from other software by
their size and the scope of their functions.
When you download a page from the Web that does something exciting on the screen
(which means it shows you something other than text that just lays there), its likely youve
invisibly downloaded an applet from the Internet. Applets let you or distant programmers
add animation and screen effects to Web pages.
Other applets are often included with or as part of application software packages (for
example, the wizard that leads you through the installation process). Some are included
with operating systems such as Windows. In fact, the chief distinction between an applet
and a full application may be little more than the fact that you dont buy applets separately
and never have.
CHAPTER 3 SOFTWARE
Operating Systems
Operating Systems
The basic level of software with which you will work on your computer is the operating
system. Its what you see when you dont have an application or utility program running.
But an operating system is much more than what you see on the screen.
As the name implies, the operating system tells your computer how to operate, how to
carry on its most basic functions. Early operating systems were designed simply to control
how you read from and wrote to files on disks and were hence termed disk operating systems
(which is why the original computer operating system was called DOS). Todays operating
systems add a wealth of functions for controlling every possible computer peripheral from
keyboard (and mouse) to monitor screen.
The operating system in todays computers has evolved from simply providing a means of
controlling disk storage into a complex web of interacting programs that perform several
functions. The most important of these is linking the various elements of your computer
system together. These linked elements include your computer hardware, your programs,
and you. In computer language, the operating system is said to provide a common hardware interface, a common programming interface, and a common user interface.
An interface, by the way, is the point where two things connect togetherfor example, the
human interface is where you, the human being, interact with your computer. The hardware interface is where your computer hardware links to its software. The programming
interface is where programs link to the operating system. And the user interface is where
you, as the user, link to the operating system. Interfaces can combine and blend together.
For example, the user interface of your operating system is part of the human interface of
your computer.
Of the operating systems many interfaces only one, the user interface, is visible to you.
The user interface is the place where you interact with your computer at its most basic
level. Sometimes this part of the operating system is called the user shell. In todays operating systems, the shell is simply another program, and you can substitute one shell for
another. Although with Windows most people stick with the shell that Microsoft gives
them, you dont have to. People who use Unix or Linux often pick their own favorite
shell.
In the way they change the appearance of your operating system, shells are like the skins
used by some applicationsfor example, the skins that give your MP3 player software the
look of an old-fashioned Wurlitzer jukebox.
In effect, the shell is a starting point to get your applications running, and its the home
base that you return to between applications. The shell is the program that paints the
desktop on the screen and lets you choose the applications you want to run.
39
40
Behind the shell, the Application Program Interface (or API) of the operating system gives
programmers a uniform set of calls, key words that instruct the operating system to execute a built-in program routine that carries out some predefined function. For example,
the API of Windows enables programmers to link their applications to the operating system to take advantage of its user interface. A program can call a routine from the operating system that draws a menu box on the screen.
Using the API offers programmers the benefit of having the complicated aspects of common program procedures already written and ready to go. Programmers dont have to
waste their time on the minutiae of moving every bit on your monitor screen or other
common operations. The use of a common base of code also eliminates duplication, which
makes todays overweight applications a bit more svelte. Moreover, because all applications
use basically the same code, they have a consistent look and work in a consistent manner.
This prevents your computer from looking like the accidental amalgamation of the latenight work of thousands of slightly aberrant engineers that it is.
At the other side of the API, the operating system links your applications to the underlying computer hardware through the hardware interface. Once we take a look at what
that hardware might be, well take a look how the operating system makes the connection
in the following section, titled Hardware Interfaces.
Outside of the shell of the user interface, you see and directly interact with little of an
operating system. The bulk of the operating system program code works invisibly (and
continuously). And thats the way its designed to be.
Operating systems must match your hardware to work on a given computer. The software
instructions must be able to operate your computer. Consequently, Apple computers and
Intel-based computers use different operating systems. Similarly, programs must match
the underlying operating system, so you must be sure that any given program you select is
compatible with your operating system.
Today, four families of operating system are popular with personal computers.
Windows 9X
The Windows 9X family includes Windows 95, Windows 98, and Windows Millennium
Edition (Me). These operating systems are built on a core of the original 16-bit DOS
(dating back to 1981) and are meant to run only on Intel microprocessors and chips that
are completely compatible with Intel architecture.
Microsoft announced Windows at a news conference in New York on November 10,
1983. But development of the graphic environment (as it was then classified) proved more
troublesome than Microsoft anticipated, and the first version of the 16-bit version of
Windows did not reach store shelves as Windows 1.0 until November 20, 1985. The last
CHAPTER 3 SOFTWARE
Operating Systems
major release of 16-bit Windows was Windows Me, which was first put on sale on
September 14, 2000.
Because DOS was originally written in a time before running several programs at once
was common, Microsofts programmers had to add on this ability and provide the power
to isolate each program from one another. In fact, they kludged together DOS, graphics,
and multitasking control into what most programmers regard as a gnarled mess of code.
Little wonder, then, that these 16-bit versions of Windows have a reputation for unreliability and are most suited to home computer users. You wouldnt want your life to depend
on them.
Windows NT
The 32-bit Windows family includes Windows NT, Windows 2000, and Windows XP, in
all its versions. When the core of this operating systems was originally conceived as
Window NT, Microsofts programmers decided to scrap all of the original Windows and
start over. Instead of DOS they used OS/2, an operating system jointly developed by
Microsoft and IBM, as its core. Microsoft called the first release of the new operating system Windows NT 3.1, cleverly skipping over early version numbers to make the new NT
seem like a continuation of the then-current 16-bit Windows version, 3.1. According to
Microsoft, Windows NT 3.1 was officially released to manufacturing on July 27, 1993.
The same 32-bit core, although extensively refined, is still in use in current versions of
Windows XP.
Instead of recycling old 16-bit code, Microsofts programmers started off with the 32-bit
code native to the 386 microprocessor that was current at the time (and is still used by
todays latest Pentium 4 microprocessors). Although Microsoft attempted to craft versions
of Windows NT to run on processors other than those using Intel architecture, no such
adaptation has proven successful, and the operating system runs exclusively on Intel-architecture microprocessors.
From the very beginning, Microsoft conceived Windows NT as a multitasking operating
system, so its engineers built in features that ensure the isolation of each application it
runs. The result was a robust operating system that had but one problemit lacked complete compatibility with old, DOS-based programs. Over the years, this compatibility
became both less necessary (as fewer people ran DOS applications) and better, as
Microsofts engineers refined their code. Most professionals regard Windows XP as a
robust operating system on par with any other, suitable for running critical applications.
Unix
Unix is not a single operating system but several incompatible families that share a common command structure. Originally written at Bell Labs to run on a 16-bit Digital
41
42
Equipment Company PDP-7 computer, Unix has been successfully adapted to nearly
every microprocessor family and includes 8-, 16-, 32-, and 64-bit versions under various
names.
Unix traces its roots to 1969, but according to Dennis M. Ritchie, one of its principal
developers (along with Ken Thompson), its development was not announced to the world
until 1974. From the beginning, Unix was designed to be a time-sharing, multiuser system
for mainframe-style computers. It proved so elegant and robust that it has been adapted to
nearly every microprocessor platform, and it runs on everything from desktop computers
to huge clusters of servers.
Nearly every major hardware platform has a proprietary version of Unix available to it.
Some of these include AIX for IBM computers, Solaris for Sun Microsystems machines,
and even OS X, the latest Apple Macintosh operating system. Note that programs written
for one version of Unix will likely not run on other hardware than for which it was
intended. It has become a sort of universal language among many computer professionals.
Linux
Unlike other operating systems, Linux can be traced back to a single individual creator,
Linus Torvalds, who wrote its kernel, the core of the operating system that integrates its
most basic functions, in 1991 while a student at the University of Helsinki. He sent early
versions around for testing and comment as early as September of that year, but a workable version (numbered 0.11) didnt become available until December. Version 1.0,
regarded as the first truly stable (and therefore usable) version of Linux came in March,
1994. Torvalds himself recommends pronouncing the name of the operating system with a
short-i sound (say Lynn-ucks).
The most notable aspect of Linux is that it is open source. That is, Torvalds has shared
the source codethe lines of program code he actually wroteof his operating system
with the world, whereas most software publishers keep their source code secret. This
open-source model, as it is called, lets other programmers help refine and develop the
operating system. It also results in what is essentially a free, although copyrighted, operating system.
Contrary to popular opinion, Linux is not a variety of Unix. It looks and works like
Unixthe commands and structure are the samebut its kernel is entirely different.
Programs written for Unix will not run under Linux (although if you have the source
code, these programs can be relatively easily converted between the two operating systems). Programmers have adapted Linux to many hardware platforms, including those
based on Intel microprocessors.
The open-source status of Linux, along with its reputation for reliability and robustness
(inherited without justification from Unix), have made it popular. In other words, its like
CHAPTER 3 SOFTWARE
Programming Languages
Unix and its free, so it has become the choice for many Web servers and small business
systems. On the other hand, the strength it gains from its similarity to Unix is also a
weakness. It is difficult for most people to install, and it lacks the wide software support of
Windows.
A computer programwhether its an applet, utility, application, or operating systemis
nothing more than a list of instructions for the brain inside your computer, the microprocessor, to carry out. A microprocessor instruction, in turn, is a specific pattern of bits, a
digital code. Your computer sends the list of instructions making up a program to its
microprocessor one at a time. Upon receiving each instruction, the microprocessor looks
up what function the code says to do, then it carries out the appropriate action.
Microprocessors by themselves only react to patterns of electrical signals. Reduced to its
purest form, the computer program is information that finds its final representation as the
ever-changing pattern of signals applied to the pins of the microprocessor. That electrical
pattern is difficult for most people to think about, so the ideas in the program are traditionally represented in a form more meaningful to human beings. That representation of
instructions in human-recognizable form is called a programming language.
As with a human language, a programming language is a set of symbols and the syntax for
putting them together. Instead of human words or letters, the symbols of the programming language correspond to patterns of bits that signal a microprocessor exactly as letters
of the alphabet represent sounds that you might speak. Of course, with the same back-tothe-real-basics reasoning, an orange is a collection of quarks squatting together with reasonable stability in the center of your fruit bowl.
The metaphor is apt. The primary constituents of an orangewhether you consider them
quarks, atoms, or moleculesare essentially interchangeable, even indistinguishable. By
itself, every one is meaningless. Only when they are taken together do they make something worthwhile (at least from a human perspective): the orange. The overall pattern, not
the individual pieces, is whats important.
Letters and words work the same way. A box full of vowels wouldnt mean anything to
anyone not engaged in a heated game of Wheel of Fortune. Match the vowels with consonants and arrange them properly, and you might make words of irreplaceable value to
humanity: the works of Shakespeare, Einsteins expression of general relativity, or the formula for Coca-Cola. The meaning is not in the pieces but their patterns.
The same holds true for computer programs. The individual commands are not as important as the pattern they make when they are put together. Only the pattern is truly meaningful.
43
44
You make the pattern of a computer program by writing a list of commands for a microprocessor to carry out. At this level, programming is like writing reminder notes for a nottoo-bright personfirst socks, then shoes.
This step-by-step command system is perfect for control freaks but otherwise is more
than most people want to tangle with. Even simple computer operations require dozens of
microprocessor operations, so writing complete lists of commands in this form can be
more than many programmerslet alone normal human beingswant to deal with. To
make life and writing programs more understandable, engineers developed higher-level
programming languages.
A higher-level language uses a vocabulary thats more familiar to people than patterns of
bits, often commands that look something like ordinary words. A special program translates each higher-level command into a sequence of bit-patterns that tells the microprocessor what to do.
Machine Language
Every microprocessor understand its own repertoire of instructions, just as a dog might
understands a few spoken commands. Whereas your pooch might sit down and roll over
when you ask it to, your processor can add, subtract, and move bit-patterns around as well
as change them. Every family of microprocessor has a set of instructions that it can recognize and carry out: the necessary understanding designed into the internal circuitry of
each microprocessor chip.
The entire group of commands that a given model of microprocessor understands and can
react to is called that microprocessors instruction set or its command set. Different microprocessor families recognize different instruction sets, so the commands meant for one
chip family would be gibberish to another. For example, the Intel family of microprocessors understands one command set; the IBM/Motorola PowerPC family of chips recognizes an entirely different command set. Thats the basic reason why programs written for
the Apple Macintosh (which is based on PowerPC microprocessors) wont work on computers that use Intel microprocessors.
That native language a microprocessor understands, including the instruction set and the
rules for using it, is called machine language. The bit-patterns of electrical signals in
machine language can be expressed directly as a series of ones and zeros, such as 0010110.
Note that this pattern directly corresponds to a binary (or base-two) number. As with any
binary number, the machine language code of an instruction can be translated into other
numerical systems as well. Most commonly, machine language instructions are expressed
in hexadecimal form (base-16 number system). For example, the 0010110 subtraction
instruction becomes 16(hex). (The (hex) indicates the number is in hexadecimal notation, that is, base 16.)
CHAPTER 3 SOFTWARE
Assembly Language
Machine language is great if youre a machine. People, however, dont usually think in
terms of bit-patterns or pure numbers. Although some otherwise normal human beings
can and do program in machine language, the rigors of dealing with the obscure codes
takes more than a little getting used to. After weeks, months, or years of machine language programming, you begin to learn which numbers do what. Thats great if you want
to dedicate your life to talking to machines, but not so good if you have better things to
do with your time.
For human beings, a better representation of machine language codes involves mnemonics
rather than strictly numerical codes. Descriptive word fragments can be assigned to each
machine language code so that 16(hex) might translate into SUB (for subtraction).
Assembly language takes this additional step, enabling programmers to write in more memorable symbols.
Once a program is written in assembly language, it must be converted into the machine
language code understood by the microprocessor. A special program, called an assembler,
handles the necessary conversion. Most assemblers do even more to make the programmers life more manageable. For example, they enable blocks of instructions to be linked
together into a block called a subroutine, which can later be called into action by using its
name instead of repeating the same block of instructions again and again.
Most of assembly language involves directly operating the microprocessor using the
mnemonic equivalents of its machine language instructions. Consequently, programmers
must be able to think in the same step-by-step manner as the microprocessor. Every
action that the microprocessor does must be handled in its lowest terms. Assembly language is consequently known as a low-level language because programmers write at the
most basic level.
High-Level Languages
Just as an assembler can convert the mnemonics and subroutines of assembly language
into machine language, a computer program can go one step further by translating more
human-like instructions into the multiple machine language instructions needed to carry
them out. In effect, each language instruction becomes a subroutine in itself.
The breaking of the one-to-one correspondence between language instruction and
machine language code puts this kind of programming one level of abstraction farther
from the microprocessor. Thats the job of the high-level languages. Instead of dealing
with each movement of a byte of information, high-level languages enable the programmer to deal with problems as decimal numbers, words, or graphic elements. The language
program takes each of these high-level instructions and converts it into a long series of
digital code microprocessor commands in machine language.
45
46
High-level languages can be classified into two types: interpreted and compiled. Batch
languages are a special kind of interpreted language.
Interpreted Languages
An interpreted language is translated from human to machine form each time it is run by
a program called an interpreter. People who need immediate gratification like interpreted
programs because they can be run immediately, without intervening steps. If the computer
encounters a programming error, it can be fixed, and the program can be tested again
immediately. On the other hand, the computer must make its interpretation each time the
program is run, performing the same act again and again. This repetition wastes the computers time. More importantly, because the computer is doing two things at onceboth
executing the program and interpreting it at the same timeit runs more slowly.
Today the most important interpreted computer language is Java, the tongue of the Web
created by Sun Microsystems. Your computer downloads a list of Java commands and converts them into executable form inside your computer. Your computer then runs the Java
code to put make some obnoxious advertisement dance and flash across your screen.
The interpreted design of Java helps make it universal. The Java code contains instructions that any computer can carry out, regardless of its operating system. The Java interpreter inside your computer converts the universal code into the specific machine
language instructions your computer and its operating system understand.
Before Java, the most popular interpreted language was BASIC, an acronym for the
Beginners All-purpose Symbolic Instruction Set. BASIC was the first language for personal computers and was the foundation upon which the Microsoft Corporation was built.
In classic form, using an interpreted language involved two steps. First, you would start
the language interpreter program, which gave you a new environment to work in, complete with its own system of commands and prompts. Once in that environment, you then
executed your program, typically starting it with a Run instruction. More modern interpreted systems such as Java hide the actual interpreter from you. The Java program
appears to run automatically by itself, although in reality the interpreter is hidden in your
Internet browser or operating system. Microsofts Visual Basic gets its interpreter support
from a runtime module, which must be available to your computers operating system for
Visual Basic programs to run.
Compiled Languages
Compiled languages execute like a program written in assembler, but the code is written
in a more human-like form. A program written with a compiled language gets translated
from high-level symbols into machine language just once. The resultant machine language
is then stored and called into action each time you run the program. The act of converting the program from the English-like compiled language into machine language is called
CHAPTER 3 SOFTWARE
compiling the program. To do this you use a language program called a compiler. The original, English-like version of the program, the words and symbols actually written by the
programmer, is called the source code. The resultant machine language makes up the programs object code.
Compiling a complex program can be a long operation, taking minutes, even hours. Once
the program is compiled, however, it runs quickly because the computer needs only to run
the resultant machine language instructions instead of having to run a program interpreter
at the same time. Most of the time, you run a compiled program directly from the DOS
prompt or by clicking an icon. The operating system loads and executes the program
without further ado. Examples of compiled languages include todays most popular computer programming language, C++, as well as other tongues left over from earlier days of
programmingCOBOL, Fortran, and Pascal.
Object-oriented languages are special compiled languages designed so that programmers
can write complex programs as separate modules termed objects. A programmer writes an
object for a specific, common task and gives it a name. To carry out the function assigned
to an object, the programmer need only put its name in the program without reiterating
all the objects code. A program may use the same object in many places and at many different times. Moreover, a programmer can put a copy of an object into different programs
without the need to rewrite and test the basic code, which speeds up the creation of complex programs. C++ is object oriented.
Optimizing compilers do the same thing as ordinary compilers, but do it better. By adding
an extra step (or more) to the program compiling process, the optimizing compiler checks
to ensure that program instructions are arranged in the most efficient order possible to
take advantage of all the capabilities of the computers processor. In effect, the optimizing
compiler does the work that would otherwise require the concentration of an assembly
language programmer.
Libraries
Inventing the wheel was difficult and probably took human beings something like a million yearsa long time to have your car sitting up on blocks. Reinventing the wheel is
easier because you can steal your design from a pattern you already know. But its far, far
easier to simply go out and buy a wheel.
Writing program code for a specific but common task often is equivalent to reinventing
your own wheel. Youre stuck with stringing together a long list of program commands,
just like all the other people writing programs have to do. Your biggest consolation is that
you need to do it only once. You can then reuse the same set of instructions the next time
you have to write a program that needs the same function.
47
48
For really common functions, you dont have to do that. Rather than reinventing the
wheel, you can buy one. Todays programming languages include collections of common
functions called libraries so that you dont have to bother with reinventing anything. You
only need pick the prepackaged code you want from the library and incorporate it into
your program. The language compiler links the appropriate libraries to your program so
that you only need to refer to a function by a code name. The functions in the library
become an extension to the language, called a meta-language.
Development Environments
Even when using libraries, youre still stuck with writing a program in the old-fashioned
waya list of instructions. Thats an effective but tedious way of building a program. The
computers strength is taking over tedious tasks, so youd think you could use some of that
power to help you write programs more easily. In fact, you might expect someone to write
a program to help you write programs.
A development environment is exactly that kind of program, one that lets you drag and drop
items from menus to build the interfaces for your own applications. The environment
includes not only all the routines of the library, but also its own, easy-to-use interface for
putting the code in those libraries to work. Most environments let you create programs by
interactively choosing the features you wantdrop-down menus, dialog boxes, and even
entire animated screensand match them with operations. The environment watches
what you do, remembers everything, and then kicks in a code generator, which creates the
series of programming language commands that achieves the same end. One example of a
development environment is Microsoft Visual Studio.
Working with a development environment is a breeze compared to traditional programming. For example, instead of writing all the commands to pop a dialog box on the screen,
you click a menu and choose the kind of box you want. After the box obediently pops on
the screen, you can choose the elements you want inside it from another menu, using your
mouse to drag buttons and labels around inside the box. When youre happy with the
results, the program grinds out the code. In a few minutes you can accomplish what it
might have taken you days to write by hand.
Batch Languages
A batch language allows you to submit a program directly to your operating system for execution. That is, the batch language is a set of operating system commands that your computer executes sequentially as a program. The resultant batch program works like an
interpreted language in that each step gets evaluated and executed only as it appears in the
program.
Applications often include their own batch languages. These, too, are merely lists of commands for the application to carry out in the order youve listed them to perform some
CHAPTER 3 SOFTWARE
Hardware Interfaces
common, everyday function. Communications programs use this type of programming to

automatically log in to the service of your choice and even retrieve files. Databases use
their own sort of programming to automatically generate reports that you regularly need.
The process of transcribing your list of commands is usually termed scripting. The commands that you can put in your program scripts are sometimes called the scripting
language.
Scripting actually is programming. The only difference is the language. Because you use
commands that are second nature to you (at least after youve learned to use the program)
and follow the syntax that youve already learned running the program, the process seems
more natural than writing in a programming language. That means if youve ever written
a script to log on to the Internet or have modified an existing script, youre a programmer
already. Give yourself a gold star.
Hardware Interfaces
Linking hardware to software has always been one of the biggest challenges facing those
charged with designing computer systemsnot just individual computers but whole families of computers, those able to run the same programs. The solution has almost invariably been to wed the two together by layering on software, so much so that the fastest
processors struggle when confronted by relatively simple tasks. An almost unbelievable
amount of computer power gets devoted to moving bytes from software to hardware
through a maze of program code, all in the name of making the widest variety of hardware
imaginable work with a universal (at least if Microsoft has its way) operating system.
The underlying problem is the same as in any mating ritual. Software is from Venus, and
hardware is from Mars (or, to ruin the allusion for sake of accuracy, Vulcan). Software is
the programmers labor of love, an ephemeral spirit that can only be represented.
Hardware is the physical reality, the stuff pounded out in Vulcans forgeenduring,
unchanging, and often priced like gold. (And, yes, for all you Trekkers out there, it is
always logical.) Bringing the two together is a challenge that even self-help books would
find hard to manage. Yet every computer not only faces that formidable task but also tackles it with aplomb, although maybe not as fast as youd like.
Heres the challenge: In the basic computer, every instruction in a program gets targeted
on the microprocessor. Consequently, the instructions can control only the microprocessor and dont themselves reach beyond. The circuitry of the rest of the computer and the
peripherals connected to it all must get their commands and data relayed to them by the
microprocessor. Somehow the microprocessor must be able to send signals to these
devices. Today the pathway of command is rarely direct. Rather, the chain of command is
a hierarchy, one that often takes on aspects of a bureaucracy.
49
50
Control Hierarchy
Perhaps the best way to get to know how software controls hardware is to look at how
your system executes a simple command. Lets start with a common situation.
Because your patience is so sorely tested by reading the electronic version of this book
using your Web browser, Internet Scapegoat, you decide to quit and go on to do something
really useful, such as playing FreeCell. Your hand wraps around your mouse, you scoot it
up to the big black at the upper-right corner of the screen, and you click your left
mouse button. Your hardware has made a link with software.
Your click is a physical act, one that closes the contacts of a switch inside the mouse,
squirting a brief pulse of electricity through its circuitry. The mouse hardware reacts by
sending a message out the mouse wire to your computer. The mouse port of your computer detects the message and warns your computer by sending a special attention signal
called a hardware interrupt squarely at your microprocessor.
At the same time, the mouse driver has been counting pulses sent out by your mouse that
indicate its motion. The mouse driver counts each pulse and puts the results into the
memory of your computer. It uses these values to find where to put the mouse pointer on
the screen.
The interrupt causes your computer to run a software interrupt routine contained in the
mouse driver software. The driver, in turn, signals to Windows that youve clicked the button. Windows checks in memory for the value the mouse driver has stored there for the
position of the mouse pointer. This value tells Windows where the mouse pointer is to
determine how to react to your button press. When it discovers youve targeted the ,
Windows sends a message to the program associated with itin this case, your browser.
The browsers reacts, muttering to itself the digital equivalent of He must be insane, and
it immediately decides to pop up a dialog box that asks whether you really, really want to
quit such a quality program as the browser. The dialog box routine is part of the program
itself, but it builds the box and its contents from graphics subroutines that are part of the
Windows operating system. The browser activates the subroutines through the Windows
application interface.
Windows itself does not draw the box. Rather it acts as a translator, converting the box
request into a series of commands to draw it. It then sends these commands to its graphics
driver. The driver determines what commands to use so that the video board will understand what to do. The driver then passes those commands to another driver, the one associated with the video board.
The video boards driver routes the command to the proper hardware ports through
which the video board accepts instructions. The driver sends a series of commands that
CHAPTER 3 SOFTWARE
Hardware Interfaces
causes the processor on the video board (a graphics accelerator) to compute where it must
change pixels to make the lines constituting the box. Once the graphic accelerator finishes
the computations, it changes the bytes that correspond to the areas on the screen where
the line will appear in a special memory area on the video board called the frame buffer.
Another part of the video board, the rasterizer, scans through the frame buffer and sends
the data it finds there to the port leading to your monitor, converting it into a serial data
stream for the journey. Using synchronizing signals sent from the video board as a guide,
the monitor takes the data from the data stream and illuminates the proper pixels on the
screen to form the pattern of the dialog box.
When you awake from your boredom-inspired daze, you finally see the warning box pop
up on the screen and react.
The journey is tortuous, but when all goes right, it takes a smaller fraction of a second
than it does for you to become aware of what youve done. And when all doesnt go right,
another tortuous chain of events will likely result, often one involving flying computers,
picture windows, and shards of glass on the front lawn.
Keys to making this chain of command function are the application program interface, the
driver software, and the device interfaces of the hardware itself. Sometimes a special part
of your computer, the BIOS, also gets involved. Lets take a deeper look at each of these
links.
Application Program Interface

A quick recap: An interface is where two distinct entities come together. The most important of the software interfaces in the Windows environment is the application program
interface or API.
Rather than a physical thing, the API is a standard set of rules for exchanging commands
and data. The Windows API comprises a set of word-like commands termed program calls.
Each of these causes Windows to take a particular action. For example, the command
DrawBox could tell Windows to draw a box on the screen, as in the preceding example.
To pass along data associated with the commandin the example, how large the box
should be and where to put it on the screenmany calls require your program to send
along parameters. Each parameter is ordinary data that is strictly formatted to meet the
expectation of Windows. That is, the order of the parameters is predefined, and the range
permitted for the data is similarly constrained.
Each software interface in the Windows system uses a similar system of calls and parameter passing. Gaining familiarity with the full repertory of the API is one of the biggest
challenges facing programmers.
51
52
Device Drivers
Device drivers are matchmakers. A device driver takes a set of standardized commands
from the operating system and matches them to the capabilities of the device that the driver controls. Typically the device that gets controlled is a piece of hardware, but, as our
example shows, one driver may control another driver that controls the hardware.
Just about every class of peripheral has some special function shared with no other device.
Printers need to switch ribbon colors; graphics boards need to put dots onscreen at high
resolution; sound boards need to blast fortissimo arpeggios; video capture boards must
grab frames; and mice have to do whatever mice do. Different manufacturers often have
widely different ideas about the best way to handle even the most fundamental functions.
No programmer or even collaborative program can ever hope to know all the possibilities.
Its even unlikely that you could fit all the possibilities into an operating system written in
code that would fit onto a stack of disks you could carry. There are just too many possibilities.
Drivers make the connection, translating generalized commands made for any hardware
into those used by the specific device in your computer. Instead of packing every control
or command you might potentially need, the driver implements only those that are appropriate for a specific type, brand, and model of product that you actually connect up to
your computer. Without the driver, your operating system could not communicate with
your computer.
Device drivers give you a further advantage. You can change them almost as often as you
change your mind. If you discover a bug in one driversay sending an uppercase F to
your printer causes it to form-feed through a full ream of paper before coming to a panting stopyou can slide in an updated driver that fixes the problem. You dont have to
replace the device or alter your software. In some cases, new drivers extend the features of
your existing peripherals because programmer didnt have enough time or inspiration to
add everything to the initial release.
The way you and your system handles drivers depends on your operating system. Older
operating systems (such as DOS and old versions of Windows) load all their drivers when
they start and stick with them all while you use your computer. Windows 95 and newer
versions treat drivers dynamically, loading them only when they are needed. Not only
does this design save memory, because you only need to load the drivers that are actually
in use, it also lets you add and remove devices while youre using your computer. For
example, when you plug in a USB scanner, Windows can determine what make and model
of scanner you have and then load the driver appropriate for it.
Although today drivers load invisibly and automatically, things were not always so easy.
The driver needs to know what system resources your hardware uses for its communications. These resources are the values you set through the Windows Add Hardware
CHAPTER 3 SOFTWARE
Hardware Interfaces
Wizard (click the Resources tab in Device Manager to see them). With old-fashioned drivers, hardware, and operating systems, you had to physically adjust settings on the hardware to assign resources and then configure the driver to match what you configured. It
wasnt pretty, often wasnt easy, and was the most common reason people could not get
hardware to workusually because they created the problem themselves. Now Windows
creates the problems automatically.
BIOS
The Basic Input/Output System, mentioned earlier as giving a personality to your computer,
has many other functions. One of the original design intentions of the BIOS was to help
match your computers hardware to its software. To that end, the BIOS was meant to act
as a special driver software included with your computer so that it can boot up right after
you take it out of the box.
Part of the BIOS is program code for drivers thats permanently or semi-permanently
recorded in special memory chips. The code acts like the hardware interface of an operating system but at a lower levelit is a hardware interface thats independent of the operating system.
Programs or operating systems send commands to the BIOS, and the BIOS sends out the
instructions to the hardware using the proper resource values. It lies waiting in your computer, ready for use.
The original goal of the BIOS was to make computer software platform independent,
meaning that programs dont care what kind of computer theyre running on. Although
that seems a trivial concern in these days of dynamic-loading drivers, it was only a dream
decades ago when computers were invented. In those dark-old days, programmers had to
write commands aimed specifically at the hardware of the computer. Change the hardwareplug in a different printerand the software wouldnt work. The BIOS, like todays
driver software, was meant to wallpaper over the difference in hardware.
The idea, although a good one, didnt work. Programmers avoided the BIOS in older
computers because it added an extra software layer that slowed things down. But the
BIOS really fell from favor when Microsoft introduced modern Windows. The old hardware-based BIOS couldnt keep up with changes in technology. Moreover, using software
drivers allowed hardware engineers to use more memory for the needed program code.
The BIOS persists in modern computers as a common means of accessing hardware
before the operating system loads. The BIOS code of every computer today still includes
the equivalent of driver software to handle accessing floppy disk drives, the keyboard,
printers, video, and parallel and serial port operation.
53
54
Device Interfaces
Regardless of whether software uses device drivers, looks through the BIOS, or accesses
hardware directly, the final link to hardware may be made in one of two ways, set by the
hardware design: input/output mapping and memory mapping. Input/output mapping relies
on sending instructions and data through ports. Memory mapping requires passing data
through memory addresses. Ports and addresses are similar in concept but different in
operation.
Input/Output Mapping
A port is an address but not a physical location. The port is a logical construct that operates as an addressing system separate from the address bus of the microprocessor, even
though it uses the same address lines. If you imagine normal memory addresses as a set of
pigeon holes for holding bytes, input/output ports act like a second set of pigeon holes on
the other side of the room. To distinguish which set of holes to use, the microprocessor
controls a flag signal on its bus called memory I/O. In one condition, it tells the rest of the
computer that the signals on the address bus indicate a memory location; in its other state,
the signals indicate an input/output port.
The microprocessors internal mechanism for sending data to a port also differs from
memory access. One instruction, move, allows the microprocessor to move bytes from any
of its registers to any memory location. Some microprocessor operations can even be performed in immediate mode, directly on the values stored at memory locations.
Ports, however, use a pair of instructions: In to read from a port, and Out to write to a
port. The values read can only be transferred into one specific register of the microprocessor (called the accumulator) and can only be written from that register. The accumulator has other functions as well. Immediate operations on values held at port locations is
impossible, which means a value stored in a port cannot be changed by the microprocessor. It must load the port value into the accumulator, alter it, and then reload the new
value back into the port.
Memory Mapping
The essence of memory mapping is sharing. The microprocessor and the hardware device
it controls share access to a specific range of memory addresses. To send data to the
device, your microprocessor simply moves the information into the memory locations
exactly as if it were storing something for later recall. The hardware device can then read
those same locations to obtain the data.
Memory-mapped devices, of course, need direct access to your computers memory bus.
Through this connection, they can gain speed and operate as fast as the memory system
CHAPTER 3 SOFTWARE
Hardware Interfaces
and its bus connection allow. In addition, the microprocessor can directly manipulate the
data at the memory location used by the connection, thus eliminating the multistep
load/change/reload process required by I/O mapping.
The most familiar memory-mapped device is your computers display. Most graphic systems allow the microprocessor to directly address the frame buffer that holds the image
that appears on your monitor screen. This design allows the video system to operate at
the highest possible speed.
The addresses used for memory mapping must be off limits to the range in which the
operating system loads your programs. If a program should transgress on the area used for
the hardware connection, it can inadvertently change the data therenearly always with
bad results. Moreover, the addresses used by the interface cannot serve any other function,
so they take away from the maximum memory addressable by a computer. Although such
deductions are insignificant with todays computers, it was a significant shortcoming for
old systems that were limited to a maximum of 1 to 16 megabytes. Because these shortcomings have diminished over the years, memory mapping has gained popularity in
interfacing.
Addressing
To the microprocessor, the difference between ports and memory is one of perception:
Memory is a direct extension of the chip. Ports are the external world. Writing to I/O
ports is consequently more cumbersome and usually requires more time and microprocessor cycles.
I/O ports give the microprocessor and computer designer greater flexibility, and they give
you a headache when you want to install multimedia accessories.
Implicit in the concept of addressing, whether memory or port addresses, is proper delivery. You expect a letter carrier to bring your mail to your address and not deliver it to
someone elses mailbox. Similarly, computers and their software assume that deliveries of
data and instructions will always go where they are supposed to. To ensure proper delivery, addresses must be correct and unambiguous. If someone types a wrong digit on a
mailing label, it will likely get lost in the postal system.
In order to use port or memory addresses properly, your software needs to know the
proper addresses used by your peripherals. Many hardware functions have fixed or standardized addresses that are the same in every computer. For example, the memory
addresses used by video boards are standardized (at least in basic operating modes), and
the ports used by most hard disk drives are similarly standardized. Programmers can write
the addresses used by this fixed-address hardware into their programs and not worry
whether their data will get where its going.
55
56
The layered BIOS approach was originally designed to eliminate the need for writing
explicit hardware addresses in programs. Drivers accomplish a similar function. They are
written with the necessary hardware addresses built in.
Resource Allocation
The basic hardware devices got assigned addresses and memory ranges early in the history
of the computer and for compatibility reasons have never changed. These fixed values
include those of serial and parallel ports, keyboards, disk drives, and the frame buffer that
stores the monitor image. Add-in devices and more recent enhancements to the traditional devices require their own assignments of system resources. Unfortunately, beyond
the original hardware assignments there are no standards for the rest of the resources.
Manufacturers consequently pick values of their own choices for new products. More
often than youd like, several products may use the same address values.
Manufacturers attempt to avoid conflicts by allowing a number of options for the
addresses used by their equipment. In days gone by, you had to select among the choices
offered by manufacturers using switches or jumpers. Modern expansion products still
require resource allocation but use a software-controlled scheme called Plug-and-Play to
set the values, usually without your intervention or knowledge.
Although a useful computer requires all threesoftware, hardware, and the glue to hold
them togetherhardware plays the enabling role. Without it, the computer doesnt exist.
It is the computer. The rest of this book will examine the individual components of that
hardware as well as related technologies that enable your computer to be the powerful
tool it is.
PART 2
Processing
The difference between a computer and a television set at first seems obviousone has a
keyboard and the other doesnt. Take a look at the modern remote control, however, and
youll see the difference is more subtle. Its something thats going on behind the screens.
When you sit around watching the television, youre gradually putting down roots as a couch
potato or some less respected vegetable, but when you sit at the computer, it watches you. It
wants to know what you want. It listens to your thoughts, relayed through the keyboard, to
carry out your commandscalculating, organizing, alphabetizing, filing away, drawing, and
editing. No other machine takes information from you and handles it so adroitly. Thats the
computers power as a data-processing machine, a thinking machine.
As a physical entity, the processing part of a desktop computer is inside the big box you may
call the system unit, CPU (for central processing unit), or simply the computer. But the processing part of the computer only takes up a fraction of the space occupied by the system
unit. All the essential processing power of the computer fits into a handful of circuit componentsthe microprocessor, the chipset, and the BIOS. Each has its own function in bringing
the processing power of the computer to life, and each is substantially different in concept,
design, and operation.
The microprocessor, the chipset, and the BIOSand the rest of the computer as wellare
themselves built the same. Each is a tiny slice of nearly pure silicon that has been carefully
fabricated into a digital electronic circuit. Because all three processing components of a computer have this same silicon/electrical foundationindeed, all of todays electronic devices
share these same design elementswell talk about the underlying technologies that make the
processing of a computer possible before looking at the individual components themselves.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Digital Electronics
First and foremost, the computer is a thinking machineand that
implies all sorts of preposterous nonsense. The thinking machine could
be a devious device, hatching plots against you as it sits on your desk. A
thinking machine must work the same unfathomable way as the human
mind, something so complicated that in thousands of years of attempts
by the best geniuses, no one has yet satisfactorily explained how it works.
A thinking machine has a brain, so you might suppose that opening it up
and working inside it is brain surgery, and the electronic patient is likely
to suffer irreversible damage at the hands of an unskilled operator.
But computers dont thinkat least not in the same way you and I do or
Albert Einstein did. The computer has no emotions or motivations. The
impulses traveling through it are no odd mixture of chemicals and electrical activity, of activation and repression. The computer deals in simple
pulses of electricity, well understood and carefully controlled. The intimate workings of the computer are probably better understood than the
seemingly simple flame that inhabits the internal combustion engine
inside your car. Nothing mysterious lurks inside the thinking machine
called the computer.
What gives those electrical pulses their thinking power is a powerful
logic system that allows electrical states to represent propositions and
combinations of them. Engineers and philosophers tailored this logic
system to precisely match the capabilities of a computer while, at the
same time, approximating the way you think and express yourselfor at
least how they thought you would think and express yourself. When you
work with a computer, you bend your thoughts to fit into this clever logical system; then the computer manipulates your ideas with the speed
and precision of digital electronics.
62
This logic system doesnt deal with full-fledged ideas such as life, liberty, and the pursuit
of happiness. It works at a much smaller level, dealing with concepts no more complex
than whether a light switch is on or off. But stacking those tiny logical operations together
in more and more complex combinations, building them layer after layer, you can
approach an approximation of human language with which to control the thoughts of the
computer. This language-like control system is the computer program, the software that
makes the computer work.
The logic and electronics of computer circuitry are intimately combined. Engineers
designed the electronics of computers exactly to suit the needs of binary logic. Through
that design, your computer comes to life.
Electronic Design
Computers are thought fearsome because they are based on electrical circuits. Electricity
can be dangerous, as the anyone struck by lightning will attest. But inside the computer,
the danger is low. At its worst, it measures 12 volts, which makes the inside of a computer
as safe as playing with an electric train. Nothing thats readily accessible inside the computer will shock you, straighten your hair, or shorten your life.
Personal computers could not existat least in their current, wildly successful form
were it not for two concepts: binary logic and digital circuitry. The binary approach
reduces data to its most minimalist form, essentially an information quantum. A binary
data bit simply indicates whether something is or is not. Binary logic provides rules for
manipulating those bits to allow them to represent and act like real-world information we
care about, such as numbers, names, and images. The binary approach involves both digitization (using binary data to represent information) and Boolean algebra (the rules for carrying out the manipulations of binary data).
The electrical circuits mimic the logic electrically. Binary logic involves two states, which
a computers digital logic circuitry mimics with two voltage levels.
The logic isnt only what the computer works with; its also what controls the computer.
The same voltages used to represent values in binary logic act as signals that control the
circuits. That means the signals flowing through the computer can control the
computerand the computer can control the signals. In other words, the computer can
control itself. This design gives the computer its power.
Digital Logic Circuitry

The essence of the digital logic that underlies the operation of the microprocessor and
motherboard is the ability to use one electrical signal to control another.
CHAPTER 4 DIGITAL ELECTRONICS
Electronic Design
Certainly there are a myriad of ways of using one electrical signal to control another, as
any student of Rube Goldberg can attest. As interesting and amusing as interspersing cats,
bellows, and bowling balls in the flow of control may be, most engineers have opted for a
more direct system that uses a more direct means based on time-proved electrical technologies.
In modern digital logic circuitry, the basis of this control is amplification, the process of
using a small current to control a larger current (or a small voltage to control a larger
voltage). The large current (or voltage) exactly mimics the controlling current (or voltage)
but is stronger or amplified. In that every change in the large signal is exactly analogous to
each one in the small signal, devices that amplify in this way are called analog. The intensity of the control signal can represent continuously variable informationfor example, a
sound level in stereo equipment. The electrical signal in this kind of equipment is therefore an analogy to the sound that it represents.
In the early years of the evolution of electronic technology, improving this analog amplification process was the primary goal of engineers. After all, without amplification, signals
eventually deteriorated into nothingness. The advent of digital information and the earliest computers made them use the power of amplification differently.
The limiting case of amplification occurs when the control signal causes the larger signal
to go from its lowest value, typically zero, to its highest value. In other words, the large
signal goes off and onswitchesunder control of the smaller signal. The two states of
the output signal (on and off) can be used as part of a binary code that represents information. For example, the switch could be used to produce a series of seven pulses to represent the number 7. Because information can be coded as groups of such numbers (digits),
electrical devices that use this switching technology are described as digital. Note that this
switching directly corresponds to other, more direct control of on/off information, such as
pounding on a telegraph key, a concept well return to in later chapters.
Strictly speaking, an electronic digital system works with signals called high and low, corresponding to a digital one and zero. In formal logic systems, these same values are often
termed true and false. In general, a digital one or logical true corresponds to an electronic
high. Sometimes, however, special digital codes reverse this relationship.
In practical electrical circuits, the high and low signals only roughly correspond to on and
off. Standard digital logic systems define both the high and low signals as ranges of voltages. High is a voltage range near the maximum voltage accepted by the system, and low
is a voltage range near (but not necessarily exactly at or including) zero. A wide range of
undefined voltages spreads between the two, lower than the lowest edge of high but
higher than the highest edge of low. The digital system ignores the voltages in the undefined range. Figure 4.1 shows how the ranges of voltages interrelate.
63
64
FIGURE 4.1
5 Volts
Significance of TTL
voltage levels.
4 Volts
3 Volts
HIGH
2 Volts
UNDEFINED
1 Volts
LOW
0 Volts
Perhaps the most widely known standard is called TTL (for transistor-transistor logic). In
the TTL system, which is still common inside computer equipment, a logical low is any
voltage below 0.8 volts. A logical high is any level above 2.0 volts. The range between 0.8
and 2.0 volts is undefined.
As modern computers shift to lower voltages, the top level of the logical high shifts
downwardfor example, from the old standard of 5.0 volts to the 3.3 volts of the latest
computer equipment (and even lower voltages of new power-conserving microprocessors)but the relationship between the high and low voltages, along with the undefined
range in between, remains the same. Modern systems usually retain the same low and
undefined rangesthey just lop the top off the figure showing the TTL voltage levels.
Electronics
Modern electronics are filled with mysterious acronyms and even stranger-sounding
names. Computer are filled with circuits made from these things, unfamiliar terms such as
CMOS and NMOS, semiconductors, and integrated circuits. A bit of historical perspective will show you what the names means and where (and how) the technologies they
describe originated.
Over the years, electrical engineers have developed a number of ways one signal can control another. The first approach to the electrical control of electrical flow evolved from
the rattling telegraph key. When a telegrapher jammed down on his key to make dots and
dashes, he actually closed an electrical circuit, which sent a voltage down the telegraph
line. At the far end of the connection, this signal powered an electromagnet that snapped
against a piece of iron to create the dot and dash sound.
In 1835 Joseph Henry saw that the electromagnet could be adapted to operate a switch
rather than just pull on a hunk of iron. The electromagnetically actuated switch allowed a
feeble current to switch on a more powerful current. The resulting device, the electrical
relay, was key to the development of the long-distance telegraph. When telegraph signals
got too weak from traveling a great distance, a relay could revitalize them.
Electronic Design
In operation, a relay is just a switch thats controlled by an electromagnet. Activating the

electromagnet with a small current moves a set of contacts that switch the flow of a larger
current. The relay doesnt care whether the control current starts off in the same box as
the relay or from a continent away. As simple in concept as the relay is, its ability to use
one signal to control another proved very powerfulpowerful enough that relays served
as the foundation of some of the first computers (or electrical calculators), such as Bell
Labs 1946 Mark V computer. Relays are still used in modern electrical equipment.
Vacuum Tubes
The vacuum tube improved on the relay design for computers by eliminating the mechanical part of the remote-action switch. Using electronics only, a tube could switch and perform logic operations faster, thousands of times faster, than relays.
Vacuum tubes developed out of Thomas Edisons 1879 invention of the incandescent light
bulb. After the public demonstration of the bulb, Edison continued to work with and
improve it. Along the way, he made a discovery in 1883 (which some historians credit as
Edisons sole original contribution to pure scientific research) of what has come to be
called the Edison Effect. Edison noted he could make a current flow through the vacuum in
the bulb from the filament to a metal plate he had introduced inside.
The Edison Effect remained a curiosity until 1904 when John Ambrose Fleming created
the diode vacuum tube. Fleming found that electrons would flow from the negatively
charged hot filament of the light bulb to a positively charged cold collector plate, but not
in the other direction. Fleming made an electrical one-way street that could operate as a
rectifier to change alternating current into direct current or as a detector that pulled modulation from carrier waves (by stripping off the carrier waves alternating cycles).
In 1907 Lee De Forest created the Audion, now known as the triode tube. De Forest
introduced an element of control to the bulb-cum-diode. He interposed a control grid
between the hot filament (the cathode) and the cold plate (the anode). De Forest found
that he could control the electron flow by varying the voltage he applied to the control
grid.
The grid allowed the Audion to harness the power of the attraction of unlike electrical
charges and the repulsion of like charges, enabling a small charge to control the flow of
electrons through the vacuum inside the tube. In the Audion, as with the relay, a small
voltage could control a much larger voltage. De Forest created the first electronic amplifier, the basis of all modern electronics.
The advantage of the vacuum tube over the relay in controlling signals is speed. The relay
operates at mechanical rates, perhaps a few thousand operations per second. The vacuum
tube can switch millions of times per second. The first recognizable computers (such as
ENIAC) were built from thousands of tubes, each configured as a digital logic gate.
65
66
Semiconductors
Using tube-based electronics in computers is fraught with problems. First is the spaceheater effect: Tubes have to glow like light bulbs to work, and they generate heat along
the way, enough to smelt rather than process data. And, like light bulbs, tubes burn out.
Large tube-based computers required daily shutdown and maintenance as well as several
technicians on the payroll. ENIAC was reported to have a mean time between failures of
5.6 hours.
In addition, tube circuits are big. ENIAC filled a room, yet the house-sized computers of
1950s vintage science fiction would easily be outclassed in computing power by todays
desktop machines. In the typical tube-based computer design, one logic gate required one
tube that took up considerably more space than a single microprocessor with tens of millions of logic gates. Moreover, physical size isnt only a matter of housing. The bigger the
computer, the longer it takes its thoughts to travel through its circuitseven at the speed
of lightand the more slowly it thinks.
Making todays practical computers took another true breakthrough in electronics: the
transistor, first created at Bell Laboratories in 1947 and announced in 1948 by the team of
John Bardeen, Walter Brattain, and William Shockley. A tiny fleck of germanium (later,
silicon) formed into three layers, the transistor was endowed with the capability to let one
electrical current applied to one layer alter the flow of another, larger current between the
other two layers. Unlike the vacuum tube, the transistor needed no hot electrons because
the current flowed entirely through a solid materialthe germanium or siliconhence,
the common name for tubeless technology, solid-state electronics.
Germanium and silicon are special materialsactually, metalscalled semiconductors. The
term describes how these materials resist the flow of electrical currents. A true electrical
conductor (such as the copper in wires) hardly resists the flow of electricity, whereas a
non-conductor (or insulator, such as the plastic wrapped around the wires) almost totally
prevents the flow of electricity. Semiconductors allow somebut not muchelectricity to
flow.
By itself, being a poor but not awful electrical conductor is as remarkable as lukewarm
water. However, infusing atoms of impurities into the semiconductors microscopic lattice
structure dramatically alters the electrical characteristics of the material and makes solidstate electronics possible.
This process of adding impurities is called doping. Some impurities add extra electrons
(carriers of negative charges) to the crystal. A semiconductor doped to be rich in electrons
is called an N-type semiconductor. Other impurities in the lattice leave holes where electrons
would ordinarily be, and these holes act as positive charge carriers. A semiconductor
doped to be rich in holes is called a P-type semiconductor.
Electronic Design
Electricity easily flows across the junction between the two materials when an N-type
semiconductor on one side passes electrons to the holes in a P-type semiconductor on the
other side. The empty holes willingly accept the electrons. They just fit right in. But electricity doesnt flow well in the opposite direction, however. If the P-type semiconductors
holes ferry electrons to the N-type material at the junction, the N-type semiconductor
will refuse delivery. It already has all the electrons it needs. It has no place to put any
more. In other words, electricity flows only in one direction through the semiconductor
junction, just as it flows only one way through a vacuum-tube diode.
The original transistor incorporated three layers with two junctions between dissimilar
materials, stacked in layers as N-P-N or P-N-P. Each layer has its own name: The top is
the emitter, the middle of the sandwich is the gate, and the bottom is the collector. (The
structure of a transistor isnt usually a three-layer cake with top and bottom, but thats
effectively how it works.)
Ordinarily no electricity could pass through such a stack from emitter to collector because
the two junctions in the middle are oriented in opposite directions. One blocks electrical
flow one way, and the second blocks the flow in the other direction.
The neat trick that makes a transistor work is changing the voltage on the middle layer.
Say you have a P-N-P transistor. Your wire dumps electrons into the holes in the P-layer,
and they travel to the junction with the N-layer. The N-layer is full of electrons, so it
wont let the current flow further. But if you drain off some of those electrons through the
gate, current can flow through the junctionand it can keep flowing through the next
junction as well. It only takes a small current to drain electrons from the gate to permit a
large current to flow from emitter to collector.
The design of transistor circuits is more complicated than our simple example. Electrical
flow through the transistor depends on the complex relationships between voltages on its
junctions, and the electricity doesnt necessarily have to flow from emitter to collector. In
fact, the junction transistor design, although essential to the first transistors, is rarely used
in computer circuits today. But the junction transistor best illustrates the core principles of
all solid-state electronics.
Modern computer circuits mostly rely on a kind of transistor in which the electrical current flow through a narrow channel of semiconductor material is controlled by a voltage
applied to a gate (which surrounds the channel) made from metal oxide. The most common variety of these transistors is made from N-type material and results in a technology
called NMOS, an acronym for N-channel Metal Oxide Semiconductor. A related technology combines both N-channel and P-channel devices and is called CMOS
(Complementary Metal Oxide Semiconductor) because the N-and P-type materials are
complements (opposites) of one another. These namesCMOS particularlypop up
occasionally in discussions of electronic circuits.
67
68
The typical microprocessor once was built from NMOS technology. Although NMOS
designs are distinguished by their simplicity and small size (even on a microchip level),
they have a severe shortcoming: They constantly use electricity whenever their gates are
turned on. Because about half of the tens or hundreds of thousands of gates in a microprocessor are switched on at any given time, an NMOS chip can draw a lot of current.
This current flow creates heat and wastes power, making NMOS unsuitable for miniaturized computers (which can be difficult to cool) and battery-operated equipment, such as
notebook computers.
Some earlier and most contemporary microprocessors now use CMOS designs. CMOS is
inherently more complex than NMOS because each gate requires more transistors, at least
a pair per gate. But this complexity brings a benefit: When one transistor in a CMOS gate
is turned on, its complementary partner is switched off, thus minimizing the current flow
through the complementary pair that make up the circuit. When a CMOS gate is idle,
just maintaining its state, it requires almost no power. During a state change, the current
flow is large but brief. Consequently, the faster the CMOS gate changes state, the more
current that flows through it and the more heat it generates. In other words, the faster a
CMOS circuit operates, the hotter it becomes. This speed-induced temperature rise is one
of the limits on the operating speed of many microprocessors.
CMOS technology can duplicate every logic function made with NMOS but with a substantial saving of electricity. On the other hand, manufacturing costs somewhat more
because of the added circuit complexity.
Integrated Circuits
The transistor overcomes several of the problems with using tubes to make a computer.
Transistors are smaller than tubes and give off less heat because they dont need to glow to
work. But every logic gate still requires one or more transistors (as well as several other
electronic components) to build. If you allocated a mere square inch to every logic gate,
the number of logic gates in a personal computer microprocessor such as the Pentium 4
(about fifty million) would require a circuit board on the order of 600 square feet.
At the very end of the 1950s, Robert N. Noyce at Fairchild Instrument and Jack S. Kilby
of Texas Instruments independently came up with the same brilliant idea of putting multiple semiconductor devices into a single package. Transistors are typically grown as crystals
from thin-cut slices of silicon called wafers. Typically, thousands of transistors are grown at
the same time on the same wafer. Instead of carving the wafer into separate transistors,
the engineer linked them together (integrated them) to create a complete electronic circuit all on one wafer. Kilby linked the devices with micro-wires; Noyce envisioned fabricating the interconnecting circuits between devices on the silicon itself. The resulting
electronic device, for which Noyce applied for a patent on July 30, 1959, became known
as the integrated circuit, or IC. Such devices now are often called chips because of their
Electronic Design
construction from a single small piece of silicona chip off the old crystal. Integrated circuit technology has been adapted to both analog and digital circuitry. Their grandest
development, however, is the microprocessor.
Partly because of the Noyce invention, Fairchild flourished as a semiconductor manufacturer throughout the 1960s. The company was acquired by Schlumberger Ltd. in 1979,
which sold it to National Semiconductor in 1987. In 1996, National spun off Fairchild as
an independent manufacturer once again, and the developer of the integrated circuit continues to operate as an independent business, Fairchild Semiconductor Corporation, based
in South Portland, Maine.
The IC has several advantages over circuits built from individual (or discrete) transistors,
most resulting from miniaturization. Most importantly, integration reduces the amount of
packaging. Instead of one metal or plastic transistor case per logic gate, multiple gates
(even millions of them) can be combined into one chip package.
Because the current inside the chip need not interact with external circuits, the chips can
be made arbitrarily small, enabling the circuits to be made smaller, too. In fact, today the
limit on the size of elements inside an integrated circuit is mostly determined by fabrication technology; internal circuitry is as small as todays manufacturing equipment can
make it affordably. The latest Intel microprocessors, which use integrated circuit technology, incorporate the equivalent of over 50 million transistors using interconnections that
measure less than 0.13 of a micron (millionths of a meter) across.
In the past, a hierarchy of names was given to ICs depending on the size of circuit elements. Ordinary ICs were the coarsest in construction. Large-scale integration (LSI) put
between 500 and 20,000 circuit elements together; very large-scale integration (VLSI) put
more than 20,000 circuit elements onto a single chip. All microprocessors use VLSI technology, although the most recent products have become so complex (Intels Pentium 4, for
example, has the equivalent of about 50 million transistors inside) that a new term has
been coined for them, ultra large-scale integration (ULSI).
Moores Law
The development of the microprocessor often is summed up by quoting Moores Law,
which sounds authoritative and predicts an exponential acceleration in computer power.
At least thats how Moores Law is most often described when you stumble over it in
books and magazines. The most common interpretation occurring these days is that
Moores Law holds that computer power doubles every 18 months.
In truth, youll never find the law concisely quoted anywhere. Thats because its a slippery
thing. It doesnt say what most people think. As originally formulated, it doesnt even
apply to microprocessors. Thats because Gordon E. Moore actually created his law well
69
70
before the microprocessor was invented. In its original form, Moores Law describes only
how quickly the number of transistors in integrated circuits was growing.
What became known as Moores Law was first published in the industry journal Electronics
on April 19, 1965, in a paper titled Cramming More Components into Integrated
Circuits, written when Moore was director of the research and development laboratories
at Fairchild Semiconductor. His observationand what Moores Law really sayswas that
the number of transistors in the most complex circuits tended to double every year.
Moores conclusion was that by 1975, ten years after he was writing, economics would
force semiconductor makers to squeeze as many as 65,000 transistors onto a single silicon
chip.
Although Moores fundamental premise, that integrated circuit complexity increases exponentially, was accurate (although not entirely obvious at the time), the actual numbers
given in Moores predictions in the paper missed the mark. He was quite optimistic. For
example, over the 31-year history of the microprocessor from 4004 to Pentium 4, the
transistor count has increased by a factor of 18,667 (from 2250 to 42 million). Thats
approximately a doubling every two years.
Thats more of a difference than you might think. At Moores rate, a microprocessor today
would have over four trillion transistors inside. Even the 18-month doubling rate so often
quoted would put about 32 billion transistors in every chip. In either case, youre talking
real computer power.
Over the years, people have gradually adapted Moores Law to better suit the factsin
other words, Moores Law doesnt predict how many transistors will be in future circuits.
It only describes how circuit complexity has increased.
The future of the law is cloudier still. A constant doubling of transistor count requires
ever-accelerating spending on circuit design research, which was possible during the 20
years of the personal computer boom but may suffer as the rapid growth of the industry
falters. Moreover, the lawor even a linear increase in the number of transistors on a
chipultimately bumps into fundamental limits imposed by the laws of physics. At some
point the increase in complexity of integrated circuits, taken to extreme, will require circuit elements smaller than atoms or quarks, which most physicists believe is not possible.
Printed Circuits
An integrated circuit is like a gear of a complex machine. By itself it does nothing. It must
be connected to the rest of the mechanism to perform its appointed task. The integrated
circuit needs a means to acquire and send out the logic and electrical signals it manipulates. In other words, each integrated circuit in a computer must be logically and electrically connectedessentially that means linked by wires.
Electronic Design
In early electrical devices, wires in fact provided the necessary link. Each wire carried one
signal from one point to another, creating a technology called point-to-point wiring.
Because people routinely soldered together these point-to-point connections by hand
using a soldering iron, they were sometimes called hand-wired. This was a workable, if not
particularly cost-effective, technology in the days of tubes, when even a simple circuit
spanned a few inches of physical space. Today, point-to-point wiring is virtually inconceivable because a computer crams the equivalent of half a million tube circuits into a few
square inches of space. Connecting them with old-fashioned wiring would take a careful
hand and some very fine wire. The time required to cut, strip, and solder in place each
wire would make building a single computer a lifetime endeavor.
Long before the introduction of the first computer, engineers found a better way of
wiring together electrical devicesthe printed circuit board. The term is sometimes confusingly shortened to computer board, even when the board is part of some other, non-computer device. Today, printed circuit boards are the standard from which nearly all
electronic devices are made. The board in the name motherboard results from the
assembly being a printed circuit board.
Fabrication
Printed circuit board technology allows all the wiring for an entire circuit assembly to be
fabricated together in a quick process that can be entirely mechanized. The wires themselves are reduced to copper traces, a pattern of copper foil bonded to the substrate that
makes up the support structure of the printed circuit board. In computers, this substrate is
usually green composite material called glass-epoxy, because it has a woven glass fiber base
thats filled and reinforced with an epoxy plastic. Less-critical electronic devices (read
cheap) substitute a simple brownish substrate of phenolic plastic for the glass-epoxy.
The simplest printed circuit boards start life as a sheet of thin copper foil bonded to a
substrate. The copper is coated with a compound called photo-resist, a light-sensitive material. When exposed to light, the photo-resist becomes resistant to the effects of compounds, such as nitric acid, that strongly react with copper. A negative image of the
desired final circuit pattern is placed over the photo-resist covered copper and exposed to
a strong light source. This process is akin to making a print of a photograph. The exposed
board is then immersed in an etchant, one of those nasty compounds that etch or eat away
the copper that is not protected by the light-exposed photo-resist. The result is a pattern
of copper on the substrate corresponding to the photographic original. The copper traces
can then be used to connect the various electronic components that will make up the final
circuit. All the wiring on a circuit board is thus fabricated in a single step.
When the electronic design on a printed circuit board is too complex to be successfully
fabricated on one side of the substrate, engineers can switch to a slightly more complex
technology to make two-sided boards. The traces on each side are separately exposed but
71
72
etched during the same bath in etchant. In general, the circuit traces on one side of the
board run parallel in one direction, and the traces on the other side run generally perpendicular. The two sides get connected together by components inserted through the board
or through plated-through holes, holes drilled through the board and then filled with solder
to provide an electrical connection
To accommodate even more complex designs, engineers have designed multilayer circuit
boards. These are essentially two or more thin double-sided boards tightly glued together
into a single assembly. Most computer system boards use multilayer technology, both to
accommodate complex designs and to improve signals characteristics. Sometimes a layer is
left nearly covered with copper to shield the signal in the layers from interacting with one
another. These shielding layers are typically held at ground potential and are consequently
called ground planes.
One of the biggest problems with the multilayer design (besides the difficulty in fabrication) is the difficulty in repair. Abnormally flexing a multilayer board can break one of the
traces hidden in the center of the board. No reasonable amount of work can repair such
damage.
Pin-in-Hole Technology
Two technologies are in wide use for attaching components to the printed circuit board.
The older technology is called pin-in-hole. Electric drills bore holes in the circuit board at
the points where the electronic components are to attach. Machines (usually) push the
leads (wires that come out of the electronic components) into and through the circuit
board holes and bend them slightly so that they hold firmly in place. The components are
then permanently fixed in place with solder, which forms both a physical and electrical
connection. Figure 4.2 shows the installation of a pin-in-hole electronic component.
FIGURE 4.2
Electronic component (resistor)
Pin-in-hole component technology.
Circuit trace
Component lead
Most mass-produced pin-in-hole boards use wave-soldering to attach pin-in-hole components. In wave-soldering, a conveyer belt slides the entire board over a pool of molten solder (a tin and lead alloy), and a wave on the solder pool extends up to the board, coating
Electronic Design
the leads and the circuit traces. When cool, the solder holds all the components firmly in
place.
Workers can also push pin-in-hole components into circuit boards and solder them individually in place by hand. Although hand fabrication is time consuming and expensive, it
can be effective when a manufacturer requires only a small number of boards. Automatic
machinery cuts labor costs and speeds production on long runs; assembly workers typically make prototypes and small production runs or provide the sole means of assembly
for tiny companies that can afford neither automatic machinery nor farming out their circuit board work.
Surface Mount Technology

The newer method of attaching components, called surface-mount technology, promises
greater miniaturization and lower costs than pin-in-hole. Instead of holes to secure them,
surface-mount components are glued to circuit boards using solder flux or paste, which
temporarily holds them in place. After all the components are affixed to a circuit board in
their proper places, the entire board assembly runs through a temperature-controlled
oven, which melts the solder paste and firmly solders each component to the board.
Figure 4.3 illustrates surface-mount construction.
FIGURE 4.3
Surface mount circuit
board construction.
Surface mount
component
Component lead
Solder flux
Board trace
Printed circuit board
Surface mount components are smaller than their pin-in-hole kin because they dont need
leads. Manufacturing is simpler because theres no need to drill holes in the circuit boards.
Without the need for large leads, the packages of the surface-mount components can be
smaller, so more components will fit in a given space with surface mount technology.
On the downside, surface mount fabrication doesnt lend itself to small production runs or
prototyping. It can also be a headache for repair workers. They have to squint and peer at
components that are often too small to be handled without tweezers and a lot of luck.
Moreover, many surface-mount boards also incorporate some pin-in-hole components, so
they still need drilling and wave-soldering.
73
74
Logic Design
Reduced to its fundamental principles, the workings of a modern silicon-based microprocessor are not difficult to understand. They are simply the electronic equivalent of a
knee-jerk. Every time you hit the microprocessor with an electronic hammer blow (the
proper digital input), it reacts by doing a specific something. Like a knee-jerk reflex, the
microprocessors reaction is always the same. When hammered by the same input and
conditions, it kicks out the same function.
The complexity of the microprocessor and what it does arises from the wealth of inputs it
can react to and the interaction between successive inputs. Although the microprocessors
function is precisely defined by its input, the output from that function varies with what
the microprocessor has to work on, and that depends on previous inputs. For example, the
result of you carrying out a specific commandSimon says lift your left legwill differ
dramatically depending on whether the previous command was Simon says sit down or
Simon says lift your right leg.
The rules for controlling the knee-jerks inside a computer are the rules of logic, and not
just any logic. Computers use a special symbolic logic system that was created about a
century and a half ago in the belief that human thinking could be mechanized much as the
production of goods had been mechanized in the Industrial Revolution.
Boolean Logic
As people began to learn to think again after the Dark Ages, they began exploring mathematics, first in the Arabian world, then Europe. They developed a rigorous, objective
systemone that was reassuring in its ability to replicate results. Carry out the same operation on the same numbers, and you always got the same answer. Mathematics delivered a
certainty that was absent from the rest of the world, one in which dragons inhabited the
unknown areas of maps and people had no conception that micro-organisms might cause
disease.
Applying the same rigor and structure, scientific methods first pioneered by the Greeks
were rediscovered. The objective scientific method found truth, the answers that eluded
the world of superstition. Science led to an understanding of the world, new processes,
new machines, and medicine.
In Victorian England, philosophers wondered whether the same objectivity and rigor
could be applied to all of human thought. A mathematician, George Boole, first proposed
applying the rigorous approach of algebra to logical decision-making. In 1847, Boole
founded the system of modern symbolic logic that we now term Boolean logic (alternately,
Logic Design
Boolean algebra). In his system, Boole reduced propositions to symbols and formal operators that followed the strict rules of mathematics. Using his rigorous approach, logical
propositions could be proven with the same certainty as mathematical equations.
Philosophers, including Ludwig Wittgenstein and Bertrand Russell, further developed the
concept of symbolic logic and showed that anything that could be known could be
expressed in its symbols. By translating what you knew and wanted to know into symbols,
you could apply the rules of Boolean logic and find an answer. Knowledge was reduced to
a mechanical process, and that made it the province of machines.
In concept, Babbages Analytical Engine could have deployed Booles symbolic logic and
become the first thinking machine. However, neither the logic nor the hardware was up to
the task at the time. But when fast calculations became possible, Boolean logic proved key
to programming the computers that carried out the tasks.
Logic Gates
Giving an electrical circuit the power to make a decision isnt as hard as you might think.
Start with that same remote signaling of the telegraph but add a mechanical arm that links
it to a light switch on your wall. As the telegraph pounds, the light flashes on and off.
Certainly youll have done a lot of work for a little return, in that the electricity could be
used to directly light the bulb. There are other possibilities, however, that produce
intriguing results. You could, for example, pair two weak telegraph arms so that their joint
effort would be required to throw the switch to turn on the light. Or you could link the
two telegraphs so that a signal on either one would switch on the light. Or you could
install the switch backwards so that when the telegraph is activated, the light would go out
instead of come on.
These three telegraph-based design examples actually provide the basis for three different
types of computer circuits, called logic gates (the AND, OR, and NOT gates, respectively).
These electrical circuits are called gates because they regulate the flow of electricity, allowing it to pass through or cutting it off, much as a gate in a fence allows or impedes your
own progress. These logic gates endow the electrical assembly with decision-making
power. In the light example, the decision is necessarily simple: when to switch on the
light. But these same simple gates can be formed into elaborate combinations that make
up a computer that can make complex logical decisions.
The three logic gates can perform the function of all the operators in Boolean logic. They
form the basis of the decision-making capabilities of the computer as well as other logic
circuitry. Youll encounter other kinds of gates, such as NAND (short for Not AND),
NOR (short for Not OR), and XOR (for Exclusive OR), but you can build any one of
the others from the basic three: AND, OR, and NOT.
75
76
In computer circuits, each gate requires at least one transistor. A microprocessor with ten
million transistors may have nearly that many gates.
Memory
These same gates also can be arranged to form memory. Start with the familiar telegraph.
Instead of operating the current for a light bulb, however, reroute the wires from the
switch so that they, too, link to the telegraphs electromagnet. In other words, when the
telegraph moves, it throws a switch that supplies itself with electricity. Once the telegraph
is supplying itself with electricity, it will stay on using that power even if you switch off
the original power that first made the switch. In effect, this simple system remembers
whether it has once been activated. You can go back at any time and see if someone has
ever sent a signal to the telegraph memory system.
This basic form of memory has one shortcoming: Its elephantine and never forgets.
Resetting this memory system requires manually switching off both the control voltage
and the main voltage source.
A more useful form of memory takes two control signals: One switches it on, the other
switches it off. In simplest form, each cell of this kind of memory is made from two
latches connected at cross-purposes so that switching one latch on cuts the other off.
Because one signal sets this memory to hold data and the other one resets it, this circuit is
sometimes called set-reset memory. A more common term is flip-flop because it alternately
flips between its two states. In computer circuits, this kind of memory is often simply
called a latch. Although the main memory of your computer uses a type of memory that
works on a different electrical principal, latch memory remains important in circuit
design.
Instructions
Although the millions of gates in a microprocessor are so tiny that you cant even discern
them with an optical microscope (you need at least an electron microscope), they act
exactly like elemental, telegraph-based circuits. They use electrical signals to control other
signals. The signals are just more complicated, reflecting the more elaborate nature of the
computer.
Todays microprocessors dont use a single signal to control their operations, rather, they
use complex combinations of signals. Each microprocessor command is coded as a pattern
of signals, the presence or absence of an electrical signal at one of the pins of the microprocessors package. The signal at each pin represents one bit of digital information.
The designers of a microprocessor give certain patterns of these bit-signals specific meanings. Each pattern is a command called a microprocessor instruction that tells the microprocessor to carry out a specific operation. The bit pattern 0010110, for example, is the
Logic Design
instruction that tells an Intel 8086-family microprocessor to subtract in a very explicit

manner. Other instructions tell the microprocessor to add, multiply, divide, move bits or
bytes around, change individual bits, or just wait around for another instruction.
Microprocessor designers can add instructions to do just about anythingfrom matrix
calculations to brewing coffee (that is, if the designers wanted to, if the instructions actually did something useful, and if they had unlimited time and resources to engineer the
chip). Practical concerns such as keeping the design work and the chip manageable constrain the range of commands given to a microprocessor.
The entire repertoire of commands that a given microprocessor model understands and
can react to is called that microprocessors instruction set or its command set. The designer
of the microprocessor chooses which pattern to assign to a given function. As a result, different microprocessor designs recognize different instruction sets, just as different board
games have different rules.
Despite their pragmatic limits, microprocessor instruction sets can be incredibly rich and
diverse, and the individual instructions incredibly specific. The designers of the original
8086-style microprocessor, for example, felt that a simple command to subtract was not
enough by itself. They believed that the microprocessor also needed to know what to subtract from what and what it should do with the result. Consequently, they added a rich
variety of subtraction instructions to the 8086 family of chips that persists into todays
Athlon and Pentium 4 chips. Each different subtraction instruction tells the microprocessor to take numbers from different places and find the difference in a slightly different
manner.
Some microprocessor instructions require a series of steps to be carried out. These multistep commands are sometimes called complex instructions because of their composite
nature. Although a complex instruction looks like a simple command, it may involve much
work. A simple instruction would be something such as pound a nail. A complex
instruction may be as far ranging as frame a house. Simple subtraction or addition of
two numbers may actually involve dozens of steps, including the conversion of the numbers from decimal to the binary (ones and zeros) notation that the microprocessor understands. For instance, the previous sample subtraction instruction tells one kind of
microprocessor that it should subtract a number in memory from another number in the
microprocessors accumulator, a place thats favored for calculations in todays most popular microprocessors.
Registers
Before the microprocessor can work on numbers or any other data, it first must know
what numbers to work on. The most straightforward method of giving the chip the variables it needs would seem to be supplying more coded signals at the same time the
77
78
instruction is given. You could dump in the numbers 6 and 3 along with the subtract
instruction, just as you would load laundry detergent along with shirts and sheets into
your washing machine. This simple method has its shortcomings, however. Somehow the
proper numbers must be routed to the right microprocessor inputs. The microprocessor
needs to know whether to subtract 6 from 3 or 3 from 6 (the difference could be significant, particularly when youre balancing your checkbook).
Just as you distinguish the numbers in a subtraction problem by where you put them in
the equation (63 versus 36), a microprocessor distinguishes the numbers on which it
works by their position (where they are found). Two memory addresses might suffice were
it not for the way most microprocessors are designed. They have only one pathway to
memory, so they can effectively see only one memory value at a time. So instead, a
microprocessor loads at least one number to an internal storage area called a register. It
can then simultaneously reach both the number in memory and the value in its internal
register. Alternatively (and more commonly today), both values on which the microprocessor is to work can be loaded into separate internal registers.
Part of the function of each microprocessor instruction is to tell the chip which registers
to use for data and where to put the answers it comes up with. Other instructions tell the
chip to load numbers into its registers to be worked on later or to move information from
a register someplace else (for instance, to memory or an output port).
A register functions both as memory and a workbench. It holds bit-patterns until they can
be worked on or sent out of the chip. The register is also connected with the processing
circuits of the microprocessor so that the changes ordered by instructions actually appear
in the register. Most microprocessors typically have several registers, some dedicated to
specific functions (such as remembering which step in a function the chip is currently carrying out; this register is called a counter or instruction pointer) and some designed for general purposes. At one time, the accumulator was the only register in a microprocessor that
could manage calculations. In modern microprocessors, all registers are more nearly equal
(in some of the latest designs, all registers are equal, even interchangeable), so the accumulator is now little more than a colorful term left over from a bygone era.
Not only do microprocessors have differing numbers of registers, but the registers may
also be of different sizes. Registers are measured by the number of bits that they can work
with at one time. A 16-bit microprocessor, for example, should have one or more registers
that each holds 16 bits of data at a time. Todays microprocessors have 32- or 64-bit registers.
Adding more registers to a microprocessor does not make it inherently faster. When a
microprocessor lacks advanced features such as pipelining or superscalar technology (discussed later in this chapter), it can perform only one operation at a time. More than two
Logic Design
registers would seem superfluous. After all, most math operations involve only two numbers at a time (or can be reduced to a series of two-number operations). Even with oldtechnology microprocessors, however, having more registers helps the software writer
create more efficient programs. With more places to put data, a program needs to move
information in and out of the microprocessor less often, which can potentially save several
program steps and clock cycles.
Modern microprocessor designs, particularly those influenced by the latest research into
design efficiency, demand more registers. Because microprocessors run much faster than
memory, every time the microprocessor has to go to memory, it must slow down.
Therefore, minimizing memory accessing helps improve performance. Keeping data in
registers instead of memory speeds things up.
The width of the registers also has a substantial effect on the performance of a microprocessor. The more bits assigned to each register, the more information that the microprocessor can process in every cycle. Consequently, a 64-bit register in the next
generation of microprocessor chips holds the potential of calculating eight times as fast as
an 8-bit register of a first generation microprocessorall else being equal.
Programs
A computer program is nothing more than a list of instructions. The computer goes
through the instruction list of the program step by step, executing each one in turn. Each
builds on the previous instructions to carry out a complex function. The program is essentially a recipe for a microprocessor or the step-by-step instructions in a how-to manual.
The challenge for the programmer is to figure out into which steps to break a complex
job and to arrange those steps in the best possible order. It can be a big job. Although a
program can be as simple as a single step (say, stop), a modern program or software package may comprise millions or tens of millions of steps. They are quite literally too complex for a single human being to understandor write. They are joint efforts. Not just the
work of many people, but the work of people and machines using development environments to divide up the work and take advantage of routines and libraries created by other
teams. A modern software package is the result of years of work in putting together simple
microprocessor instructions.
One of the most important concepts in the use of modern personal computers is multitasking,
the ability to run multiple programs at the same time, shifting your focus from one to
another. You can, for example, type a term paper on L. Frank Baum and the real meaning of
the Wizard of Oz using your word processor while your MP3 program churns out a techno
version of Over the Rainbow through your computers speakers. Today, you take that kind
of thing for granted. But thinking about it, this doesnt make sense in the context of computer programs being simple lists of instructions and your microprocessor executing the
instructions, one by one, in order. How can a computer do two things at the same time?
79
80
The answer is easy. It cannot. Computers do, in fact, process instructions as a single list.
Computers can, however, switch between lists of instructions. They can execute a series of
instructions from one list, shift to another list for a time, and then shift back to the first
list. You get the illusion that the computer is doing several things at the same time
because it shifts between instruction lists very quickly, dozens of times a second. Just as
the separate frames of an animated cartoon blur together into an illusion of continuous
motion, the computer switches so fast you cannot perceive the changes.
Multitasking is not an ability of a microprocessor. Even when youre doing six things at
once on your computer, the microprocessor is still doing one thing at a time. It just runs
all those instructions. The mediator of the multitasking is your operating system, basically
a master program. It keeps track of every program (and subprogram) thats running
including itselfand decides which gets a given moment of the microprocessors time for
executing instructions.
Interrupts
Give a computer a program, and its like a runaway freight train. Nothing can stop it. It
keeps churning through instructions until it reaches the last one. Thats great if what you
want is the answer, whenever it may arrive. But if you have a task that needs immediate
attentionsay a block of data has just arrived from the Internetyou dont want to wait
forever for one program to end before you can start another.
To add immediacy and interactivity to microprocessors, chip designers incorporate a feature called the interrupt. An interrupt is basically a signal to the microprocessor to stop
what it is doing and turn its attention to something else. Intel microprocessors understand
two kinds of interrupts: software and hardware.
A software interrupt is simply a special instruction in a program thats controlling the
microprocessor. Instead of adding, subtracting, or whatever, the software interrupt causes
program execution to temporarily shift to another section of code in memory.
A hardware interrupt causes the same effect but is controlled by special signals outside of the
normal data stream. The only problem is that the microprocessors recognize far fewer
interrupts than would be usefulonly two interrupt signal lines are provided. One of these
is a special case, the Non-Maskable Interrupt. The other line is shared by all system interrupts. The support hardware of your computer multiplies the number of hardware interrupts so that all devices that need them have the ability to interrupt the microprocessor.
Clocked Logic
Microprocessors do not carry out instructions as soon as the instruction code signals reach
the pins that connect the microprocessor to your computers circuitry. If chips did react
immediately, they would quickly become confused. Electrical signals cannot change state
Logic Design
instantly; they always go through a brief, though measurable, transition perioda period
of indeterminate voltage level during which the signals would probably perplex a microprocessor into a crash. Moreover, all signals do not necessarily change at the same rate, so
when some signals reach the right values, others may still be at odd values. As a result, a
microprocessor must live through long periods of confusion during which its signals are at
best meaningless, at worst dangerous.
To prevent the microprocessor from reacting to these invalid signals, the chip waits for an
indication that it has a valid command to carry out. It waits until it gets a Simon says
signal. In todays computers, this indication is provided by the system clock. The clock
sends out regular voltage pulses, the electronic equivalent of the ticking of a grandfathers
clock. The microprocessor checks the instructions given to it each time it receives a clock
pulse, providing it is not already busy carrying out another instruction.
Early microprocessors were unable to carry out even one instruction every clock cycle.
Vintage microprocessors may require as many as 100 discrete steps (and clock pulses) to
carry out a single instruction. The number of cycles required to carry out instructions
varies with the instruction and the microprocessor design. Some instructions take a few
cycles, others dozens. Moreover, some microprocessors are more efficient than others in
carrying out their instructions. The trend today is to minimize and equalize the number
of clock cycles needed to carry out a typical instruction.
When you want to squeeze every last bit of performance from a computer, you can sometimes tinker with its timing settings. You can up the pace at which its circuits operate, thus
making the system faster. This technique also forces circuits to operate at speeds higher
than they were intended, thus compromising the reliability of the computers operations.
Tinkers dont worry about such things, believing that most circuits have such a wide safety
margin that a little boost will do no harm. The results of their work may delight them
they might eke 10 percent or more extra performance from a computerbut these results
might also surprise them when the system operates erratically and shuts down randomly.
This game is called overclocking because it forces the microprocessor in the computer to
operate at a clock speed thats over its ratings.
Overclocking also takes a more insidious form. Unscrupulous semiconductor dealers
sometimes buy microprocessors (or memory chips or other speed-rated devices) and
change their labels to reflect higher-speed potentials (for example, buying a 2.2GHz
Pentium 4 and altering its markings to say 2.53GHz). A little white paint increases the
market value of some chips by hundreds of dollars. It also creates a product that is likely
to be operated out of its reliable range. Intel introduced internal chip serial numbers with
the Pentium III to help prevent this form of fraud. From the unalterable serial number of
the chip, the circuitry of a computer can figure out the factory-issue speed rating of the
chip and automatically adjust itself to the proper speed.
81
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Microprocessors
Microprocessors are the most complicated devices ever created by
human beings. But dont despair. A microprocessor is complicated the
same way a castle you might build from Legos is complicated. You can
easily understand each of the individual parts of a microprocessor and
how they fit together, just as you might build a fort or castle from a
legion of blocks of assorted shapes. The metaphor is more apt than you
might think. Engineers, in fact, design microprocessors as a set of functional blocksnot physical blocks like Legos, but blocks of electronic
circuits that perform specific functions. And they dont fit them together
on the floor of a giant playroom. Most of their work is mental, a lot of it
performed on computers. Yes, you need a computer to design a computer these days. Using a computer is faster; it can give engineers insight
by showing them ideas they could only before imagine, and it doesnt
leave a mess of blocks on the floor that Mom makes them clean up.
In creating a microprocessor, a team of engineers may work on each one
of the blocks that makes up the final chip, and the work of one team may
be almost entirely unfathomable by another team. Thats okay, because
all that teams need to know about the parts besides their own is what
those other parts do and how they fit together, not how they are made.
All the complex detailsthe logic gates from which each block is built
are irrelevant to anyone not designing that block. Its like those Legos.
You might know a Lego block is made from molecules of plastic, but you
dont need to know about the carbon backbones and the side-chains that
make up each molecule to put the blocks together.
84
Legos come in different shapes. The functional blocks of a microprocessor have different
functions. Well start our look at microprocessors by defining what those functions are.
Then well leave the playroom behind and look at how you make a microprocessor operate by examining the instructions that it uses. This detour is more important than you
might think. Much of the design work in creating a series of microprocessors goes into
deciding exactly what instructions it must carry outwhich instructions are most useful
for solving problems and doing it quickest. From there, well look at how engineers have
used their imaginations to design ever-faster microprocessors and at the tricks they use to
coax more speed for each new generation of chip.
Microprocessor designers cant just play with theory in creating microprocessors. They
have to deal with real-world issues. Somehow, machines must make the chips. Once they
are made, they have to operateand keep operating. Hard reality puts some tight constraints on microprocessor design. If engineers arent careful, microprocessors can become
miniature incinerators, burning themselves up. Well take a look at some of these realworld issues that guide microprocessor designincluding electricity, heat, and packaging,
all of which work together (and at times, against the design engineer).
Next, well look at real microprocessors, the chips you can actually buy. Well start with a
brief history lesson to put todays commercial offerings in perspective. And, finally, well
look at the chips in todays (and tomorrows) computers to see which is meant for what
purposeand which is best for your own computer.
Background
Every modern microprocessor starts with the basicsclocked-logic digital circuitry. The
chip has millions of separate gates combined into three basic function blocks: the
input/output unit (or I/O unit), the control unit, and the arithmetic/logic unit (ALU).
The last two are sometimes jointly called the central processing unit (CPU), although the
same term often is used as a synonym for the entire microprocessor. Some chipmakers
further subdivide these units, give them other names, or include more than one of each in
a particular microprocessor. In any case, the functions of these three units are an inherent
part of any chip. The differences are mostly a matter of nomenclature, because you can
understand the entire operation of any microprocessor as a product of these three functions.
All three parts of the microprocessor interact together. In all but the simplest microprocessor designs, the I/O unit is under the control of the control unit, and the operation
of the control unit may be determined by the results of calculations of the arithmetic/logic
unit CPU. The combination of the three parts determines the power and performance of
the microprocessor.
CHAPTER 5 MICROPROCESSORS
Background
Each part of the microprocessor also has its own effect on the processing speed of the system. The control unit operates the microprocessors internal clock, which determines the
rate at which the chip operates. The I/O unit determines the bus width of the microprocessor, which influences how quickly data and instructions can be moved in and out of
the microprocessor. And the registers in the arithmetic/control unit determine how much
data the microprocessor can operate on at one time.
Input/Output Unit
The input/output unit links the microprocessor to the rest of the circuitry of the computer, passing along program instructions and data to the registers of the control unit and
arithmetic/logic unit. The I/O unit matches the signal levels and timing of the microprocessors internal solid-state circuitry to the requirements of the other components
inside the computer. The internal circuits of a microprocessor, for example, are designed
to be stingy with electricity so that they can operate faster and cooler. These delicate
internal circuits cannot handle the higher currents needed to link to external components.
Consequently, each signal leaving the microprocessor goes through a signal buffer in the
I/O unit that boosts its current capacity.
The input/output unit can be as simple as a few buffers, or it may involve many complex
functions. In the latest Intel microprocessors used in some of the most powerful computers, the I/O unit includes cache memory and clock-doubling or -tripling logic to match
the high operating speed of the microprocessor to slower external memory.
The microprocessors used in computers have two kinds of external connections to their
input/output units: those connections that indicate the address of memory locations to or
from which the microprocessor will send or receive data or instructions, and those connections that convey the meaning of the data or instructions. The former is called the
address bus of the microprocessor; the latter, the data bus.
The number of bits in the data bus of a microprocessor directly influences how quickly it
can move information. The more bits that a chip can use at a time, the faster it is. The
first microprocessors had data buses only four bits wide. Pentium chips use a 32-bit data
bus, as do the related Athlon, Celeron, and Duron chips. Itanium and Opteron chips have
64-bit data buses.
The number of bits available on the address bus influences how much memory a microprocessor can address. A microprocessor with 16 address lines, for example, can directly
16
work with 2 addresses; thats 65,536 (or 64K) different memory locations. The different
microprocessors used in various computers span a range of address bus widths from 32 to
64 or more bits.
85
86
The range of bit addresses used by a microprocessor and the physical number of address
lines of the chip no longer correspond. Thats because people and microprocessors look at
memory differently. Although people tend to think of memory in terms of bytes, each
comprising eight bits, microprocessors now deal in larger chunks of data, corresponding
to the number of bits in their data buses. For example, a Pentium chip chews into data 32
bits at a time, so it doesnt need to look to individual bytes. It swallows them four at a
time. Chipmakers consequently omit the address lines needed to distinguish chunks of
memory smaller than their data buses. This bit of frugality saves the number of connections the chip needs to make with the computers circuitry, an issue that becomes important once you see (as you will later) that the modern microprocessor requires several
hundred external connectionseach prone to failure.
Control Unit
The control unit of a microprocessor is a clocked logic circuit that, as its name implies,
controls the operation of the entire chip. Unlike more common integrated circuits, whose
function is fixed by hardware design, the control unit is more flexible. The control unit
follows the instructions contained in an external program and tells the arithmetic/logic
unit what to do. The control unit receives instructions from the I/O unit, translates them
into a form that can be understood by the arithmetic/logic unit, and keeps track of which
step of the program is being executed.
With the increasing complexity of microprocessors, the control unit has become more
sophisticated. In the basic Pentium, for example, the control unit must decide how to
route signals between what amounts to two separate processing units called pipelines. In
other advanced microprocessors, the function of the control unit is split among other
functional blocks, such as those that specialize in evaluating and handling branches in the
stream of instructions.
Arithmetic/Logic Unit
The arithmetic/logic unit handles all the decision-making operations (the mathematical
computations and logic functions) performed by the microprocessor. The unit takes the
instructions decoded by the control unit and either carries them out directly or executes
the appropriate microcode (see the section titled Microcode later in this chapter) to
modify the data contained in its registers. The results are passed back out of the microprocessor through the I/O unit.
The first microprocessors had but one ALU. Modern chips may have several, which commonly are classed into two types. The basic form is the integer unit, one that carries out
only the simplest mathematical operations. More powerful microprocessors also include
one or more floating-point units, which handle advanced math operations (such as trigonometric and transcendental functions), typically at greater precision.
Background
Floating-Point Unit
Although functionally a floating-point unit is part of the arithmetic/logic unit, engineers
often discuss it separately because the floating-point unit is designed to process only
floating-point numbers and not to take care of ordinary math or logic operations.
Floating-point describes a way of expressing values, not a mathematically defined type of
number such as an integer, rational, or real number. The essence of a floating-point number is that its decimal point floats between a predefined number of significant digits
rather than being fixed in place the way dollar values always have two decimal places.
Mathematically speaking, a floating-point number has three parts: a sign, which indicates
whether the number is greater or less than zero; a significant (sometimes called a mantissa),
which comprises all the digits that are mathematically meaningful; and an exponent, which
determines the order of magnitude of the significant (essentially the location to which the
decimal point floats). Think of a floating-point number as being like those represented by
scientific notation. But whereas scientists are apt to deal in base-10 (the exponents in scientific notation are powers of 10), floating-point units think of numbers digitally in base-2
(all ones and zeros in powers of two).
As a practical matter, the form of floating-point numbers used in computer calculations
follows standards laid down by the Institute of Electrical and Electronic Engineers
(IEEE). The IEEE formats take values that can be represented in binary form using 80
bits. Although 80 bits seems somewhat arbitrary in a computer world thats based on powers of two and a steady doubling of register size from 8 to 16 to 32 to 64 bits, its exactly
the right size to accommodate 64 bits of the significant, with 15 bits leftover to hold an
exponent value and an extra bit for the sign of the number held in the register. Although
the IEEE standard allows for 32-bit and 64-bit floating-point values, most floating-point
units are designed to accommodate the full 80-bit values. The floating-point unit (FPU)
carries out all its calculations using the full 80 bits of the chips registers, unlike the integer unit, which can independently manipulate its registers in byte-wide pieces.
The floating-point units of Intel-architecture processors have eight of these 80-bit registers in which to perform their calculations. Instructions in your programs tell the microprocessor whether to use its ordinary integer ALU or its floating-point unit to carry out a
mathematical operation. The different instructions are important because the eight 80-bit
registers in Intel floating-point units also differ from integer units in the way they are
addressed. Commands for integer unit registers are directly routed to the appropriate register as if sent by a switchboard. Floating-point unit registers are arranged in a stack, sort
of an elevator system. Values are pushed onto the stack, and with each new number the
old one goes down one level. Stack machines are generally regarded as lean and mean computers. Their design is austere and streamlined, which helps them run more quickly. The
same holds true for stack-oriented floating-point units.
87
88
Until the advent of the Pentium, a floating-point unit was not a guaranteed part of a
microprocessor. Some 486 and all previous chips omitted floating-point circuitry. The
floating-point circuitry simply added too much to the complexity of the chip, at least for
the state of fabrication technology at that time. To cut costs, chipmakers simply left the
floating-point unit as an option.
When it was necessary to accelerate numeric operations, the earliest microprocessors used
in computers allowed you to add an additional, optional chip to your computer to accelerate the calculation of floating-point values. These external floating-point units were
termed math coprocessors.
The floating-point units of modern microprocessors have evolved beyond mere numbercrunching. They have been optimized to reflect the applications for which computers
most often crunch floating-point numbersgraphics and multimedia (calculating dots,
shapes, colors, depth, and action on your screen display).
Instruction Sets
Instructions are the basic units for telling a microprocessor what to do. Internally, the circuitry of the microprocessor has to carry out hundreds, thousands, or even millions of
logic operations to carry out one instruction. The instruction, in effect, triggers a cascade
of logical operations. How this cascade is controlled marks the great divide in microprocessor and computer design.
The first electronic computers used a hard-wired design. An instruction simply activated
the circuits appropriate for carrying out all the steps required. This design has its advantages. It optimizes the speed of the system because the direct hard-wire connection adds
nothing to slow down the system. Simplicity means speed, and the hard-wired approach is
the simplest. Moreover, the hard-wired design was the practical and obvious choice. After
all, computers were so new that no one had thought up any alternative.
However, the hard-wired computer design has a significant drawback. It ties the hardware
and software together into a single unit. Any change in the hardware must be reflected in
the software. A modification to the computer means that programs have to be modified.
A new computer design may require that programs be entirely rewritten from the
ground up.
Microcode
The inspiration for breaking away from the hard-wired approach was the need for flexibility in instruction sets. Throughout most of the history of computing, determining exactly
what instructions should make up a machines instruction set was more an art than a science. IBMs first commercial computers, the 701 and 702, were designed more from intuition than from any study of which instructions programmers would need to use. Each
Background
machine was tailored to a specific application. The 701 ran instructions thought to serve
scientific users; the 702 had instructions aimed at business and commercial applications.
When IBM tried to unite its many application-specific computers into a single, more general-purpose line, these instruction sets were combined so that one machine could satisfy
all needs. The result was, of course, a wide, varied, and complex set of instructions. The
new machine, the IBM 360 (introduced in 1964), was unlike previous computers in that it
was created not as hardware but as an architecture. IBM developed specifications and rules
for how the machine would operate but enabled the actual machine to be created from
any hardware implementation designers found most expedient. In other words, IBM
defined the instructions that the 360 would use but not the circuitry that would carry
them out. Previous computers used instructions that directly controlled the underlying
hardware. To adapt the instructions defined by the architecture to the actual hardware
that made up the machine, IBM adopted an idea called microcode, originally conceived by
Maurice Wilkes at Cambridge University.
In the microcode design, an instruction causes a computer to execute a small program to
carry out the logic instructions required by the instruction. The collection of small programs for all the instructions the computer understands is its microcode.
Although the additional layer of microcode made machines more complex, it added a
great deal of design flexibility. Engineers could incorporate whatever new technologies
they wanted inside the computer, yet still run the same software with the same instructions originally written for older designs. In other words, microcode enabled new hardware designs and computer systems to have backward compatibility with earlier machines.
After the introduction of the IBM 360, nearly all mainframe computers used microcode.
When the microprocessors came along, they followed the same design philosophy, using
microcode to match instructions to hardware. Using this design, a microprocessor actually
has a smaller microprocessor inside it, which is sometimes called a nanoprocessor, running
the microcode.
This microcode-and-nanoprocessor approach makes creating a complex microprocessor
easier. The powerful data-processing circuitry of the chip can be designed independently
of the instructions it must carry out. The manner in which the chip handles its complex
instructions can be fine-tuned even after the architecture of the main circuits are laid into
place. Bugs in the design can be fixed relatively quickly by altering the microcode, which
is an easy operation compared to the alternative of developing a new design for the whole
chip (a task thats not trivial when millions of transistors are involved). The rich instruction set fostered by microcode also makes writing software for the microprocessor (and
computers built from it) easier, thus reducing the number of instructions needed for each
operation.
89
90
Microcode has a big disadvantage, however. It makes computers and microprocessors

more complicated. In a microprocessor, the nanoprocessor must go through several of its
own microcode instructions to carry out every instruction you send to the microprocessor.
More steps means more processing time taken for each instruction. Extra processing time
means slower operation. Engineers found that microcode had its own way to compensate
for its performance penaltycomplex instructions.
Using microcode, computer designers could easily give an architecture a rich repertoire of
instructions that carry out elaborate functions. A single, complex instruction might do the
job of half a dozen or more simpler instructions. Although each instruction would take
longer to execute because of the microcode, programs would need fewer instructions
overall. Moreover, adding more instructions could boost speed. One result of this microcode more is merrier instruction approach is that typical computer microprocessors
have seven different subtraction commands.
RISC
Although long the mainstay of computer and microprocessor design, microcode is not
necessary. While system architects were staying up nights concocting ever more powerful
and obscure instructions, a counter force was gathering. Starting in the 1970s, the microcode approach came under attack by researchers who claimed it takes a greater toll on
performance than its benefits justify.
By eliminating microcode, this design camp believed, simpler instructions could be executed at speeds so much higher that no degree of instruction complexity could compensate. By necessity, such hard-wired machines would offer only a few instructions because
the complexity of their hard-wired circuitry would increase dramatically with every additional instruction added. Practical designs are best made with small instruction sets.
John Cocke at IBMs Yorktown Research Laboratory analyzed the usage of instructions by
computers and discovered that most of the work done by computers involves relatively
few instructions. Given a computer with a set of 200 instructions, for example, two-thirds
of its processing involves using as few as 10 of the total instructions. Cocke went on to
design a computer that was based on a few instructions that could be executed quickly. He
is credited with inventing the Reduced Instruction Set Computer (RISC) in 1974. The term
RISC itself is credited to David Peterson, who used it in a course in microprocessor design
at the University of California at Berkeley in 1980.
The first chip to bear the label and to take advantage of Cockes discoveries was RISC-I, a
laboratory design that was completed in 1982. To distinguish this new design approach
from traditional microprocessors, microcode-based systems with large instruction sets
have come to be known as Complex Instruction Set Computers (CISC).
Background
Cockes research showed that most of the computing was done by basic instructions, not
by the more powerful, complex, and specialized instructions. Further research at Berkeley
and Stanford Universities demonstrated that there were even instances in which a
sequence of simple instructions could perform a complex task faster than a single complex
instruction could. The result of this research is often summarized as the 80/20 Rule,
meaning that about 20 percent of a computers instructions do about 80 percent of the
work. The aim of the RISC design is to optimize a computers performance for that 20
percent of instructions, speeding up their execution as much as possible. The remaining
80 percent of the commands could be duplicated, when necessary, by combinations of the
quick 20 percent. Analysis and practical experience has shown that the 20 percent could be
made so much faster that the overhead required to emulate the remaining 80 percent was
no handicap at all.
To enable a microprocessor to carry out all the required functions with a handful of
instructions requires a rethinking of the programming process. Instead of simply translating human instructions into machine-readable form, the compilers used by RISC processors attempt to find the optimum instructions to use. The compiler takes a more in-depth
look at the requested operations and finds the best way to handle them. The result was
the creation of optimizing compilers discussed in Chapter 3, Software.
If effect, the RISC design shifts a lot of the processing from the microprocessor to the
compilera lot of the work in running a program gets taken care of before the program
actually runs. Of course, the compiler does more work and takes longer to run, but thats a
fair tradeoffa program needs to be compiled only once but runs many, many times when
the streamlined execution really pays off.
RISC microprocessors have several distinguishing characteristics. Most instructions execute in a single clock cycleor even faster with advanced microprocessor designs with
several execution pathways. All the instructions are the same length with similar syntax.
The processor itself does not use microcode; instead, the small repertory of instructions is
hard-wired into the chip. RISC instructions operate only on data in the registers of the
chip, not in memory, making what is called a load-store design. The design of the chip
itself is relatively simple, with comparatively few logic gates that are themselves constructed from simple, almost cookie-cutter designs. And most of the hard work is shifted
from the microprocessor itself to the compiler.
Micro-Ops
Both CISC and RISC have a compelling design rationale and performance, desirable enough
that engineers working on one kind of chip often looked over the shoulders of those working
in the other camp. As a result, they developed hybrid chips embodying elements of both
the CISC and RISC design. All the latest processorsfrom the Pentium Pro to the
91
92
Pentium 4, Athlon, and Duron as wellhave RISC cores mated with complex instruction
sets.
The basic technique involves converting the classic Intel instructions into RISC-style
instructions to be processed by the internal chip circuitry. Intel calls the internal RISClike instructions micro-ops. The term is often abbreviated as uops (strictly speaking, the initial u should be the Greek letter mu, which is an abbreviation for micro) and pronounced
you-ops. Other companies use slightly different terminology.
By design, the micro-ops sidestep the primary shortcomings of the Intel instruction set by
making the encoding of all commands more uniform, converting all instructions to the
same length for processing, and eliminating arithmetic operations that directly change
memory by loading memory data into registers before processing.
The translation to RISC-like instructions allows the microprocessor to function internally
as a RISC engine. The code conversion occurs in hardware, completely invisible to your
applications and out of the control of programmers. In other words, it shifts back from
the RISC shift to doing the work in the compiler. Theres a good reason for this backward
shift: It lets the RISC code deal with existing programsthose compiled before the RISC
designs were created.
Single Instruction, Multiple Data

In a quest to improve the performance of Intel microprocessors on common multimedia
tasks, Intels hardware and software engineers analyzed the operations multimedia programs most often required. They then sought the most efficient way to enable their chips
to carry out these operations. They essentially worked to enhance the signal-processing
capabilities of their general-purpose microprocessors so that they would be competitive
with dedicated processors, such as digital signal processor (DSP) chips. They called the
technology they developed Single Instruction, Multiple Data (SIMD). In effect a new class
of microprocessor instructions, SIMD is the enabling element of Intels MultiMedia
Extensions (MMX) to its microprocessor command set. Intel further developed this technology to add its Streaming SIMD Extensions (SSE, once known as the Katmai New
Instructions) to its Pentium III microprocessors to enhance their 3D processing power.
The Pentium 4 further enhances SSE with more multimedia instructions to create what
Intel calls SSE2.
As the name implies, SIMD allows one microprocessor instruction to operate across several bytes or words (or even larger blocks of data). In the MMX scheme of things, the
SIMD instructions are matched to the 64-bit data buses of Intels Pentium and newer
microprocessors. All data, whether it originates as bytes, words, or 16-bit double-words,
gets packed into 64-bit form. Eight bytes, four words, or two double-words get packed
Performance-Enhancing Architectures
into a single 64-bit package that, in turn, gets loaded into a 64-bit register in the microprocessor. One microprocessor instruction then manipulates the entire 64-bit block.
Although the approach at first appears counterintuitive, it improves the handling of common graphic and audio data. In video processor applications, for example, it can trim the
number of microprocessor clock cycles for some operations by 50 percent or more.
Very Long Instruction Words

Just as RISC started flowing into the product mainstream, a new idea started designers
thinking in the opposite direction. Very long instruction word (VLIW) technology at first
appears to run against the RISC stream by using long, complex instructions. In reality,
VLIW is a refinement of RISC meant to better take advantage of superscalar microprocessors. Each very long instruction word is made from several RISC instructions. In a
typical implementation, eight 32-bit RISC instructions combine to make one instruction
word.
Ordinarily, combining RISC instructions would add little to overall speed. As with RISC,
the secret of VLIW technology is in the softwarethe compiler that produces the final
program code. The instructions in the long word are chosen so that they execute at the
same time (or as close to it as possible) in parallel processing units in the superscalar
microprocessor. The compiler chooses and arranges instructions to match the needs of the
superscalar processor as best as possible, essentially taking the optimizing compiler one
step further. In essence, the VLIW system takes advantage of preprocessing in the compiler to make the final code and microprocessor more efficient.
VLIW technology also takes advantage of the wider bus connections of the latest generation of microprocessors. Existing chips link to their support circuitry with 64-bit buses.
Many have 128-bit internal buses. The 256-bit very long instruction words push a little
further yet enable a microprocessor to load several cycles of work in a single memory
cycle. Transmetas Crusoe processor uses VLIW technology.
Functionally, the first microprocessors operated a lot like meat grinders. You put something in such as meat scraps, turned a crank, and something new and wonderful came
outa sausage. Microprocessors started with data and instructions and yielded answers,
but operationally they were as simple and direct as turning a crank. Every operation carried out by the microprocessor clicked with a turn of the crankone clock cycle, one
operation.
93
94
Such a design is straightforward and almost elegant. But its wonderful simplicity imposes
a heavy constraint. The computers clock becomes an unforgiving jailor, locking up the
performance of the microprocessor. A chip with this turn-the-crank design is locked to
the clock speed and can never improve its performance beyond one operation per clock
cycle. The situation is worse than that. The use of microcode almost ensures that at least
some instructions will require multiple clock cycles.
One way to speed up the execution of instructions is to reduce the number of internal
steps the microprocessor must take for execution. That idea was the guiding principle
behind the first RISC microprocessors and what made them so interesting to chip designers. Actually, however, step reduction can take one of two forms: making the microprocessor more complex so that steps can be combined or making the instructions simpler so
that fewer steps are required. Both approaches have been used successfully by microprocessor designersthe former as CISC microprocessors, the latter as RISC.
Ideally, it would seem, executing one instruction every clock cycle would be the best anyone could hope for, the ultimate design goal. With conventional microprocessor designs,
that would be true. But engineers have found another way to trim the clock cycles
required by each instructionby processing more than one instruction at the same time.
Two basic approaches to processing more instructions at once are pipelining and superscalar architecture. All modern microprocessors take advantage of these technologies as
well as several other architectural refinements that help them carry out more instructions
for every cycle of the system clock.
Clock Speed
The operating speed of a microprocessor is usually called its clock speed, which describes
the frequency at which the core logic of the chip operates. Clock speed is usually measured in megahertz (one million hertz or clock cycles per second) or gigahertz (a billion
hertz). All else being equal, a higher number in megahertz means a faster microprocessor.
Faster does not necessarily mean the microprocessor will compute an answer more
quickly, however. Different microprocessor designs can execute instructions more efficiently because theres no one-to-one correspondence between instruction processing and
clock speed. In fact, each new generation of microprocessor has been able to execute more
instructions per clock cycle, so a new microprocessor can carry out more instructions at a
given megahertz rating. At the same megahertz rating, a Pentium 4 is faster than a
Pentium III. Why? Because of pipelining, superscalar architecture, and other design features.
Sometimes microprocessor-makers take advantage of this fact and claim that megahertz
doesnt matter. For example, AMDs Athlon processors carry out more instructions per
clock cycle than Intels Pentium III, so AMD stopped using megahertz numbers to
describe its chips. Instead, it substituted model designations that hinted at the speed of a
comparable Pentium chips. An Athlon XP 2200+ processes data as quickly as a Pentium 4
at 2200MHz chip, although the Athlon chip actually operates at less than 2000MHz.
With the introduction of its Itanium series of processors, Intel also made assertions that
megahertz doesnt matter because Itanium chips have clock speeds substantially lower
than Pentium chips.
A further complication is software overhead. Microprocessor speed doesnt affect the performance of Windows or its applications very much. Thats because the performance of
Windows depends on the speed of your hard disk, video system, memory system, and
other system resources as well as your microprocessor. Although a Windows system using
a 2GHz processor will appear faster than a system with a 1GHz processor, it wont be
anywhere near twice as fast.
In other words, the megahertz rating of a microprocessor gives only rough guidance in
comparing microprocessor performance in real-world applications. Faster is better, but a
comparison of megahertz (or gigahertz) numbers does not necessarily express the relationship between the performance of two chips or computer systems.
Pipelining
In older microprocessor designs, a chip works single-mindedly. It reads an instruction
from memory, carries it out, step by step, and then advances to the next instruction. Each
step requires at least one tick of the microprocessors clock. Pipelining enables a microprocessor to read an instruction, start to process it, and then, before finishing with the
first instruction, read another instruction. Because every instruction requires several steps,
each in a different part of the chip, several instructions can be worked on at once and
passed along through the chip like a bucket brigade (or its more efficient alternative, the
pipeline). Intels Pentium chips, for example, have four levels of pipelining. Up to four different instructions may be undergoing different phases of execution at the same time
inside the chip. When operating at its best, pipelining reduces the multiple-step/multipleclock-cycle processing of an instruction to a single clock cycle.
Pipelining is very powerful, but it is also demanding. The pipeline must be carefully organized, and the parallel paths kept carefully in step. Its sort of like a chorus singing a canon
such as Frr Jacquesone missed beat and the harmony falls apart. If one of the execution stages delays, all the rest delay as well. The demands of pipelining push microprocessor designers to make all instructions execute in the same number of clock cycles. That
way, keeping the pipeline in step is easier.
In general, the more stages to a pipeline, the greater acceleration it can offer. Intel has
added superlatives to the pipeline name to convey the enhancement. Super-pipelining is
Intels term for breaking the basic pipeline stages into several steps, resulting in a 12-stage
95
96
design used for its Pentium Pro through Pentium III chips. Later, Intel further sliced the
stages to create the current Pentium 4 chip with 20 stages, a design Intel calls hyperpipelining.
Real-world programs conspire against lengthy pipelines, however. Nearly all programs
branch. That is, their execution can take alternate paths down different instruction
streams, depending on the results of calculations and decision-making. A pipeline can load
up with instructions of one program branch before it discovers that another branch is the
one the program is supposed to follow. In that case, the entire contents of the pipeline
must be dumped and the whole thing loaded up again. The result is a lot of logical wheelspinning and wasted time. The bigger the pipeline, the more time thats wasted. The
waste resulting from branching begins to outweigh the benefits of bigger pipelines in the
vicinity of five stages.
Branch Prediction
Todays most powerful microprocessors are adopting a technology called branch prediction
logic to deal with this problem. The microprocessor makes its best guess at which branch a
program will take as it is filling up the pipeline. It then executes these most likely instructions. Because the chip is guessing at what to do, this technology is sometimes called speculative execution.
When the microprocessors guesses turn out to be correct, the chip benefits from the multiple-pipeline stages and is able to run through more instructions than clock cycles. When
the chips guess turns out wrong, however, it must discard the results obtained under speculation and execute the correct code. The chip marks the data in later pipeline stages as
invalid and discards it. Although the chip doesnt lose timethe program would have executed in the same order anywayit does lose the extra boost bequeathed by the pipeline.
Speculative Execution
To further increase performance, more modern microprocessors use speculative execution.
That is, the chip may carry out an instruction in a predicted branch before it confirms
whether it has properly predicted the branch. If the chips prediction is correct, the
instruction has already been executed, so the chip wastes no time. If the prediction was
incorrect, the chip will have to execute a different instruction, which it would have to have
done anyhow, so it suffers no penalty.
Superscalar Architectures
The steps in a program normally are listed sequentially, but they dont always need to be
carried out exactly in order. Just as tough problems can be broken into easier pieces, program code can be divided as well. If, for example, you want to know the larger of two
rooms, you have to compute the volume of each and then make your comparison. If you
had two brains, you could compute the two volumes simultaneously. A superscalar microprocessor design does essentially that. By providing two or more execution paths for programs, it can process two or more program parts simultaneously. Of course, the chip
needs enough innate intelligence to determine which problems can be split up and how to
do it. The Pentium, for example, has two parallel, pipelined execution paths.
The first superscalar computer design was the Control Data Corporation 6600 mainframe, introduced in 1964. Designed specifically for intense scientific applications, the initial 6600 machines were built from eight functional units and were the fastest computers
in the world at the time of their introduction.
Superscalar architecture gets its name because it goes beyond the incremental increase in
speed made possible by scaling down microprocessor technology. An improvement to the
scale of a microprocessor design would reduce the size of the microcircuitry on the silicon
chip. The size reduction shortens the distance signals must travel and lowers the amount
of heat generated by the circuit (because the elements are smaller and need less current to
effect changes). Some microprocessor designs lend themselves to scaling down.
Superscalar designs get a more substantial performance increase by incorporating a more
dramatic change in circuit complexity.
Using pipelining and superscalar architecture cycle-saving techniques has cut the number
of clock cycles required for the execution of a typical microprocessor instruction dramatically. Early microprocessors needed, on average, several cycles for each instruction.
Todays chips can often carry out multiple instructions in a single clock cycle. Engineers
describe pipelined, superscalar chips by the number of instructions they can retire per
clock cycle. They look at the number of instructions that are completed, because this best
describes how much work the chip has (or can) actually accomplish.
Out-of-Order Execution
No matter how well the logic of a superscalar microprocessor divides up a program, each
pipeline is unlikely to get an equal share of the work. One or another pipeline will grind
away while another finishes in an instant. Certainly the chip logic can shove another
instruction down the free pipeline (if another instruction is ready). But if the next instruction depends on the results of the one before it, and that instruction is the one stuck
grinding away in the other pipeline, the free pipeline stalls. It is available for work but can
do no work, thus potential processor power gets wasted.
Like a good Type-A employee who always looks for something to do, microprocessors can
do the same. They can check the program for the next instruction that doesnt depend on
previous work thats not finished and work on the new instruction. This sort of ambitious
97
98
approach to programs is termed out-of-order execution, and it helps microprocessors take

full advantage of superscalar designs.
This sort of ambitious microprocessor faces a problem, however. It is no longer running
the program in the order it was written, and the results might be other than the programmer had intended. Consequently, microprocessors capable of out-of-order execution dont
immediately post the results from their processing into their registers. The work gets carried out invisibly and the results of the instructions that are processed out of order are
held in a buffer until the chip has finished the processing of all the previous instructions.
The chip puts the results back into the proper order, checking to be sure that the out-oforder execution has not caused any anomalies, before posting the results to its registers.
To the program and the rest of the outside world, the results appear in the microprocessors registers as if they had been processed in normal order, only faster.
Register Renaming
Out-of-order execution often runs into its own problems. Two independently executable
instructions may refer to or change the same register. In the original program, one would
carry out its operation, then the other would do its work later. During superscalar out-oforder execution, the two instructions may want to work on the register simultaneously.
Because that conflict would inevitably lead to confusing results and errors, an ordinary
superscalar microprocessor would have to ensure the two instructions referencing the
same register executed sequentially instead of in parallel, thus eliminating the advantage of
its superscalar design.
To avoid such problems, advanced microprocessors use register renaming. Instead of a
small number of registers with fixed names, they use a larger bank of registers that can be
named dynamically. The circuitry in each chip converts the references made by an instruction to a specific register name to point instead to its choice of physical register. In effect,
the program asks for the EAX register, and the chip says, Sure, and gives the program a
register it calls EAX. If another part of the program asks for EAX, the chip pulls out a different register and tells the program that this one is EAX, too. The program takes the
microprocessors word for it, and the microprocessor doesnt worry because it has several
million transistors to sort things out in the end.
And it takes several million transistors because the chip must track all references to registers. It does this to ensure that when one program instruction depends on the result in a
given register, it has the right register and results dished up to it.
Explicitly Parallel Instruction Computing

With Intels shift to 64-bit architecture for its most powerful line of microprocessors
(aimed, for now, at the server market), the company introduced a new instruction set to
compliment the new architecture. Called Explicitly Parallel Instruction Computing (EPIC),
the design follows the precepts of RISC architecture by putting the hard work into software (the compiler), while retaining the advantages of longer instructions used by SIMD
and VLIW technologies.
The difference between EPIC and older Intel chips is that the compiler takes a swipe at
each program and determines where parallel processes can occur. It then optimizes the
program code to sort out separate streams of execution that can be routed to different
microprocessor pipelines and carried out concurrently. This not only relieves the chip
from working to figure out how to divide up the instruction stream, but it also allows the
software to more thoroughly analyze the code rather than trying to do it on the fly.
By analyzing and dividing the instruction streams before they are submitted to the microprocessor, EPIC trims the need and use of speculative execution and branch prediction.
The compiler can look ahead in the program, so it doesnt have to speculate or predict. It
knows how best to carry out a complex program.
Front-Side Bus
The instruction stream is not the only bottleneck in a modern computer. The core logic
of most microprocessors operates much faster than other parts of most computers, including the memory and support circuitry. The microprocessor links to the rest of the computer through a connection called the system bus or the front-side bus. The speed at which
the system bus operates sets a maximum limit on how fast the microprocessor can send
data to other circuits (including memory) in the computer.
When the microprocessor needs to retrieve data or an instruction from memory, it must
wait for this to come across the system bus. The slower the bus, the longer the microprocessor has to wait. More importantly, the greater the mismatch between microprocessor speed and system bus speed, the more clock cycles the microprocessor needs to wait.
Applications that involve repeated calculations on large blocks of datagraphics and video
in particularare apt to require the most access to the system bus to retrieve data from
memory. These applications are most likely to suffer from a slow system bus.
The first commercial microprocessors of the current generation operated their system
buses at 66MHz. Through the years, manufacturers have boosted this speed and increased
the speed at which the microprocessor can communicate with the rest of the computer.
Chips now use clock speeds of 100MHz or 133MHz for their system buses.
With the Pentium 4, Intel added a further refinement to the system bus. Using a technology called quad-pumping, Intel forces four data bits onto each clock cycle of the system
bus. A quad-pumped system bus operating at 100MHz can actually move 400Mb of
data in a second. A quad-pumped system bus running at 133MHz achieves a 533Mbps
data rate.
99
100
The performance of the system bus is often described in its bandwidth, the number of total
bytes of data that can move through the bus in one second. The data buses of all current
computers are 64 bits widethats 8 bytes. Multiplying the clock speed or data rate of the
bus by its width yields its bandwidth. A 100MHz system bus therefore has an 800MBps
(megabytes per second) bandwidth. A 400MHz bus has a 3.2GBps bandwidth.
Intel usually locks the system bus speed to the clock speed of the microprocessor. This
synchronous operation optimizes the transfer rate between the bus and the microprocessor. It also explains some of the odd frequencies at which some microprocessors are
designed to operate (for example, 1.33GHz or 2.56GHz). Sometimes the system bus operates at a speed thats not an even divisor of the microprocessor speedfor example, the
microprocessor clock speed may be 4.5, 5, or 5.5 times the system bus speed. Such a mismatch can slow down system performance, although such mismatches are minimized by
effective microprocessor caching (see the upcoming section Caching for more information).
Translation Look-Aside Buffers

Modern pipelined, superscalar microprocessors need to access memory quickly, and they
often repeatedly go to the same address in the execution of a program. To speed up such
operations, most newer microprocessors include a quick lookup list of the pages in memory that the chip has addressed most recently. This list is termed a translation look-aside
and is a small block of fast memory inside the microprocessor that stores a table that
cross-references the virtual addresses in programs with the corresponding real addresses in
physical memory that the program has most recently used. The microprocessor can take a
quick glance away from its normal address-translation pipeline, effectively looking aside,
to fetch the addresses it needs.
The translation look-aside buffer (TLB) appears to be very small in relation to the memory of most computers. Typically, a TLB may be 64 to 256 entries. Each entry, however,
refers to an entire page of memory, which with todays Intel microprocessors, totals four
kilobytes. The amount of memory that the microprocessor can quickly address by checking the TLB is the TLB address space, which is the product of the number of entries in the
TLB and the page size. A 256-entry TLB can provide fast access to a megabyte of memory (256 entries times 4KB per page).
Caching
The most important means of matching todays fast microprocessors to the speeds of
affordable memory, which is inevitably slower, is memory caching. A memory cache interposes a block of fast memorytypically high-speed static RAMbetween the microprocessor and the bulk of primary storage. A special circuit called a cache controller (which
current designs make into an essential part of the microprocessor) attempts to keep the
cache filled with the data or instructions that the microprocessor is most likely to need
next. If the information the microprocessor requests next is held within the cache, it can
be retrieved without waiting.
This fastest possible operation is called a cache hit. If the needed data is not in the cache
memory, it is retrieved from outside the cache, typically from ordinary RAM at ordinary
RAM speed. The result is called a cache miss.
Not all memory caches are created equal. Memory caches differ in many ways: size, logical arrangement, location, and operation.
Cache Level
Caches are sometimes described by their logical and electrical proximity to the microprocessors core logic. The closest physically and electrically to the microprocessors core
logic is the primary cache, also called a Level One cache. A secondary cache (or Level Two cache)
fits between the primary cache and main memory. The secondary cache usually is larger
than the primary cache but operates at a lower speed (to make its larger mass of memory
more affordable). Rarely is a tertiary cache (or Level Three cache) interposed between the
secondary cache and memory.
In modern microprocessor designs, both the primary and secondary caches are part of the
microprocessor itself. Older designs put the secondary cache in a separate part of a microprocessor module or in external memory.
Primary and secondary caches differ in the way they connect with the core logic of the
microprocessor. A primary cache invariably operates at the full speed of the microprocessors core logic with the widest possible bit-width connection between the core logic and
the cache. Secondary caches often operate at a rate slower than the chips core logic,
although all current chips operate the secondary cache at full core speed.
Cache Size
A major factor that determines how successful the cache will be is how much information
it contains. The larger the cache, the more data that is in it and the more likely any
needed byte will be there when you system calls for it. Obviously, the best cache is one
thats as large as, and duplicates, the entirety of system memory. Of course, a cache that
big is also absurd. You could use the cache as primary memory and forget the rest. The
smallest cache would be a byte, also an absurd situation because it guarantees the next
read is not in the cache. Chipmakers try to make caches as large as possible within the
constraints of fabricating microprocessors affordably.
Todays primary caches are typically 64 or 128KB. Secondary caches range from 128 to
512KB for chips for desktop and mobile applications and up to 2MB for server-oriented
microprocessors.
101
102
Instruction and Data Caches

Modern microprocessors subdivide their primary caches into separate instruction and data
caches, typically with each assigned one-half the total cache memory. This separation
allows for a more efficient microprocessor design. Microprocessors handle instructions
and data differently and may even send them down different pipelines. Moreover, instructions and data typically use memory differentlyinstructions are sequential whereas data
can be completely random. Separating the two allows designers to optimize the cache
design for each.
Write-Through and Write-Back Caches

Caches also differ in the way they treat writing to memory. Most caches make no attempt
to speed up write operations. Instead, they push write commands through the cache
immediately, writing to cache and main memory (with normal wait-state delays) at the
same time. This write-through cache design is the safe approach because it guarantees that
main memory and cache are constantly in agreement. Most Intel microprocessors through
the current versions of the Pentium use write-through technology.
The faster alternative is the write-back cache, which allows the microprocessor to write
changes to its cache memory and then immediately go back about its work. The cache
controller eventually writes the changed data back to main memory as time allows.
Cache Mapping
The logical configuration of a cache involves how the memory in the cache is arranged
and how it is addressed (that is, how the microprocessor determines whether needed
information is available inside the cache). The major choices are direct-mapped, full associative, and set-associative.
The direct-mapped cache divides the fast memory of the cache into small units, called
lines (corresponding to the lines of storage used by Intel 32-bit microprocessors, which
allow addressing in 16-byte multiples, blocks of 128 bits), each of which is identified by an
index bit. Main memory is divided into blocks the size of the cache, and the lines in the
cache correspond to the locations within such a memory block. Each line can be drawn
from a different memory block, but only from the location corresponding to the location
in the cache. Which block the line is drawn from is identified by a tag. For the cache controllerthe electronics that ride herd on the cachedetermining whether a given byte is
stored in a direct-mapped cache is easy. It just checks the tag for a given index value.
The problem with the direct-mapped cache is that if a program regularly moves between
addresses with the same indexes in different blocks of memory, the cache needs to be continually refreshedwhich means cache misses. Although such operation is uncommon in
single-tasking systems, it can occur often during multitasking and slow down the directmapped cache.
Electrical Characteristics
The opposite design approach is the full-associative cache. In this design, each line of the
cache can correspond to (or be associated with) any part of main memory. Lines of bytes
from diverse locations throughout main memory can be piled cheek-by-jowl in the cache.
The major shortcoming of the full-associative approach is that the cache controller must
check the addresses of every line in the cache to determine whether a memory request
from the microprocessor is a hit or miss. The more lines there are to check, the more
time it takes. A lot of checking can make cache memory respond more slowly than main
memory.
A compromise between direct-mapped and full-associative caches is the set-associative
cache, which essentially divides up the total cache memory into several smaller directmapped areas. The cache is described as the number of ways into which it is divided. A
four-way set-associative cache, therefore, resembles four smaller direct-mapped caches.
This arrangement overcomes the problem of moving between blocks with the same
indexes. Consequently, the set-associative cache has more performance potential than a
direct-mapped cache. Unfortunately, it is also more complex, making the technology more
expensive to implement. Moreover, the more ways there are to a cache, the longer the
cache controller must search to determine whether needed information is in the cache.
This ultimately slows down the cache, mitigating the advantage of splitting it into sets.
Most computer-makers find a four-way set-associative cache to be the optimum
compromise between performance and complexity.
At its heart, a microprocessor is an electronic device. This electronic foundation has
important ramifications in the construction and operation of chips. The free lunch principle (that is, there is none) tells us that every operation has its cost. Even the quick electronic thinking of a microprocessor takes a toll. The thinking involves the switching of
the state of tiny transistors, and each state change consumes a bit of electrical power,
which gets converted to heat. The transistors are so small that the process generates a
minuscule amount of heat, but with millions of them in a single chip, the heat adds up.
Modern microprocessors generate so much heat that keeping them cool is a major concern in their design.
Thermal Constraints
Heat is the enemy of the semiconductor because it can destroy the delicate crystal structure of a chip. If a chip gets too hot, it will be irrevocably destroyed. Packing circuits
tightly concentrates the heat they generate, and the small size of the individual circuit
components makes them more vulnerable to damage.
103
104
Heat can cause problems more subtle than simple destruction. Because the conductivity of
semiconductor circuits also varies with temperature, the effective switching speed of transistors and logic gates also changes when chips get too hot or too cold. Although this temperature-induced speed change does not alter how fast a microprocessor can compute (the
chip must stay locked to the system clock at all times), it can affect the relative timing
between signals inside the microprocessor. Should the timing get too far off, a microprocessor might make a mistake, with the inevitable result of crashing your system. All
chips have rated temperature ranges within which they are guaranteed to operate without
such timing errors.
Because chips generate more heat as speed increases, they can produce heat faster than it
can radiate away. This heat buildup can alter the timing of the internal signals of the chip
so drastically that the microprocessor will stop working andas if you couldnt guess
cause your system to crash. To avoid such problems, computer manufacturers often attach
heatsinks to microprocessors and other semiconductor components to aid in their cooling.
A heatsink is simply a metal extrusion that increases the surface area from which heat can
radiate from a microprocessor or other heat-generating circuit element. Most heatsinks
have several fins (rows of pins) or some geometry that increases its surface area. Heatsinks
are usually made from aluminum because that metal is one of the better thermal conductors, enabling the heat from the microprocessor to quickly spread across the heatsink.
Heatsinks provide passive cooling (passive because cooling requires no power-using mechanism). Heatsinks work by convection, transferring heat to the air that circulates past the
heatsink. Air circulates around the heatsink because the warmed air rises away from the
heatsink and cooler air flows in to replace it.
In contrast, active cooling involves some kind of mechanical or electrical assistance to
remove heat. The most common form of active cooling is a fan, which blows a greater
volume of air past the heatsink than would be possible with convection alone. Nearly all
modern microprocessors require a fan for active cooling, typically built into the chips
heatsink.
The makers of notebook computers face another challenge in efficiently managing the
cooling of their computers. Using a fan to cool a notebook system is problematic. The fan
consumes substantial energy, which trims battery life. Moreover, the heat generated by the
fan motor itself can be a significant part of the thermal load of the system. Most designers
of notebook machines have turned to more innovative passive thermal controls, such as
heat pipes and using the entire chassis of the computer as a heatsink.
Operating Voltages
In desktop computers, overheating rather than excess electrical consumption is the major
power concern. Even the most wasteful of microprocessors uses far less power than an
ordinary light bulb. The most that any computer-compatible microprocessor consumes is
about nine watts, hardly more than a night light and of little concern when the power grid
supplying your computer has megawatts at its disposal.
If you switch to battery power, however, every last milliwatt is important. The more
power used by a computer, the shorter the time its battery can power the system or the
heavier the battery it will need to achieve a given life between charges. Every degree a
microprocessor raises its case temperature clips minutes from its battery runtime.
Battery-powered notebooks and sub-notebook computers consequently caused microprocessor engineers to do a quick about-face. Whereas once they were content to use bigger and bigger heatsinks, fans, and refrigerators to keep their chips cool, today they focus
on reducing temperatures and wasted power at the source.
One way to cut power requirements is to make the design elements of a chip smaller.
Smaller digital circuits require less power. But shrinking chips is not an option; microprocessors are invariably designed to be as small as possible with the prevailing technology.
To further trim the power required by microprocessors to make them more amenable to
battery operation, engineers have come up with two new design twists: low-voltage operation and system-management mode. Although founded on separate ideas, both are often
used together to minimize microprocessor power consumption. All new microprocessor
designs incorporate both technologies.
Since the very beginning of the transistor-transistor logic family of digital circuits (the
design technology that later blossomed into the microprocessor), digital logic has operated with a supply voltage of 5 volts. That level is essentially arbitrary. Almost any voltage
would work. But 5-volt technology offers some practical advantages. Its low enough to be
both safe and frugal with power needs but high enough to avoid noise and allow for several diode drops, the inevitable reduction of voltage that occurs when a current flows
across a semiconductor junction.
Every semiconductor junction, which essentially forms a diode, reduces or drops the voltage flowing through it. Silicon junctions impose a diode drop of about 0.7 volts, and there
may be one or more such junctions in a logic gate. Other materials impose smaller
dropsthat of germanium, for example, is 0.4 voltsbut the drop is unavoidable.
Theres nothing magical about using 5 volts. Reducing the voltage used by logic circuits
dramatically reduces power consumption because power consumption in electrical circuits
increases by the square of the voltage. That is, doubling the voltage of a circuit increases
the power it uses by fourfold. Reducing the voltage by one-half reduces power consumption by three-quarters (providing, of course, that the circuit will continue to operate at the
lower voltage).
105
106
All current microprocessor designs operate at about 2 volts or less. The Pentium 4 operates at just over 1.3 volts with minor variations, depending on the clock frequency of the
chip. For example, the 2.53GHz version requires 1.325 volts. Microprocessors designed
for mobile applications typically operate at about 1.1 volts; some as low as 0.95 volt.
To minimize power consumption, Intel sets the operating voltage of the core logic of its
chips as low as possiblesome, such as Intels ultra-low-voltage Mobile Pentium III-M, to
just under 1 volt. The integral secondary caches of these chips (which are fabricated separately from the core logic) usually require their own, often higher, voltage supply. In fact,
operating voltage has become so critical that Intel devotes several pins of its Pentium II
and later microprocessors to encoding the voltage needs of the chip, and the host computer must adjust its supply to the chip to precisely meet those needs.
Most bus architectures and most of todays memory modules operate at the 3.3 volt level.
Future designs will push that level lower. Rambus memory systems, for example, operate
at 2.5 volts (see Chapter 6, Chipsets, for more information).
Power Management
Trimming the power need of a microprocessor both reduces the heat the chip generates
and increases how long it can run off a battery supply, an important consideration for
portable computers. Reducing the voltage and power use of the chip is one way of keeping
the heat down and the battery running, but chipmakers have discovered they can save
even more power through frugality. The chips use power only when they have to, thus
managing their power consumption.
Chipmakers have two basic power-management strategies: shutting off circuits when they
are not needed and slowing down the microprocessor when high performance is not
required.
The earliest form of power-savings built into microprocessors was part of system management mode (SMM), which allowed the circuitry of the chip to be shut off. In terms of
clock speed, the chip went from full speed to zero. Initially, chips switched off after a
period of system inactivity and woke up to full speed when triggered by an appropriate
interrupt.
More advanced systems cycled the microprocessor between on and off states as they
required processing power. The chief difficulty with this design is that nothing gets done
when the chip isnt processing. This kind of power management only works when youre
not looking (not exactly a salesmans dream) and is a benefit you should never be able to
see. Intel gives this technique the name QuickStart and claims that it can save enough
energy between your keystrokes to significantly reduce overall power consumption by
briefly cutting the microprocessors electrical needs by 95 percent. Intel introduced
QuickStart in the Mobile Pentium II processor, although it has not widely publicized the
technology.
In the last few years, chipmakers have approached the power problem with more advanced
power-saving systems that take an intermediary approach. One way is to reduce microprocessor power when it doesnt need it for particular operations. Intel slightly reduces
the voltage applied to its core logic based on the activity of the processor. Called Intel
Mobile Voltage Positioning (IMVP), this technology can reduce the thermal design power
which means the heat produced by the microprocessorby about 8.5 percent. According
to Intel, this reduction is equivalent to reducing the speed of a 750MHz Mobile Pentium
III by 100MHz.
Another technique for saving power is to reduce the performance of a microprocessor
when its top speed is not required by your applications. Instead of entirely switching off
the microprocessor, the chipmakers reduce its performance to trim power consumption.
Each of the three current major microprocessor manufacturers puts its own spin on this
performance-as-needed technology, labeling it with a clever trademark. Intel offers
SpeedStep, AMD offers PowerNow!, and Transmeta offers LongRun. Although at heart all
three are conceptually much the same, in operation youll find distinct differences
between them.
Intel SpeedStep
Internal mobile microprocessor power savings started with SpeedStep, introduced by Intel
on January 18, 2000, with the Mobile Pentium III microprocessors, operating at 600MHz
and 650MHz. To save power, these chips can be configured to reduce their operating
speed when running on battery power to 500MHz. All Mobile Pentium III and Mobile
Pentium 4 chips since that date have incorporated SpeedStep into their designs. Mobile
Celeron processors do not use SpeedStep.
The triggering event is a reduction of power to the chip. For example, the initial Mobile
Pentium III chips go from the 1.7 volts that is required for operating at their top speeds
to 1.35 volts. As noted earlier, the M-series step down from 1.4 to 1.15 volts, the lowvoltage M-series from 1.35 to 1.1 volts, and the ultra-low-voltage chips from 1.1 to 0.975
volts. Note that a 15-percent reduction in voltage in itself reduces power consumption by
about 29 percent, with a further reduction thats proportional to the speed decrease. The
600MHz Pentium III, for example, cuts its power consumption an additional 17 percent
thanks to voltage reduction when slipping down from 600MHz to 500MHz.
Intel calls the two modes Maximum Performance Mode (for high speed) and Battery
Optimized Mode (for low speed). According to Intel, switching between speeds requires
about one two-thousandths of a second. The M-series of Mobile Pentium III adds an
additional step to provide an intermediary level of performance when operating on
battery. Intel calls this technology Enhanced SpeedStep. Table 5.1 lists the SpeedStep capabilities of many Intel chips.
107
108
TA B L E 5 . 1
The Effect of SpeedStep in Intel Mobile Pentium-III Microprocessors
Pentium III
Original speed (MHz)
500
600
650
700
750
850
900
1000
After SpeedStep reduction (MHz)
300
500
500
500
550
750
700
700
Ultra-Low-Voltage Pentium III-M

700
300
Low-Voltage Pentium III-M

733
750
800
466
450
533
800A
866
933
1000
1060
1130
1200
500
667
733
733
733
733
800
Pentium III-M
AMD PowerNow!
Advanced Micro Devices devised its own power-saving technology, called PowerNow!, for
its mobile processors. The AMD technology differs from Intels SpeedStep by providing
up to 32 levels of speed reduction and power savings. Note that 32 levels is the design
limit. Actual implementations of the technology from AMD have far fewer levels. All current AMD mobile processorsboth the Mobile Athlon and Mobile Duron linesuse
PowerNow! technology.
PowerNow! operates in one of three modes:
Hi-Performance. This mode runs the microprocessor at full speed and full voltage
so that the chip maximizes its processing power.
Battery Saver. This mode runs the chip at a lower speed using a lower voltage to
conserve battery power, exactly as a SpeedStep chip would, but with multiple levels.
The speed and voltage are determined by the chips requirements and programming
of the BIOS (which triggers the change).
Automatic. This mode makes the changes in voltage and clock speed dynamic,
responding to the needs of the system. When an application requires maximum processing power, the chip runs at full speed. As the need for processing declines, the
chip adjusts its performance to match. Current implementations allow for four
discrete speeds at various operating voltages. Automatic mode is the best compromise for normal operation of portable computers.
The actual means of varying the clock frequency involves dynamic control of the clock
multiplier inside the AMD chip. The external oscillator or clock frequency the computer
supplies the microprocessor does not change, regardless of the performance demand. In
the case of the initial chip to use PowerNow! (the Mobile K6-2+ chip, operating at
550MHz), its actual operating speed could vary from 200MHz to 550MHz in a system
that takes full advantage of the technology.
Control of PowerNow! starts with the operating system (which is almost always
Windows). Windows monitors processor usage, and when it dips below a predetermined
level, such as 50 percent, Windows signals the PowerNow! system to cut back the clock
multiplier inside the microprocessor and then signals a programmable voltage regulator to
trim the voltage going to the chip. Note that even with PowerNow!, the chips supply
voltage must be adjusted externally to the chip to achieve the greatest power savings.
If the operating system detects that the available processing power is still underused, it
signals to cut back another step. Similarly, should processing needs reach above a predetermined level (say, 90 percent of the available ability), the operating system signals
PowerNow! to kick up performance (and voltage) by a notch.
Transmeta LongRun
Transmeta calls its proprietary power-saving technology LongRun. It is a feature of both
current Crusoe processors, the TM5500 and TM5800. In concept, LongRun is much like
AMDs PowerNow! The chief difference is control. Because of the design of Transmetas
Crusoe processors, Windows instructions are irrelevant to their power usageCrusoe
chips translate the Windows instructions into their own format. To gauge its power needs,
the Crusoe chip monitors the flow of its own native instructions and adjusts its speed to
match the processing needs of that code stream. In other words, the Crusoe chip does its
own monitoring and decision-making regarding power savings without regard to
Windows power-conservation information.
According to Transmeta, LongRun allows its microprocessors to adjust their power consumption by changing their clock frequency on the fly, just as PowerNow! does, as well as
to adjust their operating voltage. The processor core steps down processor speed in
33MHz increments, and each step holds the potential of reducing chip voltage. For example, trimming the speed of a chip from 667MHz to 633MHz also allows for reducing the
operating voltage from 1.65 to 1.60 volts.
Packaging
The working part of a microprocessor is exactly what the nickname chip implies: a small
flake of a silicon crystal no larger than a postage stamp. Although silicon is a fairly robust
material with moderate physical strength, it is sensitive to chemical contamination. After
109
110
all, semiconductors are grown in precisely controlled atmospheres, the chemical content
of which affects the operating properties of the final chip. To prevent oxygen and contaminants in the atmosphere from adversely affecting the precision-engineered silicon, the
chip itself must be sealed away. The first semiconductors, transistors, were hermetically
sealed in tiny metal cans.
The art and science of semiconductor packaging has advanced since those early days.
Modern integrated circuits (ICs) are often surrounded in epoxy plastic, an inexpensive
material that can be easily molded to the proper shape. Unfortunately, microprocessors
can get very hot, sometimes too hot for plastics to safely contain. Most powerful modern
microprocessors are consequently cased in ceramic materials that are fused together at
high temperatures. Older, cooler chips reside in plastic. The most recent trend in chip
packaging is the development of inexpensive tape-based packages optimized for automated
assembly of circuit boards.
The most primitive of microprocessors (that is, those of the early generation that had neither substantial signal nor power requirements) fit in the same style housing popular for
other integrated circuitsthe infamous dual inline pin (DIP) package. The packages grew
more pinsor legs, as engineers sometimes call themto accommodate the ever-increasing number of signals in data and address buses.
The DIP package is far from ideal for a number of reasons. Adding more connections, for
example, makes for an ungainly chip. A centipede microprocessor would be a beast measuring a full five inches long. Not only would such a critter be hard to fit onto a reasonably sized circuit board, it would require that signals travel substantially farther to reach
the end pins than those in the center. At modern operating frequencies, that difference in
distance can amount to a substantial fraction of a clock cycle, potentially putting the pins
out of sync.
Modern chip packages are compact squares that avoid these problems. Engineers developed several separate styles to accommodate the needs of the latest microprocessors.
The most common is the pin grid array (PGA), a square package that varies in size with
the number of pins that it must accommodate (typically about two inches square). The
first PGA chips had 68 pins. Pentium 4 chips in PGA packages have up to 478.
No matter their number, the pins are spaced as if they were laid out on a checkerboard,
making the grid array of the package name (see Figure 5.1).
To fit the larger number of pins used by wider-bus microprocessors into a reasonable
space, Intel rearranged the pins of some processors (notably the Pentium Pro), staggering
them so that they can fit closer together. The result is a staggered pin grid array (SPGA)
package, as shown in Figure 5.2.
FIGURE 5.1
Pin-grid array socket
(with PGA chip).
FIGURE 5.2
A Multi-Cavity
Module (MCM) SPGA
package.
Pins take up space and add to the cost of fabrication, so chipmakers have developed a
number of pinless packages. The first of these to find general use was the Leadless Chip
Carrier (LCC) socket. Instead of pins, this style of package has contact pads on one of its
surfaces. The pads are plated with gold to avoid corrosion or oxidation that would impede
the flow of the minute electrical signals used by the chip (see Figure 5.3). The pads are
designed to contact special springy mating contacts in a special socket. Once installed, the
chip itself may be hidden in the socket, under a heat sink, or perhaps only the top of the
chip may be visible, framed by the four sides of the socket.
FIGURE 5.3
Leadless Chip Carrier
microprocessor, top
and bottom views.
A related design, the Plastic Leaded Chip Carrier (PLCC), substitutes epoxy plastic for the
ceramic materials ordinarily used for encasing chips. Plastic is less expensive and easier to
work with. Some microprocessors with low thermal output sometimes use a housing
designed to be soldered downthe Plastic Quad Flat Package (PQFP), sometimes called
111
112
simply the quad flat pack because the chips are flat (they fit flat against the circuit board)
and they have four sides (making them a quadrilateral, as shown in Figure 5.4).
The Tape Carrier Package takes the advantage of the quad flat pack a step further, reducing
the chip to what looks like a pregnant bulge in the middle of a piece of photographic film
(see Figure 5.5).
FIGURE 5.4
Plastic Quad Flat
Package microprocessor.
FIGURE 5.5
Tape Carrier Package
microprocessor.
Another way to deal with the problem of pins is to reduce them to vestigial bumps, substituting precision-formed globs of solder that can mate with socket contacts. Alternately,
the globs can be soldered directly to a circuit board using surface-mount technology.
Because the solder contacts start out as tiny balls but use a variation on the PGA layout,
the package is termed solder-ball grid array. (Note that solder is often omitted from the
name, thus yielding the abbreviation BGA.)
When Intels engineers first decided to add secondary caches to the companys microprocessors, they used a separately housed slice of silicon for the cache. Initially Intel put
the CPU and cache chips in separate chambers in one big, black chip. The design, called
the Multi-Cavity Module (MCM), was used only for the Pentium Pro chip.
Next, Intel shifted to putting the CPU and cache on a small circuit board inside a cartridge, initially called the Single Edge Contact cartridge or SEC cartridge (which Intel often
abbreviates SECC) when it was used for the Pentium II chip. Figure 5.6 shows the
Pentium II microprocessor SEC cartridge.
Intel used a functionally similar but physically different design for its Pentium II Xeon
chips and a similar cartridge but slightly different bus design for the Pentium III.
To cut the cost of the cartridge for the inexpensive Celeron line, Intel eliminated the case
around the chip to make the Singe-Edge Processor (SEP) package (see Figure 5.7).
FIGURE 5.6
The SEC cartridge as
used in the Intel
Pentium II.
FIGURE 5.7
The SEP package as
used by the Intel
Celeron microprocessor.
When Intel developed the capability to put the CPU and secondary cache on a single
piece of silicon, called a die, the need for cartridges disappeared. Both later Celeron and
Pentium III had on-die caches and were packaged both as cartridges and as individual chips
in PGA and similar packages. With the Pentium 4, the circle was complete. Intel offers
the latest Pentiums only in compact chip-style packages.
The package that the chip is housed in has no effect on its performance. It can, however,
be important when you want to replace or upgrade your microprocessor with a new chip
or upgrade card. Many of these enhancement products require that you replace your systems microprocessor with a new chip or adapter cable that links to a circuit board. If you
want the upgrade or a replacement part to fit on your motherboard, you may have to
specify which package your computer uses for its microprocessor.
Ordinarily you dont have to deal with microprocessor sockets unless youre curious and
want to pull out the chip, hold it in your hand, and watch a static discharge turn a $300
circuit into epoxy-encapsulated sand. Choose to upgrade your computer to a new and better microprocessor, and youll tangle with the details of socketry, particularly if you want
to improve your Pentium.
Intel recognizes nine different microprocessor sockets for its processors, from the 486 to
the Pentium Pro. In 1999, it added a new socket for some incarnations of the Pentium II
Celeron. Other Pentium II and Pentium III chips, packaged as modules or cartridges,
mate with slots instead of sockets. Table 5.2 summarizes these socket types, the chips that
use them, and the upgrades appropriate to them.
113
114
TA B L E 5 . 2
Slots and Sockets for Intel Microprocessors
Socket/Slot Number
Pins
Layout
Voltage
Microprocessor
OverDrives
Sockets for Older Intel Microprocessors and Upgrades

0
168
Inline
5V
486DX
DX2, DX4
169
Inline
5V
486DX, 486SX
DX2, DX4
238
Inline
5V
486DX, 486SX,
DX2
DX2, DX4,
Pentium
237
Inline
3V or 5V
486DX, 486SX,
DX2, DX4
DX2, DX4,
Pentium
273
Inline
5V
60MHz or 66MHz
Pentium
Pentium
320
Staggered
3V
Other Pentium
Pentium
235
Inline
3V
DX4
Pentium
321
Staggered
3V
Other Pentium
Pentium
387
Staggered
3V
Pentium Pro
Pentium Pro
Slots for Intel Architecture Microprocessors

1
242
Inline
3V
Pentium II,
Celeron
330
Inline
3V
Xeon
A (EV6)
NA
Inline
3V
AMD K7
NA
Inline
3V
Merced
SC242
242
Inline
3V
Pentium III,
Celeron
History
No matter the designation or origin, all microprocessors in todays Windows-based computers share a unique characteristic and heritage. All are direct descendents of the very
first microprocessor. The instruction set used by all current computer microprocessors is
rooted in the instructions selected for that first-ever chip. Even the fastest of todays
Pentium 4 chips has, hidden in its millions of transistors, the capability of acting exactly
like that first chip.
In a way, thats good because this backward-looking design assures us that each new generation of microprocessor remains compatible with its predecessors. When a new chip
History
arrives, manufacturers can plug it into a computer and give you reasonable expectations
that all your old software will still work. But holding to the historical standard also heaps
extra baggage on chip designs that holds back performance. By switching to a radically
new design, engineers could create a faster, simpler microprocessorone that could run
circles around any of todays chips, but, alas, one that cant use any of your current programs or operating systems.
Initial Development
The history of the microprocessor stretches back to a 1969 request to Intel by a nowdefunct Japanese calculator company, Busicom. The original plan was to build a series of
calculators, each one different and each requiring a custom integrated circuit. Using conventional IC technology, the project would have required the design of 12 different chips.
The small volumes of each design would have made development costs prohibitive.
Intel engineer Mercian E. (Ted) Hoff had a better idea, one that could slash the necessary
design work. Instead of a collection of individually tailored circuits, he envisioned creating
one general-purpose device that would satisfy the needs of all the calculators. Hoff laid
out an integrated circuit with 2,300 transistors using 10-micron design rules with four-bit
registers and a four-bit data bus. Using a 12-bit multiplexed addressing system, it was able
to address 640 bytes of memory for storing subproducts and results.
Most amazing of all, once fabricated, the chip worked. It became the first general-purpose
microprocessor, which Intel put on sale as the 4004 on November 15, 1971.
The chip was a success. Not only did it usher in the age of low-cost calculators, it also
gave designers a single solid-state programmable device for the first time. Instead of
designing the digital decision-making circuits in products from scratch, developers could
buy an off-the-shelf component and tailor it to their needs simply by writing the appropriate program.
With the microprocessors ability to handle numbers proven, the logical next step was to
enable chips to deal with a broader range of data, including text characters. The 4004s
narrow four-bit design was sufficient for encoding only numbers and basic operationsa
total of 16 symbols. The registers would need to be wider to accommodate a wider repertory. Rather than simply bump up the registers a couple of bits, Intels engineers chose to
go double and design a full eight-bit microprocessor with eight-bit registers and an eightbit data bus. In addition, this endowed the chip with the ability to address a full 16KB of
memory using 14 multiplexed address lines. The result, which required a total of 3450
transistors, was the Intel 8008, introduced in April 1972.
Intel continued development (as did other integrated circuit manufacturers) and, in April
1974, created a rather more drastic revision, the 8080, which required nearly twice as
115
116
many transistors (6,000) as the earlier chip. Unlike the 8008, the new 8080 chip was
planned from the start for byte-size data. Intel gave the 8080 a 16-bit address bus that
could handle a full 64KB of memory and a richer command set, one that embraced all the
commands of the 8008 but went further. This set a pattern for Intel microprocessors:
Every increase in power and range of command set enlarged on what had gone before
rather than replacing it, thus ensuring backward compatibility (at least to some degree) of
the software. To this day, the Intel-architecture chips used in personal computers can run
program code written using 8080 instructions. From the 8080 on, the story of the microprocessor is simply one of improvements in fabrication technology and increasingly complex designs.
With each new generation of microprocessor, manufacturers relied on improving technology in circuit design and fabrication to increase the number and size of the registers in
each microprocessor, broadening the data and address buses to match. When that strategy
stalled, they moved to superscalar designs with multiple pipelines. Improvements in semiconductor fabrication technology made the increasing complexity of modern microprocessor designs both practical and affordable. In the three decades since the introduction of
the first microprocessor, the linear dimensions of semiconductor circuits have decreased
to 1/50th their original size, from 10-micron design rules to 0.13 micron, which means
microprocessor-makers can squeeze 6,000 transistors where only one fit originally. This
size reduction also facilitates higher speeds. Todays microprocessors run nearly 25,000
times faster than the first chip out of the Intel foundry, 2.5GHz in comparison to the
108KHz of the first 4004 chip.
Personal Computer Influence

The success of the personal computer marked a major turning point in microprocessor
design. Before the PC, microprocessor engineers designed what they regarded as the best
possible chips. Afterward, they focused their efforts on making chips for PCs. This change
came between what is now regarded as the first generation of Intel microprocessors and
the third generation, in the years 1981 to 1987.
8086 Family
The engineers who designed the IBM Personal Computer chose to use a chip from the
Intel 8086 family. Intel introduced the 8086 chip in 1978 as an improvement over its first
chips. Intels engineers doubled the size of the registers in its 8080 to create a chip with
16-bit registers and about 10 times the performance. The 16-bit design carried through
completely, also doubling the size of the data bus of earlier chips to 16 bits to move information in and out twice as fast.
In addition, Intel broadened the address bus from 16-bit to 20-bits to allow the 8086
to directly address up to one megabyte of RAM. Intel divided this memory into 64KB
History
segments to make programming and the transition to the new chip easier. A single 16-bit
register could address any byte in a given segment. Another, separate register indicated
which of the segments that address was in.
A year after the introduction of the 8086, Intel introduced the 8088. The new chip was
identical to the 8086 in every way16-bit registers, 20 address lines, and the same command setexcept one. Its data bus was reduced to eight bits, enabling the 8088 to exploit
readily available eight-bit support hardware. At that, the 8088 broke no new ground and
should have been little more than a footnote in the history of the microprocessor.
However, its compromise design that mated 16-bit power with cheap 8-bit support chips
made the 8088 IBMs choice for its first personal computer. With that, the 8088 entered
history as the second most important product in the development of the microprocessor,
after the ground-breaking 4004.
286 Family
After the release of the 8086, Intels engineers began to work on a successor chip with
even more power. Designated the 80286, the new chip was to feature several times the
speed and 16 times more addressable memory than its predecessors. Inherent in its design
was the capability of multitasking, with new instructions for managing tasks and a new
operating mode, called protected mode, that made its full 16MB of memory fodder for
advanced operating systems.
The 80286 chip itself was introduced in 1982, but its first major (and most important)
application didnt come until 1984 with the introduction of IBMs Personal Computer AT.
Unfortunately, this development work began before the PC arrived, and few of the new
features were compatible with the personal computer design. The DOS operating system
for PCs and all the software that ran under it could not take advantage of the chips new
protected modewhich effectively put most of the new chips memory off limits to PC
programs.
With all its innovations ignored by PCs, the only thing the 80286 had going for it was its
higher clock speed, which yielded better computer performance. Initially released running
at 6MHz, computers powered by the 80286 quickly climbed to 8MHz, and then 10MHz.
Versions operating at 12.5MHz, 16MHz, 20MHz, and ultimately 24MHz were eventually
marketed.
The 80286 proved to be an important chip for Intel, although not because of any enduring success. It taught the companys engineers two lessons. First was the new importance
of the personal computer to Intels microprocessor market. Second was licensing.
Although the 80286 was designed by Intel, the company licensed the design to several
manufacturers, including AMD, Harris Semiconductor, IBM, and Siemens. Intel granted
these licenses not only for income but also to assure the chip buyers that they had alternate sources of supply for the 80286, just in case Intel went out of business. At the time,
117
118
Intel was a relatively new company, one of many struggling chipmakers. With the success
of the PC and its future ensured, however, Intel would never again license its designs so
freely.
Even before the 80286 made it to the marketplace, Intels engineers were working on its
successor, a chip designed with the power of hindsight. By then they could see the importance that the personal computers primeval DOS operating system had on the microprocessor market, so they designed to match DOS instead of some vaguely conceived
successor. They also added in enough power to make the chip a fearsome competitor.
386 Family
The next chip, the third generation of Intel design, was the 80386. Two features distinguish it from the 80286: a full 32-bit design, for both data and addressing, and the new
Virtual 8086 mode. The first gave the third generation unprecedented power. The second
made that power useful.
Moreover, Intel learned to tailor the basic microprocessor design to specific niches in the
marketplace. In addition to the mainstream microprocessor, the company saw the need to
introduce an entry level chip, which would enable computer makers to sell lower-cost
systems, and a version designed particularly for the needs of battery-powered portable
computers. Intel renamed the mainstream 80386 as the 386DX, designated an entry-level
chip the 386SX (introduced in 1988), and reengineered the same logic core for low-power
applications as the 386SL (introduced in 1990).
The only difference between the 386DX and 386SX was that the latter had a 16-bit external data bus whereas the former had a 32-bit external bus. Internally, however, both chips
had full 32-bit registers. The origin of the D/S nomenclature is easily explained. The
external bus of the 386DX handled double words (32 bits), and that of the 386SX, single
words (16 bits).
Intel knew it had a winner and severely restricted its licensing of the 386 design. IBM
(Intels biggest customer at the time) got a license only by promising not to sell chips. It
could only market the 386-based microprocessors it built inside complete computers or on
fully assembled motherboards. AMD won its license to duplicate the 386 in court based
on technology-sharing agreements with Intel dating before even the 80286 had been
announced. Another company, Chip and Technologies, reverse-engineered the 386 to
build clones, but these were introduced too latewell after Intel advanced to its fourth
generation of chipsto see much market success.
Age of Refinement
The 386 established Intel Architecture in essentially its final form. Later chips differ only
in details. They have no new modes. Although Intel has added new instructions to the
History
basic 386 command set, almost any commercial software written today will run on any
Intel processor all the way back to the 386but not likely any earlier processor, if the
software is Windows based. The 386 design had proven itself and had become the foundation for a multibillion-dollar software industry. The one area for improvement was performance. Todays programs may run on a 386-based machine, but they are likely to run very
slowly. Current chips are about 100 times faster than any 386.
486 Family
The next major processor after the 386 was, as you might expect, the 486. Even Intel conceded its new chip was basically an improved 386. The most significant difference was
that Intel added three features that could boost processing speed by working around handicaps in circuitry external to the microprocessor. These innovations included an integral
Level One cache that helped compensate for slow memory systems, pipelining within the
microprocessor to get more processing power from low clock speeds, and an integral
floating-point unit that eliminated the handicap of an external connection. As this generation matured, Intel added one further refinement that let the microprocessor race ahead
of laggardly support circuitssplitting the chip so that its core logic and external bus
interface could operate at different speeds.
Intel introduced the first of this new generation in 1989 in the form of a chip then designated 80486, continuing with its traditional nomenclature. When the company added
other models derived from this basic design, it renamed the then-flagship chip as the
486DX and distinguished lower-priced models by substituting the SX suffix and lowpower designs for portable computers using the SL designation, as it had with the third
generation. Other manufacturers followed suit, using the 486 designation for their similar
productsand often the D/S indicators for top-of-the-line and economy models.
In the 486 family, however, the D/S split does not distinguish the width of the data bus.
The designations had become disconnected from their origins. In the 486 family, Intel
economized on the SX version by eliminating the integral floating-point unit. The savings
from this strategy was substantialwithout the floating-point circuitry, the 486SX
required only about half the silicon of the full-fledged chip, making it cheaper to make. In
the first runs of the 486SX, however, the difference was more marketing. The SX chips
were identical to the DX chips except that their floating-point circuitry was either defective or deliberately disabled to make a less capable processor.
As far as hardware basics are concerned, the 486 series retained the principal features of
the earlier generation of processors. Chips in both the third and fourth generations have
three operating modes (real, protected, and virtual 8086), full 32-bit registers, and a 32-bit
address bus enabling up to 4GB of memory to be directly addressed. Both support virtual
memory that extends their addressing to 64TB. Both have built-in memory-management
units that can remap memory in 4KB pages.
119
120
But the hardware of the 486 also differs substantially from the 386 (or any previous Intel
microprocessor). The pipelining in the core logic allows the chip to work on parts of several instructions at the same time. At times the 486 could carry out one instruction every
clock cycle. Tighter silicon design rules (smaller details etched into the actual silicon that
makes up the chip) gave the 486 more speed potential than preceding chips. The small
but robust 8KB integral primary cache helped the 486 work around the memory wait
states that plagued faster 386-based computers.
The streamlined hardware design (particularly pipelining) meant that the 486-level microprocessors could think faster than 386 chips when the two operated at the same clock
speed. On most applications, the 486 proved about twice as fast as a 386 at the same clock
rate, so a 20MHz 486 delivered about the same program throughput as a 40MHz 386.
Pentium Family
In March 1993, Intel introduced its first superscalar microprocessor, the first chip to bear
the designation Pentium. At the time the computer industry expected Intel to continue its
naming tradition and label the new chip the 80586. In fact, the competition was banking
on it. Many had already decided to use that numerical designation for their next generation of products. Intel, however, wanted to distinguish its new chip from any potential
clones and establish its own recognizable brand on the marketplace. Getting trademark
protection for the 586 designation was unlikely. A federal court had earlier ruled that the
386 numeric designation was genericthat is, it described a type of product rather than
something exclusive to a particular manufacturerso trademark status was not available
for it. Intel coined the word Pentium because it could get trademark protection. It also
implied the number 5, signifying fifth generation, much as 586 would have.
Intel has used the Pentium name quite broadly as the designation for mainstream (or
desktop performance) microprocessors, but even in its initial usage the singular Pentium
designation obscured changes in silicon circuitry. Two very different chips wear the plain
designation Pentium. The original Pentium began its life under the code name P5 and
was the designated successor to the 486DX. Characterized by 5-volt operation, low operating speeds, and high power consumption, the Intel made the P5 available only at three
speeds: 60MHz, 66MHz, and 90MHz. Later, Intel refined the initial Pentium design as
the P54C (another internal code name), with tighter design rules and lower voltage operation. These innovations raised the speed potential of the design, and commercial chips
gradually stepped up from 100MHz to 200MHz. The same basic design underlies the
Pentium OverDrive (or P24T) processor used for upgrading 486-based PCs.
In January 1997, Intel enhanced the Pentium instruction set to better handle multimedia
applications and created Pentium Processor with MMX Technology (code-named P55C during
development). These chips also incorporated a larger on-chip primary memory cache,
32KB.
History
To put the latest in Pentium power in the field. Intel reengineered the Pentium with
MMX Technology chip for low-power operation to make the Mobile Pentium with MMX
Technology chip, also released in January 1997. Unlike the deskbound version, the addressing capability of the mobile chip was enhanced by four more lines to allow direct access to
64GB of physical memory.
Mature Design
The Pentium was Intels last CISC design. Other manufacturers were adapting RISC
designs to handle the Intel instruction set and achieving results that put Intel on notice.
The company responded with its own RISC-based design in 1995 that became the standard Intel core logic until the introduction of the Pentium 4 in the year 2000. Intel developed this logic core under the code name P6, and it has appeared in a wide variety of
chips, including those bearing the names Pentium Pro, Pentium II, Celeron, Xeon, and
Pentium III.
Thats not to say all these chips are the same. Although the entire series uses essentially
the same execution units, the floating-point unit continued to evolve throughout the
series. The Pentium Pro incorporates a traditional floating-point unit. That of the
Pentium II is enhanced to handle the MMX instruction set. The Pentium III adds
Streaming SIMD Extensions. In addition, Intel altered the memory cache and bus of these
chips to match the requirements of particular market segments to distinguish the Celeron
and Xeon lines from the plain Pentium series.
The basic P6 design uses its own internal circuits to translate classic Intel instructions into
micro-ops that can be processed in a RISC-based core, which has been tuned using all the
RISC design tricks to massage extra processing speed from the code. Intel called this
design Dynamic Execution. In the standard language of RISC processors, Dynamic
Execution merely indicates a combination of out-of-order instruction execution and the
underlying technologies that enable its operation (branch prediction, register renaming,
and so on).
The P6 pipeline has 12 stages, divided into three sections: an in-order fetch/decode stage,
an out-of-order execution/dispatch stage, and an in-order retirement stage. The design is
superscalar, incorporating two integer units and one floating-point unit.
Pentium Pro
One look and theres no mistaking the Pentium Pro. Instead of a neat square chip, its a
rectangular giant. Intel gives this package the name Multi-Chip Module (MCM). It is also
termed a dual-cavity PGA (pin-grid array) package because it holds two distinct slices of
silicon, the microprocessor core and secondary cache memory. This was Intels first chip
with an integral secondary cache. Notably, this design results in more pins than any previous Intel microprocessor and a new socket requirement, Socket 8 (discussed earlier).
121
122
The main processor chip of the Pentium Pro uses the equivalent of 5.5 million transistors.
About 4.5 million of them are devoted to the actual processor itself. The other million
provide the circuitry of the chips primary cache, which provides a total of 16KB storage
bifurcated into separate 8KB sections for program instructions and data. Compared to
true RISC processors, the Pentium Pro uses about twice as many transistors. The circuitry
that translates instructions into RISC-compatible micro-ops requires the additional transistor logic.
The integral secondary RAM cache fits onto a separate slice of silicon in the other cavity
of the MCM. Its circuitry involves another 15.5 million transistors for 256KB of storage
and operates at the same speed as the core logic of the rest of the Pentium Pro.
The secondary cache connects with the microprocessor core logic through a dedicated 64bit bus, termed a back-side bus, that is separate and distinct from the 64-bit front-side bus
that connects to main memory. The back-side bus operates at the full internal speed of the
microprocessor, whereas the front-side bus operates at a fraction of the internal speed of
the microprocessor.
The Pentium Pro bus design superficially appears identical to that of the Pentium with
32-bit addressing, a 64-bit data path, and a maximum clock rate of 66MHz. Below the
surface, however, Intel enhanced the design by shifting to a split-transaction protocol.
Whereas the Pentium (and, indeed, all previous Intel processors) handled memory accessing as a two-step process (on one clock cycle the chip sends an address out the bus, and
reads the data at the next clock cycle), the Pentium Pro can put an address on the bus at
the same time it reads data from a previously posted address. Because the address and data
buses use separate lines, these two operations can occur simultaneously. In effect, the
throughput of the bus can nearly double without an increase in its clock speed.
The internal bus interface logic of the Pentium Pro is designed for multiprocessor systems. Up to four Pentium Pro chips can be directly connected together, pin for pin, without any additional support circuitry. The computers chipset arbitrates the combination.
Pentium II
The chief distinctions of the Pentium II are its addition of MMX technology and new
package called the Single Edge Contact cartridge, or SEC cartridge. The socket it plugs into
is termed Slot 1.
One underlying reason for the cartridge-style design is to accommodate the Pentium IIs
larger secondary cache, which is not integral to the chip package but rather co-mounted
on the circuit board inside the cartridge. The 512KB of static cache memory connect
through a 64-bit back-side bus. Note that the secondary cache memory of a Pentium II
operates at one-half the speed of the core logic of the chip itself. This reduced speed is,
of course, a handicap. It was a design expediency. It lowers the cost of the technology,
History
allowing Intel to use off-the-shelf cache memory (from another manufacturer, at least initially) in a lower-cost package. The Pentium II secondary cache design has another limitation. Although the Pentium II can address up to 64GB of memory, its cache can track
only 512MB. The Pentium II also has a 32KB primary cache thats split with 16KB
assigned to data and 16KB to instructions. Table 5.3 summarizes the Intel Pentium II line.
TA B L E 5 . 3
Introduction Dates and Speeds of Intel Pentium II Microprocessors
Speed
Introduction Date
Cache Size
Front-Side
Bus Speed
Instruction Set
Design Rules
233
May 7, 1997
512KB
66MHz
IA32
0.35 micron
266
May 7, 1997
512KB
66MHz
IA32
0.35 micron
300
May 7, 1997
512KB
66MHz
IA32
0.35 micron
333
January 26, 1997
512KB
66MHz
IA32
0.25 micron
350
April 15, 1998
512KB
66MHz
IA32
0.25 micron
400
April 15, 1998
512KB
66MHz
IA32
0.25 micron
450
August 24, 1998
512KB
66MHz
IA32
0.25 micron
Mobile Pentium II
To bring the power of the Pentium II processor to notebook computers, Intel reengineered the desktop chip to reduce its power consumption and altered its packaging to fit
slim systems. The resulting chipthe Mobile Pentium II, introduced on April 2, 1997
preserved the full power of the Pentium II while sacrificing only its multiprocessor support. The power savings comes from two changes. The core logic of the Mobile Pentium
II is specifically designed for low-voltage operation and has been engineered to work well
with higher external voltages. It also incorporates an enriched set of power-management
modes, including a new QuickStart mode that essentially shuts down the chip, except for
the logic that monitors for bus activity by the PCI bridge chip, and allows the chip to
wake up when its needed. This design, because it does not monitor for other processor
activity, prevents the Mobile Pentium II from being used in multiprocessor applications.
The Mobile Pentium II can also switch off its cache clock during its sleep or QuickStart
states.
Initially, the Mobile Pentium II shared the same P6 core logic design and cache design
with the desktop Pentium II (full-speed 32KB primary cache and half-speed 512KB secondary cache inside its mini-cartridge package). However, as fabrication technology
improved, Intel was able to integrate the secondary cache on the same die as the processor
core, and on January 25, 1999, the company introduced a new version of the Mobile
Pentium II with an integral 256KB cache operating at full core speed. Unlike the Pentium
123
124
II, the mobile chip has the ratio between its core and bus clocks fixed at the factory to
operate with a 66MHz front-side bus. Table 5.4 lists the introduction dates and basic
characteristics of the Mobile Pentium II models.
TA B L E 5 . 4
Introduction Dates and Speed of Intel Mobile Pentium II Microprocessors
Speed
Introduction Date
Cache Size
Front-Side Bus Speed
Design Rules
233MHz
April 2, 1998
512KB
66MHz
0.25 micron
266MHz
April 2, 1998
512KB
66MHz
0.25 micron
266MHz
January 25, 1999
256KB
66MHz
0.25 micron
300MHz
September 9, 1998
512KB
66MHz
0.25 micron
300MHz
January 25, 1999
256KB
66MHz
0.25 micron
333MHz
January 25, 1999
256KB
66MHz
0.25 micron
366MHz
January 25, 1999
256KB
66MHz
0.25 micron
400MHz
June 14, 1999
256KB
66MHz
0.25 micron
Pentium II Celeron
Introduced on March 4, 1998, the Pentium II Celeron was Intels entry-level processor
derived from the Pentium II. Although it had the same processor core as what was at the
time Intels premier chip (the second-generation Pentium II with 0.45-micron design
rules), Intel trimmed the cost of building the chip by eliminating the integral 512KB secondary (Level Two) memory cache installed in the Pentium II cartridge. The company
also opted to lower the packaging cost of the chip by omitting the metal outer shell of the
full Pentium II and instead leaving the Celerons circuit board substrate bare. In addition,
the cartridge-based Celeron package lacked the thermal plate of the Pentium II and the
latches that secure it to the slot. Intel terms the Celeron a Single Edge Processor Package to
distinguish it from the Single Edge Contact cartridge used by the Pentium II.
In 1999, Intel introduced a new, lower-cost package for the Celeron, a plastic pin-grid
array (PPGA) shell that looks like a first generation Pentium on steroids. It has 370 pins
and mates with Intels PGA370 socket. The chip itself measures just under two inches
square (nominally 49.5 millimeters) and about three millimeters thick, not counting the
pins, which hang down another three millimeters or so (the actual specification is 3.05 to
3.30 millimeters).
When the Celeron chip was initially introduced, the absence of a cache made such a hit
on the performance that Intel was forced by market pressure to revise its design. In
August, 1998, the company added a 128KB cache operating at one-half core speed to the
Celeron. Code names distinguished the two chips: The first Celeron was code-named
History
Covington during development; the revised chip was code-named Mendocino. Intel further
increased the cache to 256KB on October 2, 2001, with the introduction of a 1.2GHz
Celeron variant.
Intel also distinguished the Celeron from its more expensive processor lines by limiting its
front-side bus speed to 66MHz. All Celerons sold before January 3, 2001 were limited to
that speed. With the introduction of the 800MHz Celeron, Intel kicked the chips frontside bus up to 100MHz. With the introduction of a 1.7GHz Celeron on May 2, 2002,
Intel started quad-clocking the chips front-side bus, yielding an effective data rate of
400MHz.
Intel also limited the memory addressing of the Celeron to 4GB of physical RAM by
omitting the four highest address bus signals used by the Pentiums II and III from the
Celeron pin-out. The Celeron does not support multiprocessor operation, and, until Intel
introduced the Streaming SIMD Extensions to the 1.2GHz version, the Celeron understood only the MMX extension to the Intel instruction set.
Table 5.5 lists the features and introduction dates of various Celeron models.
TA B L E 5 . 5
Introduction Dates and Specifications of Intel Pentium II Celeron Microprocessors
Speed
Introduction Date
Cache Size
Front-Side
Bus
Instruction
Set
Design Rules
266
March 4, 1998
None
66MHz
MMX
0.35 micron
300
June 8, 1998
None
66MHz
MMX
0.35 micron
300A
August 24, 1998
128KB
66MHz
MMX
0.25 micron
333
August 24, 1998
128KB
66MHz
MMX
0.25 micron
366
January 4, 1999
128KB
66MHz
MMX
0.25 micron
400
January 4, 1999
128KB
66MHz
MMX
0.25 micron
433
March 22, 1999
128KB
66MHz
MMX
0.25 micron
450
August 2, 1999
128KB
66MHz
MMX
0.25 micron
566
March 29, 2000
128KB
66MHz
MMX
0.25 micron
600
March 29, 2000
128KB
66MHz
MMX
0.18 micron
633
June 26, 2000
128KB
66MHz
MMX
0.18 micron
667
June 26, 2000
128KB
66MHz
MMX
0.18 micron
700
June 26, 2000
128KB
66MHz
MMX
0.18 micron
733
November 13, 2000 128KB
66MHz
MMX
0.18 micron
continues
125
126
TA B L E 5 . 5
Continued
Front-Side
Bus
Instruction
Set
Design Rules
November 13, 2000 128KB
66MHz
MMX
0.18 micron
800
January 3, 2001
128KB
100MHz
MMX
0.18 micron
850
May 21, 2001
128KB
100MHz
MMX
0.18 micron
900
July 2, 2001
128KB
100MHz
MMX
0.18 micron
950
August 31, 2001
128KB
100MHz
MMX
0.18 micron
1000
August 31, 2001
128KB
100MHz
MMX
0.18 micron
1100
August 31, 2001
128KB
100MHz
MMX
0.18 micron
1200
October 2, 2001
256KB
100MHz
SSE
0.13 micron
1300
January 3, 2002
256KB
100MHz
SSE
0.13 micron
1400
May 15, 2002
256KB
100MHz
SSE
0.13 micron
1700
May 15, 2002
256KB
400MHz
SSE2
0.13 micron
1800
June 12, 2002
256KB
400MHz
SSE2
0.13 micron
Speed
Introduction Date
750
Cache Size
Pentium II Xeon
In 1998, Intel sought to distinguish its higher performance microprocessors from its economy line. In the process, the company created the Xeon, a refined Pentium II microprocessor core enhanced by a higher-speed memory cache, one that operated at the same
clock rate as the core logic of the chip.
At heart, the Xeon is a full 32-bit microprocessor with a 64-bit data bus, as with all
Pentium-series processors. Its address bus provides for direct access to up to 64GB of
RAM. The internal logic of the chip allows for up to four Xeons to be linked together
without external circuitry to form powerful multiprocessor systems.
A sixth generation processor, the Xeon is a Pentium Pro derivative by way of the standard
Pentium II. It incorporates two 12-stage pipelines to make what Intel terms Dynamic
Execution micro-architecture.
The Xeon incorporates two levels of caching. One is integral to the logic core itself, a primary 32KB cache split 16KB for instructions, 16KB for data. In addition, a separate secondary cache is part of the Xeon processor module but is mounted separately from the
core logic on the cartridge substrate. This integral-but-separate design allows flexibility in
configuring the Xeon. Current chips are available equipped with either 512KB or 1MB of
L2 cache, and the architecture and slot design allow for secondary caches of up to 2MB.
This integral cache runs at the full core speed of the microprocessor.
History
This design required a new interface, tagged Slot 2 by Intel.

Initially the core operating speed of the Xeon started where the Pentium II left off (at the
time, 400MHz) and followed the Pentium II up to 450MHz.
The front-side bus of the Xeon was initially designed for 100MHz operation, although
higher speeds are possible and expected. A set of contacts on the SEC cartridge allows the
motherboard to adjust the multiplier that determines the ratio between front-side bus and
core logic speed.
The independence of the logic core and cache is emphasized by the power requirements
of the Xeon. Each section requires its own voltage level. The design of the Xeon allows
Intel flexibility in the power requirements of the chip through a special coding scheme. A
set of pins indicates the core voltage and the cache voltage required by the chip, and the
chip expects the motherboard to determine the requirements of the board and deliver the
required voltages. The Xeon design allows for core voltages as low as 1.8 volts or as high
as 2.1 volts (the level required by the first chips). Cache voltage requirements may reach
as high as 2.8 volts. Nominally, the Xeon is a 2-volt chip.
Overall, the Xeon is optimized for workstations and servers and features built-in provide
support for up to four identical chips in a single computer. Table 5.6 summarizes the original Xeon product line.
TA B L E 5 . 6
Introduction Dates and Specification of Intel Pentium II Xeon Microprocessors
Speed
Introduction Date
Cache Size
Front-Side
Bus Speed
Instruction
Set
Design Rules
400MHz
June 29, 1998
512KB
100MHz
IA32
0.25 micron
400MHz
June 29, 1998
1MB
100MHz
IA32
0.25 micron
450MHz
October 6, 1998
512KB
100MHz
IA32
0.25 micron
450MHz
January 5, 1999
512KB
100MHz
IA32
0.25 micron
450MHz
January 5, 1999
1MB
100MHz
IA32
0.25 micron
450MHz
January 5, 1999
2MB
100MHz
IA32
0.25 micron
Pentium II OverDrive
To give an upgrade path for systems originally equipped with the Pentium Pro processor,
Intel developed a new OverDrive line of direct-replacement upgrades. These Pentium II
OverDrive chips fit the same zero-insertion force Socket 8 used by the Pentium Pro, so
you can slide one chip out and put the other in. Dual-processor systems can use two
OverDrive upgrades. Intel warns that some systems may require a BIOS upgrade to
accommodate the OverDrive upgrade.
127
128
The upgrade offers the revised design of the Pentium II (which means better 16-bit operation) as well as higher clock speeds. The chip also can earn an edge over ordinary
Pentium II chips operating at the same speedsthe 512KB secondary cache in the
OverDrive chip operates at full core logic speed, not half speed as in the Pentium II.
Current Products
Although more than a dozen companies make microprocessors, any new personal computer you buy will likely be based on a chip from one of only three companies: Advanced
Micro Devices, Intel, or Transmeta. Intel, the largest semiconductor-maker in the world,
makes the majority of computer processorsabout 80 percent in the first quarter of the
year 2002, according to Mercury Research. In the same period, AMD sold 18.2 percent of
the chips destined for personal computers.
Intel earned its enviable position by not only inventing the microprocessor but also by a
quirk of fate. IBM chose one of its chips for its first Personal Computer in 1981, the
machine that all modern personal computers have been patterned after. Computers must
use Intel chips or chips designed to match the Intel Architecture to be able to run todays
most popular software. Because microprocessors are so complicated, designing and building them requires a huge investment, which prevents new competitors from edging into
the market.
Intel Microprocessors
Currently Intel sells microprocessors under several brand names. The most popular is the
Pentium 4, which Intel claims as its best-selling microprocessor, ever. But the Pentium 4
is not a single chip design. Rather, its a trade name. Intel has marketed two distinctly different chip designs as Pentium 4. More confusing still, Intel markets the same core logic
under more than one name. At one time, the Pentium, Celeron, and Xeon all used essentially the same internal design. The names designated the market segment Intel hoped to
carve for the respective chipsCeleron for the price-sensitive low end of the personal
computer marketplace, Pentium for the mainstream, and Xeon for the pricey, high-end
server marketplace. Although Intel did tailor some features to justify its market positioning of the chips, they nevertheless shared the same circuitry deep inside.
Today, that situation has changed. Intel now designs Xeon chips separately, and Celerons
often retain older designs and technologies longer than mainstream Pentiums. The market position assigned the microprocessor names remains the same. Intel offers Celeron
chips for the budget conscious. It sacrifices the last bit of performance to make computers
more affordable. Pentium processors are meant for the mainstream (most computer purchasers) and deliver full performance for single-user computers. Xeons are specialized
Current Products
microprocessors designed primarily for high-powered server systems. Intel has added a
further name, Itanium, to its trademark lineup. The Itanium uses a new architecture (usually termed IA64, shorthand for 64-bit Intel Architecture) that is not directly compatible
with software meant for other Intel chips.
The important lesson is that the names you see on the market are only brand names and
do not reflect whats inside a chip. Some people avoid confusion by using the code name
the manufacturer called a given microprocessor design during its development. These
code names allow consumers to distinguish the Northwood processor from the
Willamette, both of which are sold as Pentium 4. The Northwood is a newer design with
higher-speed potentials. Table 5.7 lists many of Intels microprocessor code names.
TA B L E 5 . 7
Intel Microprocessor Code Names
Code Name
Corresponding Brand Name
Klamath
Original Pentium II
Deschutes
Improved Pentium II (100MHz bus)
Katmai
Original Pentium III (0.25 micron)
Coppermine
Pentium III (0.18 micron)
Willamette
Original Pentium 4 (0.18 micron)
Northwood
Pentium 4 (0.13 micron)
Foster
Pentium 4 Xeon (0.18 micron)
Prestonia
Pentium 4 Xeon (0.13 micron)
Nocona
Proposed Pentium 4 Xeon (0.09 micron)
Gallatin
Multiprocessor Prestonia (0.13 micron)
Merced
Original Itanium (0.18 micron)
McKinley
Itanium 2 (0.18 micron)
Madison
Proposed Itanium 2 (0.13 micron)
Deerfield
Proposed Desktop Itanium 2
Banias
Proposed high-performance IA-32 portable chip
Pentium III
Announced in January and officially released on February 26, 1999, the Pentium III was
the swan song for Intels P6 processor core, developed for the Pentium Pro. Code-named
Katmai during its development, the Pentium III chip is most notable for adding SSE to
129
130
the Intel microprocessor repertory. SSE is a compound acronym for Streaming SIMD
Extensions (SIMD itself being an acronym for Single Instruction, Multiple Data). SIMD
technology allows one microprocessor instruction to operate across several bytes or words
(or even larger blocks of data). In the Pentium III, SSE (formerly known as the Katmai
New Instructions or KNI) is a set of 70 new SIMD codes for microprocessor instructions
that allows programs to specify elaborate three-dimensional processing functions with a
single command.
Unlike the MMX extensions, which added no new registers to the basic Pentium design
and instead simply redesignated the floating-point unit registers for multimedia functions,
Intels Streaming SIMD Extensions add new registers to Intel architecture, pushing the
total number of transistors inside the core logic of the chip above 9.5 million.
At heart, the Pentium III uses the same core logic as its Pentium II and Pentium Pro forebears, amended to handle its larger instruction set. The enhancements chiefly involve the
floating-point unit, which does double-duty processing multimedia instructions. In other
words, the Pentium III does not mark a new generation of microprocessor technology or
performance. Even Intel noted that on programs that do not take advantage of the
Streaming SIMD Extensions of the Pentium III, the chips deliver performance thats
about the same as a Pentium II.
Although the initial fabrication used 0.25-micron technology, Intel rapidly shifted to 0.18micron technology with the Coppermine design. The result is that the circuitry of the
chip takes less silicon (making fabrication less expensive), requires less power, and is able
to operate at higher speeds. The initial Pentium III releases, and all versions through at
least the 600MHz chip, operate with a 100MHz memory bus. Many of the new chips
using the Coppermine core ratchet the maximum memory speed to 133MHz using
Rambus memory technology, although some retain a 100MHz maximum memory speed.
In going to the new Coppermine design, Intel replaced earlier Pentium III chips with
0.25-micron design features, in particular those at the 450, 533, 550, and 600MHz speeds.
The newer chips are designed with the suffix E. In addition, to distinguish chips with
133MHz front-side bus capability (when 100MHz versions of the chip were once offered),
Intel added a B suffix to the designation of 533 and 600MHz chips that are capable of
running their memory buses at the higher 133MHz speed.
The Pentium III was the first Intel processor to cross the line at a 1GHz clock speed with
a chip released on March 8, 2000. The series ended its development run at 1.13GHz on
July 31, 2000, although Intel continued to manufacturer the chip into 2002.
The Pentium III was designed to plug into the same Slot 1 as the Pentium II, however,
the Pentium III now comes in three distinct packages. One, the SEC cartridge, is familiar
Current Products
from the Pentium II. Pentium III is also available in the SEC cartridge 2 (or SECC2). In
all aspects, the SECC2 is identical to the SEC cartridge and plugs into the same slot (Slot
1, which Intel has renamed SC242), but the SECC2 package lacks the thermal plate of the
earlier design. Instead, the SECC2 is designed to mate with an external heatsink, and,
because of the lack of the thermal plate, it will make a better thermal connection for more
effective cooling. (Well, it makes sense when Intel explains it.) In addition, the 450 and
550E are available in a new package design termed FC-PGA (for Flip-Chip Pin Grid
Array), which is more compact and less expensive than the cartridge design.
As with the Pentium II, the 0.25-micron versions of the Pentium III have a 32KB integral
primary cache and include a 512KB secondary cache on the same substrate but not in the
same hermetic package at the core CPU. The secondary cache runs at half the chip speed.
The newer 0.18-micron versions have a smaller, 256KB secondary cache but operate it at
full chip speed and locate it on the same silicon as the core logic. In addition, Intel has
broadened the data path between the core logic and the cache to enhance performance
(Intel calls this Advanced Transfer Cache technology). According to Intel, these improvements give the newer Coppermine-based (0.18-micron) Pentium III chips a 25-percent
performance advantage over older Pentium III chips operating at the same clock speed.
The entire Pentium III line supports multiprocessing with up to two chips.
The most controversial aspect of the Pentium III lineup is its internal serial number.
Hard-coded into the chip, this number is unique to each individual microprocessor.
Originally Intel foresaw that a single commandincluding a query from a distant Web
sitewould cause the chip to send out its serial number for positive identification (of the
chip, of the computer it is in, and of the person owning or using the computer). Intel
believed the feature would improve Internet security, not to mention allowing the company to track its products and detect counterfeits. Consumer groups saw the feature as
in invasion of privacy, and under threat of boycott Intel changed its policy. Where formerly the Pentium III would default to making the identification information available,
after the first production run of the new chip, the identification would default to off and
require a specific software command to make the serial number accessible. Whether the
chip serial number is available becomes a setup feature of the BIOS in PCs using the
Pentium III chip, although a software command can override that setting. In other words,
someone can always interrogate your PC to discover your Pentium IIIs serial number.
Therefore, you might want to watch what you say online when you run with the Pentium
III. Table 5.8 summarizes the history of the Pentium III.
131
132
TA B L E 5 . 8
Introduction Dates and Specification of Intel Pentium III Microprocessors
Speed
Introduction Date
Cache Size
Front-Side
Bus Speed
Instruction
Set
Design Rules
450
February 26, 1999
512KB
100MHz
SSE
0.25 micron
500
February 26, 1999
512KB
100MHz
SSE
0.25 micron
500
October 25, 1999
256KB
100MHz
SSE
0.18 micron
533
October 25, 1999
256KB
133MHz
SSE
0.18 micron
550
May 17, 1999
512KB
100MHz
SSE
0.25 micron
550
October 25, 1999
256KB
100MHz
SSE
0.18 micron
600
August 2, 1999
512KB
100MHz
SSE
0.25 micron
600E
October 25, 1999
256KB
100MHz
SSE
0.18 micron
600EB October 25, 1999
256KB
133MHz
SSE
0.18 micron
650
October 25, 1999
256KB
100MHz
SSE
0.18 micron
667
October 25, 1999
256KB
133MHz
SSE
0.18 micron
700
October 25, 1999
256KB
100MHz
SSE
0.18 micron
733
October 25, 1999
256KB
133MHz
SSE
0.18 micron
750
December 20, 1999
256KB
100MHz
SSE
0.18 micron
800
December 20, 1999
256KB
133MHz
SSE
0.18 micron
850
March 20, 2000
256KB
100MHz
SSE
0.18 micron
866
March 20, 2000
256KB
133MHz
SSE
0.18 micron
933
May 24, 2000
256KB
133MHz
SSE
0.18 micron
1000
March 8, 2000
256KB
133MHz
SSE
0.18 micron
1130
July 31, 2000
256KB
133MHz
SSE
0.18 micron
Pentium III Xeon

To add its Streaming SIMD Extensions to its server products, on March 17, 1999, Intel
introduced the Pentium III Xeon. As with the Pentium III itself, the new instructions are
the chief change, but they are complemented by a shift to finer technology. As a result,
the initial new Xeons start with a speed of 500MHz. At this speed, Intel offers the chip
with either a 512KB, 1MB, or 2MB integral secondary cache operating at core speed. The
new Slot 2 chips also incorporate the hardware serial number feature of the Pentium III
chip.
Current Products
Developed under the code name Tanner, the Pentium III Xeon improved upon the original (Pentium II) Xeon with additions to the core logic to handle Intels Streaming SIMD
Extensions. Aimed at the same workstation market as the original Xeon, the Pentium III
Xeon is distinguished from the ordinary Pentium III by its larger integral cache, its Slot 2
packaging, and its wider multiprocessor supportthe Pentium III Xeon design allows for
servers with up to eight processors.
When Intel introduced its Coppermine 0.18-micron technology on October 25, 1999, it
unveiled three new Pentium III Xeon versions with speed ratings up to 733MHz. Except
for packaging, however, these new Xeons differed little from the ordinary Pentium III
line. As with the mainstream processors, the Xeons supported a maximum of two processors per system and had cache designs identical to the ordinary Pentium III with a 256KB
secondary cache operating at full processor speed using wide-bus Advanced Transfer
Cache technology. In May of 2000, Intel added Xeons with larger, 1MB and 2MB caches
as well as higher-speed models with 256KB caches. Table 5.9 lists the characteristics of all
of Intels Pentium III Xeon chips.
TA B L E 5 . 9
Introduction Dates and Specification of Intel Pentium III Xeon Microprocessors
Speed
Introduction Date
Cache Size
Front-Side
Bus Speed
Instruction
Set
Design Rules
500MHz
March 17, 1999
512KB
100MHz
SSE
0.25 micron
500MHz
March 17, 1999
1MB
100MHz
SSE
0.25 micron
500MHz
March 17, 1999
2MB
100MHz
SSE
0.25 micron
550MHz
March 17, 1999
512KB
100MHz
SSE
0.25 micron
550MHz
March 17, 1999
1MB
100MHz
SSE
0.25 micron
550MHz
March 17, 1999
2MB
100MHz
SSE
0.25 micron
600MHz
October 25, 1999
256KB
133MHz
SSE
0.18 micron
667MHz
October 25, 1999
256KB
133MHz
SSE
0.18 micron
700MHz
May 22, 2000
1MB
100MHz
SSE
0.18 micron
700MHz
May 22, 2000
2MB
100MHz
SSE
0.18 micron
733MHz
October 25, 1999
256KB
133MHz
SSE
0.18 micron
800MHz
January 12, 2000
256KB
133MHz
SSE
0.18 micron
866MHz
April 10, 2000
256KB
133MHz
SSE
0.18 micron
900MHz
March 21, 2001
2MB
100MHz
SSE
0.18 micron
933MHz
May 24, 2000
256KB
133MHz
SSE
0.18 micron
133
134
Intel Pentium 4
Officially introduced on November 20, 2000, the Pentium 4 is Intels newest and most
powerful microprocessor core for personal computers. According to Intel, the key advance
made by the new chip is its use of NetBurst micro-architecture, which can be roughly
explained as a better way of translating program instructions into the micro-ops that the
chip actually carries out. NetBurst is the first truly new Intel core logic design since the
introduction of the P6 (Pentium Pro) in 1995.
Part of the innovation is an enhancement to the instruction set; another part is an
improvement to the underlying hardware. All told, the first Pentium 4 design required the
equivalent of 42 million transistors.
Chips designated Pentium 4 actually use one of two designs. Intel code-named the early
chips Willamette and used 0.18-micron design rules in their fabrication. At initial release,
these chips operated at 1.4GHz and 1.5GHz, but Intel soon upped their clock speeds. At
the time, Intel and AMD were in a horserace for the fastest microprocessor, and the title
shifted between the Athlon and Pentium 4 with each new chip release.
In 2002, Intel shifted to 0.13-micron design rules with a new processor code designed
under the name Northwood. This shift resulted in a physically smaller chip that also allows
more space for cache memorywhereas the Willamette chips boasts 256KB of on-chip
Level Two cache operating at full core speed, the Northwood design doubles that to
512KB. The difference is in size only. Both chips have a 256-bit-wide connection with the
caches, which use an eight-way set-associative design.
Of particular note, the Pentium 4 uses a different system bus from that of the Pentium III
chip. As a practical matter, that means the Pentium 4 requires different chipsets and
motherboards from those of the Pentium III. Although this is ordinarily the concern of
the computer manufacturer, the new design has important benefits. It adds extra speed to
the system (memory) bus by shifting data up to four times faster than older designs using
a technology Intel calls Source-Synchronous Transfer. In effect, this signaling system packs
four bits of information into each clock cycle, so a bus with a 133MHz nominal clock
speed can shift data at an effective rate of 533MHz. The address bus is double-clocked,
signaling twice in each clock cycle, yielding an effective rate of 266MHz. In that the
Pentium 4, like earlier Pentiums, has a 64-bit-wide data bus, that speed allows the
Pentium 4 to move information at a peak rate of 4.3GBps (that is, 8 bytes times
533MHz).
Only Northwood chips rated at 2.26GHz, 2.4GHz, and 2.53GHz have 533MHz system
buses. Other Northwood chips, as well as all Willamette versions, use a 400MHz system
bus (that is, a quadruple-clocked 100MHz bus). Note that chips operating at 2.4GHz may
Current Products
have either a 400MHz or 533MHz system bus. The system bus speed is set in the system
design and cannot be varied through hardware or software.
The Pentium 4 has three execution units. The two integer arithmetic/logic units (ALUs)
comprise what Intel calls a rapid execution engine. They are rapid because they operate at
twice the speed of the rest of the core logic (that is, 5.06GHz in a 2.53GHz chip), executing up to two instructions in each clock cycle. The registers in each ALU are 32 bits wide.
Unlike previous Intel floating-point units, the registers in the Pentium 4 FPU are 128 bits
wide. The chief benefit of these wider registers is in carrying out multimedia instructions
using a further enhancement on Intels Streaming SIMD Extensions (SSE), a set of 144
new instructions (mainly aimed at moving bytes in and out of the 128-bit registers but
also including double-precision floating-point and memory-management instructions)
called SSE2.
Intel lengthened the pipelines pumping instructions into the execution units to 20 stages,
the longest of any microprocessor currently in production. Intel calls this design
hyperpipelined technology.
One way Intel pushes more performance from the Pentium 4 is by double-clocking the
integer units in the chip. They operate at twice the external clock frequency applied to the
chip (that is, at 3GHz in the 1.5GHz Pentium 4 chip). Balancing the increased speed is a
400MHz system bus throughput as well as an improved primary cache and integral
256KB secondary cache. Intel connects this secondary cache to the rest of the chip
through a new 256-bit-wide bus, double the size of those in previous chips.
Intel also coins the term hyperpipelining to describe the Pentium 4. The term refers to
Intels doubling the depth of the instruction pipeline, as compared to the previous lineleader, the Pentium III. One (but not all) of the pipelines in the Pentium 4 stretches out
for 20 stages. Intel claims that the new NetBurst micro-architecture enabled the successful
development of the long pipeline because it minimizes the penalties associated with mispredicting instruction branches.
The Pentium 4 recognizes the same instruction set as previous Intel microprocessors,
including the Streaming SIMD Extensions introduced with the Pentium III, but the
Pentium 4 adds 144 more instructions to the list. The result is termed by Intel SSE2. The
chip has the same basic data and address bus structure as the Pentium III, allowing it to
access up to 64GB of physical memory eight bytes (64-bits) at a time.
Table 5.10 summarizes the Intel Pentium 4 line.
135
136
TA B L E 5 . 1 0
Introduction Dates and Specification of Intel Pentium 4 Microprocessors
Speed (MHz) Introduction Date
Cache Size
Front-Side
Bus Speed
Instruction
Set
Design Rules
1400
November 20, 2000
256KB
400MHz
SSE2
0.18 micron
1500
November 20, 2000
256KB
400MHz
SSE2
0.18 micron
1600
July 2, 2001
256KB
400MHz
SSE2
0.18 micron
1700
April 23, 2001
256KB
400MHz
SSE2
0.18 micron
1800
July 2, 2001
256KB
400MHz
SSE2
0.18 micron
1900
August 27, 2001
256KB
400MHz
SSE2
0.18 micron
2000
August 27, 2001
256KB
400MHz
SSE2
0.18 micron
2000
August 27, 2001
512KB
400MHz
SSE2
0.13 micron
2200
January 7, 2002
512KB
400MHz
SSE2
0.13 micron
2260
May 6, 2002
512KB
533MHz
SSE2
0.13 micron
2400
April 2, 2002
512KB
400MHz
SSE2
0.13 micron
2400
May 6, 2002
512KB
533MHz
SSE2
0.13 micron
2530
May 6, 2002
512KB
533MHz
SSE2
0.13 micron
2660
August 26, 2002
512KB
533MHz
SSE2
0,13 micron
2800
August 26, 2002
512KB
533MHz
SSE2
0.13 micron
3060
November 14, 2002
512KB
533MHz
SSE2
0.13 micron
Xeon (Pentium 4)
On May 21, 2001, Intel released the first of its Xeon microprocessors built using the
NetBurst core logic of the Pentium 4 microprocessor. To optimize the chip for use in computer servers, the company increased the secondary cache size of the chip, up to 2MB of onchip cache. For multiprocessor systems, Intel later derived a separate chip, the Xeon MP
processor, for servers with two to four microprocessors. The chief difference between the MP
chip and the base Xeon chip is the former chips cachinga three-level design. The primary
cache is 8KB, the secondary cache is 256KB, and the tertiary cache is either 512KB or 1MB.
Table 5.11 summarizes the Pentium 4 Xeon, including Xeon MP processors.
TA B L E 5 . 1 1
Summary of Intel Pentium 4 Xeon and Xeon MP Microprocessors
Processor
Clock Speed
Introduction Date
Technology
Xeon
2.8GHz
September 11, 2002
0.13 micron
Xeon
2.6GHz
September 11, 2002
0.13 micron
Current Products
Processor
Clock Speed
Introduction Date
Technology
Xeon
2.4GHz
April 23, 2002
0.13 micron
Xeon
2.2GHz
February 25, 2002
0.13 micron
Xeon
2.0GHz
February 25, 2002
0.13 micron
Xeon
2.0GHz
September 25, 2001
0.13 micron
Xeon
1.8GHz
February 25, 2002
0.13 micron
Xeon
1.7GHz
May 21, 2001
0.13 micron
Xeon
1.5GHz
May 21, 2001
0.13 micron
Xeon
1.4GHz
May 21, 2001
0.13 micron
Xeon MP
1.6GHz
March 12, 2002
0.13 micron
Xeon MP
1.5GHz
March 12, 2002
0.13 micron
Xeon MP
1.4GHz
March 12, 2002
0.13 micron
Mobile Processors
The diversity of models that Intel puts on the desktop is exceeded only by the number of
microprocessors it makes for portable computers. The current lineup includes four major
models, each of which includes chips meant to operate on three different voltage levels, in
a wide range of frequencies. The choices include Mobile Celeron, Mobile Pentium III
(soon to fall from the product line, as of this writing), Mobile Pentium III-M, and Mobile
Pentium 4-M. Each has its target market, with Celeron at the low end, Pentium 4 at the
highest, and the various Pentium III chips for everything in between. Each chip shares
essentially the same core as the desktop chip bearing the same designation. But mobile
chips have added circuitry for power management and, in some models, different packaging.
Microprocessors for portable computers differ from those meant for desktop systems in
three ways: operating power, power management, and performance. The last is a result of
the first and, in normal operation, the second.
To help portable computers run longer from a single charge of their batteries, Intel and
other microprocessor manufacturers reduce the voltage at which their chips operate. Intel,
in fact, produces chips in three voltage ranges, which it calls very (or sometimes, ultra) low
voltage, low voltage, and nothing. Very-low-voltage chips operate as low as 0.95 volts.
Low-voltage chips dip down to about 1.1. Low-voltage operation necessitates lower-speed
operation, which limits performance. Of the chips that Intel produces in all three power
levels, the very-low-voltage chips inevitably have the slowest megahertz rating.
137
138
Power management aids in the same endprolonging battery lifeand incidentally prevents your laptop computer from singeing you should you operate it on your lap. Some
machines still get uncomfortably hot.
Mobile Celeron
Intel aims its Mobile Celeron at the budget market. The chips are not only restricted to
lower clock speeds than their Pentium siblings, but most Mobile Celeron chips lack Intels
SpeedStep power-saving technology. The Mobile Celeron perpetually lags Intels desktop
processors in other performance indexes. For example, much like the desktop Celerons,
the mobile chips usually are one step out-of-date in front-side bus speed. Although Intel
has bumped its top memory bus speed to 533MHz, it only endows the latest versions of
the Mobile Celeron with the older quad-pumped 400MHz rating. Most Mobile Celerons
still have 100MHz front-side buses.
Much like the desktop line, the first Mobile Celeron was a modest alteration of the
Pentium II. Introduced on January 25, 1999, it used the same core logic but with a different cache design, a 128KB cache on the chip substrate operating at full core speed.
Otherwise, the chip followed the Pentium II design and followed it up in speed, from
266MHz up to 466MHz, using the same 66MHz front-side bus as the desktop chips and
the same 0.25-micro design rules. It differs chiefly by operating at a lower, power-saving
voltage, 1.6 volts.
On February 14, 2000, Intel revised the Mobile Celeron design to take advantage of the
Pentium III core and its 0.18-micro design rules. The newer Mobile Celerons gained two
advantages: the higher, 100MHz front-side bus speed and the Streaming SIMD
Extensions to the instruction set. In addition to a faster chip (at 500MHz), Intel also
introduced a 450MHz chip, slower than the quickest of the old Mobile Celeron design
but able to take advantage of the higher bus speed. Intel continued to upgrade this core
up to a speed of 933MHz, introduced on October 1, 2001.
When Intel move the Mobile Celeron line to 0.13-micron technology, the company cut
the core voltage of the chips down to 1.45 volts while pushing up its top clock speed to
1.2GHz. The smaller design rules left more space on the chips silicon, which Intel utilized for an enhanced secondary cache, pushing it to 256KB.
On June 24, 2002, Intel switched the Mobile Celeron core once again, bringing in a
design derived from the Pentium 4. The new core allowed Intel to trim the chips operating voltage once again, down to 1.3 volts. In addition, new versions of the Mobile Celeron
boast the quad-pumped 400MHz front-side bus speed of the Pentium 4 as well as its
enhanced Streaming SIMD Extensions 2 instruction set.
Table 5.12 summarizes the life history of Intels Mobile Celeron product line.
Current Products
TA B L E 5 . 1 2
Summary of Intel Mobile Celeron Features
Speed
Introduction Date
Cache
Front-Side
Core
Bus Speed Design Rules Voltage
Instruction
Set
266MHz
January 25, 1999
128KB
66MHz
0.25 micron
1.6 volts
MMX
300MHz
January 25, 1999
128KB
66MHz
0.25 micron
1.6 volts
MMX
333MHz
April 5, 1999
128KB
66MHz
0.25 micron
1.6 volts
MMX
366MHz
May 17, 1999
128KB
66MHz
0.25 micron
1.6 volts
MMX
400MHz
June 14, 1999
128KB
66MHz
0.18 micron
1.6 volts
MMX
433MHz
September 15, 1999 128KB
66MHz
0.25 micron
1.9 volts
MMX
450MHz
February 14, 2000
100MHz
0.18 micron
1.6 volts
SSE
466MHz
66MHz
0.25 micron
1.9 volts
MMX
500MHz
February 14, 2000
128KB
100MHz
0.18 micron
1.6 volts
SSE
550MHz
April 24, 2000
128KB
100MHz
0.18 micron
1.6 volts
SSE
600MHz
June 19, 2000
128KB
100MHz
0.18 micron
1.6 volts
SSE
650MHz
June 19, 2000
128KB
100MHz
0.18 micron
1.6 volts
SSE
700MHz
100MHz
0.18 micron
1.6 volts
SSE
733MHz
October 1, 2001
128KB
100MHz
0.18 micron
1.7 volts
SSE
750MHz
March 19, 2001
128KB
100MHz
0.18 micron
1.6 volts
SSE
800MHz
May 21, 2001
128KB
100MHz
0.18 micron
1.6 volts
SSE
800AMHz
October 1, 2001
128KB
100MHz
0.18 micron
1.7 volts
SSE
850MHz
July 2, 2001
128KB
100MHz
0.18 micron
1.6 volts
SSE
866MHz
October 1, 2001
128KB
100MHz
0.18 micron
1.7 volts
SSE
900MHz
October 1, 2001
128KB
100MHz
0.18 micron
1.7 volts
SSE
933MHz
October 1, 2001
128KB
100MHz
0.18 micron
1.7 volts
SSE
1000MHz
April 17, 2002
256KB
133MHz
0.13 micron
1.4 volts
SSE
1060MHz
January 21, 2002
256KB
133MHz
0.13 micron
1.45 volts SSE
1130MHz
January 21, 2002
256KB
133MHz
0.13 micron
1.45 volts SSE
1200MHz
January 21, 2002
256KB
133MHz
0.13 micron
1.45 volts SSE
1330MHz
June 24, 2002
256KB
133MHz
0.13 micron
1.5 volts
SSE
1400MHz
June 24, 2002
256KB
400MHz
0.13 micron
1.3 volts
SSE2
1500MHz
June 24, 2002
256KB
400MHz
0.13 micron
1.3 volts
SSE2
128KB
139
140
Mobile Pentium III

For about a year, Intels most powerful mobile chips wore the Pentium III designation. As
the name implies, they were derived from the desktop series with several features added to
optimize them for mobile applications and, incidentally, to bring the number of transistors
on their single slice of silicon to 28 million. Table 5.13 summarizes the Mobile Pentium III
product line.
TA B L E 5 . 1 3
Summary of Intel Mobile Pentium III Features
Core Speed Introduction Date
Cache
Size
Front-Side
Operating Instruction
Set
400MHz
October 25, 1999
256KB
100MHz
0.18 micron
1.35 volts
SSE
450MHz
October 25, 1999
256KB
100MHz
0.18 micron
1.6 volts
SSE
500MHz
October 25, 1999
256KB
100MHz
0.18 micron
1.6 volts
SSE
600MHz
January 18, 2000
256KB
100MHz
0.18 micron
1.6 volts
SSE
650MHz
January 18, 2000
256KB
100MHz
0.18 micron
1.6 volts
SSE
700MHz
April 24, 2000
256KB
100MHz
0.18 micron
1.6 volts
SSE
750MHz
June 19, 2000
256KB
100MHz
0.18 micron
1.35 volts
SSE
800MHz
100MHz
0.18 micron
1.35 volts
SSE
850MHz
100MHz
0.18 micron
1.35 volts
SSE
900MHz
March 19, 2001
256KB
100MHz
0.18 micron
1.35 volts
SSE
1.0GHz
March 19, 2001
256KB
100MHz
0.18 micron
1.35 volts
SSE
Mobile Pentium III-M

When Intel shifted to new fabrication, the company altered the core logic design of the Pentium
III. Although the basic design remained the same, the tighter design rules allowed for higherspeed operation. In addition, Intel improved the power management of the Mobile Pentium III
with Enhanced SpeedStep technology, which allows the chip to shift down in speed in increments to conserve power. Table 5.14 summarizes the Mobile Pentium III-M lineup.
TA B L E 5 . 1 4
Summary of Intel Mobile Pentium III-M Features
Core Speed
Introduction
Date
Cache
Size
Front-Side
Bus Speed
Design Rules Voltage
Set
866MHz
July 30, 2001
512KB
133MHz
0.13 micron
1.15 volts
SSE
933MHz
July 30, 2001
512KB
133MHz
0.13 micron
1.15 volts
SSE
1000MHz
July 30, 2001
512KB
133MHz
0.13 micron
1.4 volts
SSE
Current Products
Core Speed
Introduction
Date
Cache
Size
Front-Side
Bus Speed
Design Rules Voltage
Set
1060MHz
July 30, 2001
512KB
133MHz
0.13 micron
1.4 volts
SSE
1130MHz
July 30, 2001
512KB
133MHz
0.13 micron
1.4 volts
SSE
1200MHz
October 1, 2001
512KB
133MHz
0.13 micron
1.4 volts
SSE
Mobile Pentium 4-M

Intels highest performance mobile chip is the Mobile Pentium 4-M. Based on the same core
logic as the line-leading Pentium 4, the mobile chip is enhanced with additional power-management features and lower-voltage operation. Table 5.15 summarizes the Pentium 4-M lineup.
TA B L E 5 . 1 5
Summary of Intel Mobile Pentium 4-M Features
Introduction
Core Speed Date
Cache
Size
Front-Side
Operating
Instruction
Set
1400MHz
April 23, 2002
512KB 400MHz
0.13 micron
1.3 volts
SSE2
1500MHz
April 23, 2002
512KB 400MHz
0.13 micron
1.3 volts
SSE2
1600MHz
March 4, 2002
512KB 400MHz
0.13 micron
1.3 volts
SSE2
1700MHz
March 4, 2002
512KB 400MHz
0.13 micron
1.3 volts
SSE2
1800MHz
April 23, 2002
512KB 400MHz
0.13 micron
1.3 volts
SSE2
1900MHz
June 24, 2002
512KB 400MHz
0.13 micron
1.3 volts
SSE2
2000MHz
June 24, 2002
512KB 400MHz
0.13 micron
1.3 volts
SSE2
2200MHz
September 26, 2002 512KB 400MHz
0.13 micron
1.3 volts
SSE2
Low-Voltage Microprocessors
For use in computers with strict power budgetseither because the manufacturer decided
to devote little space to batteries or because the maker opted for extremely long
runtimesIntel has developed several lines of low-voltage microprocessors. These mobile
chips have operating voltages substantially lower than the mainstream chips. Such lowvoltage chips have been produced in three major mobile processor lines: the Mobile
Celeron, the Mobile Pentium III, and the Mobile Pentium III-M.
Ultra-Low-Voltage Microprocessors
For systems in which power consumption is absolutely critical, Intel has offered versions
of its various mobile chips designed for ultra-low-voltage operation. Ultra, like beauty, is
in the eye of the beholderthese chips often operate at voltages only a fraction lower
than ordinary low-voltage chips. The lower operating voltage limits the top speed of these
141
142
chips; they are substantially slower than the ordinary low-voltage chips at the time of their
introduction. But again, they are meant for systems where long battery life is more important than performance. Table 5.16 summarizes Intels ultra-low-voltage microprocessors.
TA B L E 5 . 1 6
Summary of Intel Low-Voltage Mobile Microprocessor Features
Core
Designation Speed
Introduction
Date
Cache
Size
Front-Side
Bus Speed
Design
Rules
Voltage
Set
Mobile
Celeron
500MHz
January 30,
2001
128KB
100MHz
0.18
micron
1.1 volts
MMX
Mobile
Celeron
600MHz
May 21,
2001
128KB
100MHz
0.18
micron
1.1 volts
MMX
Mobile
Celeron
650MHz
January 21,
2002
256KB
133MHz
0.13
micron
1.1 volts
SSE
Mobile
Pentium III
500MHz
January 30,
2001
256KB
100MHz
0.18
micron
1.1 volts
SSE
Mobile
Pentium III
600MHz
May 21,
2001
256KB
100MHz
0.18
micron
1.1 volts
SSE
Mobile
Pentium III
750MHz
January 21,
2002
512KB
133MHz
0.13
micron
1.1 volts
SSE
Mobile
Pentium III
512KB
700MHz
November 13, 512KB

2001
100MHz
0.13
micron
1.1 volts
SSE
Mobile
Pentium
III-M
800MHz
April 17,
2002
512KB
133MHz
0.13
micron
1.15 volts
SSE
Mobile
Pentium
III-M
800MHz
April 17,
2002
512KB
100MHz
0.13
micron
1.15 volts
SSE
Itanium
Intels Itanium microprocessor line marks an extreme shift for Intel, entirely breaking with
the Intel Architecture of the pastwhich means that the Itanium cannot run programs or
operating systems designed for other Intel chips. Instead of using the old Intel design dating back to the 4004, the Itanium introduces Intels Explicitly Parallel Instruction
Computing architecture, which is based on the Precision Architecture originally developed by Hewlett-Packard Corporation for its line of RISC chips. In short, that means
everything you know about Intel processors doesnt apply to the Itanium, especially the
performance you should expect from the megahertz ratings of the chips. Itanium chips
look slow on paper but perform fast in computers.
Current Products
The original Itanium was code-named Merced and was introduced in mid-2001, with an
announcement from Intel on May 29, 2001, that systems soon would be shipping. The
original Itanium was sold in two speeds: 733MHz and 800MHz. Equipped with a
266MHz system bus (double-clocked 133MHz), the Itanium further enhanced performance with a three-level on-chip cache design with 32KB in its primary cache, 96KB in
its secondary cache, and 2MB or 4MB in its tertiary cache (depending on chip model). All
aspects of the chip feature a full 64-bit bus width, both data and address lines. Meant for
high-performance servers, the Itanium allows for up to 512 processors in a single computer.
Introduced on July 8, 2002, the Itanium 2 (code-named McKinley) pushed up the clock
speed of the same basic design as the original Itanium by shifting down the design rules to
0.13 micron. Intel increased the secondary cache of the Itanium 2 to 256KB and allowed
for tertiary caches of 1.5MB or 3MB. The refined design also increased the system bus
speed to 400MHz (actually, a double-clocked 200MHz bus) with a bus width of 128 bits.
Initially, Itanium 2 chips were offered at speeds of 900MHz and 1.0GHz. Table 5.17 summarizes the Itanium line.
TA B L E 5 . 1 7
Summary of Intel Itanium Product Line
Primary Secondary Tertiary System

Clock Speed Cache Cache
Cache
Bus
Design
Rules
Model
Introduction
Itanium
May 29, 2001 733MHz
32KB
96KB
2MB
266MHz
0.18 micron
Itanium
May 29, 2001 733MHz
32KB
96KB
4MB
266MHz
0.18 micron
Itanium
May 29, 2001 800MHz
32KB
96KB
2MB
266MHz
0.18 micron
Itanium
May 29, 2001 800MHz
32KB
96KB
4MB
266MHz
0.18 micron
Itanium 2 July 8, 2002
900MHz
32KB
256KB
1.5MB
400MHz
0.13 micron
900MHz
32KB
256KB
3MB
400MHz
0.13 micron
1000MHz
32KB
256KB
1.5MB
400MHz
0.13 micron
1000MHz
32KB
256KB
3MB
400MHz
0.13 micron
Advanced Micro Devices Microprocessors

Advanced Micro Devices currently fields two lines of microprocessor. Duron chips correspond to Intels Celeron line, targeting the budget-minded consumer. Athlon chips are
mainstream, full-performance microprocessors. Around the end of the year 2002, AMD
will add the Opteron name to its product line. Meant to compete with the performance of
Intels Itanium, the Opteron will differ with a design meant to run todays software as well
or better than current Intel processors. Opteron will become the new top-end of the
AMD lineup.
143
144
Athlon
AMDs answer to the Pentium III and its P6 core was the Athlon. The Athlon is built on a
RISC core with three integer pipelines (compared to the two inside the Pentium III),
three floating-point pipelines (versus one in the Pentium III), and three instruction
decoders (compared one in the Pentium III). The design permits the Athlon to achieve up
to nine operations per clock cycle, compared to five for the Pentium III. AMD designed
the floating-point units specifically for multimedia and endowed them with both the
MMX (under Intel license) and 3DNow! instruction sets.
Program code being what it isnot very amendable to superscalar processingthe
Athlons advantage proved more modest in reality. Most people gave the Athlon a slight
edge on the Pentium III, megahertz for megahertz. The Athlon chip was more than powerful enough to challenge Intel for leadership in processing power. For more than a year,
AMD and Intel ran a speed race for the fastest processor, with AMD occasionally edging
ahead even in pure megahertz.
The Athlon has several other features that help to boost its performance. It has both primary (L1) and secondary (L2) caches on-chip, operating at the full speed rating of the
core logic. A full 128KB is devoted to the primary cache, half for instructions and half for
data, and 256KB is devoted to the secondary cache, for a total (the figure that AMD usually quotes) of 384KB of cache. The secondary cache connects to the chip through a 64bit back-side bus operating at the core speed of the chip.
The system bus of the Athlon also edged past that used by Intel for the Pentium III. At
introduction, the Athlon allowed for a 200MHz system bus (and 133MHz memory).
Later, in March, 2001, the system bus interface was bumped up to 266MHz. This bus
operates asynchronously with the core logic, so AMD never bothered with some of the
odd speeds Intel used for its chips. The instruction set of the Athlon includes an enhanced
form of AMDs 3DNow! The Athlon recognizes 45 3D instructions, compared to 21 for
AMDs previous-generation K6-III chip.
The design of the Athlon requires more than 22 million transistors. As with other chips in
its generation, it has registers 32 bits wide but connects to its primary cache through a
128-bit bus and to the system through a 64-bit data bus. It can directly address up to 8TB
of memory through an address bus thats effectively 43-bits wide.
AMD has introduced several variations on the Athlon namethe basic Athlon, the Athlon
4 (to parallel the introduction of the Pentium 4), and the Athlon XP (paralleling
Microsofts introduction of Windows XP). The difference between the Athlon and Athlon
4 are in name alone. The basic core of all these Athlons is the same. Only the speed rating
has increased with time. With the XP designation, however, AMD added Intels Streaming
SIMD Extensions to the instruction set of the chip, giving it better multimedia performance.
Current Products
The Athlon comes in cartridge form and slides into AMDs Slot A. Based on the EV6 bus
design developed by Digital Equipment Corporation (now part of Compaq) for the Alpha
chip (a microprocessor originally meant for minicomputers but now being phased out),
the new socket is physically the same as Intels Slot 1, but the signals are different and the
AMD chip is incompatible with slots for Intel processors.
AMD fabricated its initial Athlon chips using 0.25-micron design rules. In November,
1999, the company shifted to new fabrication facilities that enabled it to build the Athlon
with 0.18-micron design rules.
Table 5.18 summarizes the features of the AMD Athlon line.
TA B L E 5 . 1 8
Summary of AMD Athlon Microprocessor Features
Processor
Model
orMHz
Introduction
Date
Front-Side
Bus Speed
Cache Size
Design Rules
Athlon XP
2100+
March 13, 2002
266MHz
128KB/256KB
0.18 micron
Athlon XP
2000+
January 7, 2002
266MHz
128KB/256KB
0.18 micron
Athlon XP
1900+
November 5, 2001
266MHz
128KB/256KB
0.18 micron
Athlon XP
1800+
October 9, 2001
266MHz
128KB/256KB
0.18 micron
Athlon XP
1700+
October 9, 2001
266MHz
128KB/256KB
0.18 micron
Athlon XP
1600+
October 9, 2001
266MHz
128KB/256KB
0.18 micron
Athlon XP
1500+
October 9, 2001
266MHz
128KB/256KB
0.18 micron
Athlon
1400
June 6, 2001
266MHz
128KB/256KB
0.18 micron
Athlon
1330
March 21, 2001
266MHz
128KB/256KB
0.18 micron
Athlon
1300
March 21, 2001
266MHz
128KB/256KB
0.18 micron
Athlon
1200
October 17, 2000
200MHz
128KB/256KB
0.18 micron
Athlon
1100
August 14, 2000
200MHz
128KB/256KB
0.18 micron
Athlon
1000
March 6, 2000
200MHz
128KB/256KB
0.18 micron
Athlon
950
March 6, 2000
200MHz
128KB/256KB
0.18 micron
Athlon
900
March 6, 2000
200MHz
128KB/256KB
0.18 micron
Athlon
850
February 11, 2000
200MHz
128KB/256KB
0.18 micron
Athlon
800
January 6, 2000
200MHz
128KB/256KB
0.18 micron
Athlon
750
November 29, 1999
200MHz
128KB/256KB
0.18 micron
Athlon
700
October 4, 1999
200MHz
128KB/256KB
0.25 micron
continues
145
146
TA B L E 5 . 1 8
Continued
Processor
Model
orMHz
Introduction
Date
Front-Side
Bus Speed
Cache Size
Design Rules
Athlon
650
August 9, 1999
200MHz
128KB/256KB
0.25 micron
Athlon
600
June 23, 1999
200MHz
128KB/256KB
0.25 micron
Athlon
550
June 23, 1999
200MHz
128KB/256KB
0.25 micron
Athlon
500
June 23, 1999
200MHz
128KB/256KB
0.25 micron
Athlon MP
For multiprocessor applications, AMD adapted the core logic of the Athlon chip with bus
control circuitry meant for high-bandwidth transfers. These chips are specifically aimed at
servers rather than desktop computers. Table 5.19 summarizes the AMD offerings.
TA B L E 5 . 1 9
Summary of AMD Athlon MP Features
Processor
Introduction Date
FSB Speed
Technology
Athlon MP 2000+
March 13, 2002
266MHz
0.18 micron
Athlon MP 1900+
December 12, 2001
266MHz
0.18 micron
Athlon MP 1800+
October 15, 2001
266MHz
0.18 micron
Athlon MP 1600+
October 15, 2001
266MHz
0.18 micron
Athlon MP 1500+
October 15, 2001
266MHz
0.18 micron
Duron
To take on Intels budget-priced Celeron chips, AMD slimmed down the Athlon to make
a lower-priced product. Although based on the same logic core as the Athlon, the Duron
skimps on cache. Although it retains the same 128KB primary cache, split with half handling data and half instructions, the secondary cache is cut to 64KB. As with the Athlon,
however, both caches operate at full core logic speed. The smaller secondary cache
reduces the size of the silicon die required to make the chip, allowing more Durons than
Athlons to be fabricated from each silicon wafer, thus cutting manufacturing cost.
The basic architecture of the Duron core matches the Athlon with three integer pipelines,
three floating-point pipelines (which also process both 3DNow! and Intels MMX instruction sets), and three instruction/address decoders. Duron chips even share the same
0.18-micron technology used by the higher-priced Athlon. For now, however, Durons are
Current Products
restricted to lower speeds than the Athlon line and have not benefited from AMDs
higher-speed 266MHz system bus. All Durons use a 200MHz bus.
During development AMD used the code name Spitfire for the Duron. The company
explains the official name of the chip as derived from the Latin root durare, meaning to
last and on, meaning unit. The root is the same as the English word durability.
Table 5.20 summarizes the characteristics of the AMD Duron line.
TA B L E 5 . 2 0
Summary of AMD Duron Microprocessor Features
Processor
Speed
(MHz) Introduction
Front-Side
Bus Speed
Cache Size
(L1/L2)
Technology
Duron
1300
January 21, 2002
200MHz
128KB/64KB
0.18 micron
Duron
1200
November 15, 2001
200MHz
128KB/64KB
0.18 micron
Duron
1100
October 1, 2001
200MHz
128KB/64KB
0.18 micron
Duron
950
June 6, 2001
200MHz
128KB/64KB
0.18 micron
Duron
900
April 2, 2001
200MHz
128KB/64KB
0.18 micron
Duron
850
January 8, 2001
200MHz
128KB/64KB
0.18 micron
Duron
800
October 17, 2000
200MHz
128KB/64KB
0.18 micron
Duron
750
September 5, 2000
200MHz
128KB/64KB
0.18 micron
Duron
700
June 19, 2000
200MHz
128KB/64KB
0.18 micron
Duron
650
June 19, 2000
200MHz
128KB/64KB
0.18 micron
Duron
600
June 19, 2000
200MHz
128KB/64KB
0.18 micron
Mobile Processors
As with its desktop processors, AMD has two lines of chips for portable computers, the
Athlon and Duron, for the high and low ends of the market, respectively. Unlike Intel,
AMD puts essentially the same processors as used on the desktop in mobile packages. The
AMD chips operate at the same low voltages as chips specifically designed for mobile
applications, and AMDs desktop (and therefore, mobile) products all use its power-saving
PowerNow! technology.
The one difference: AMD shifted to 0.13-micron technology for its portable Athlon XP
while the desktop chip stuck with 0.18-micron technology. Table 5.21 summarizes AMDs
Mobile Athlon product line.
147
148
TA B L E 5 . 2 1
Summary of AMD Mobile Athlon Microprocessor Features
Processor
Model
orMHz Introduction Date
Cache Size
FSB Speed (L1/L2)
Technology
Mobile Athlon XP
1800+
July 15, 2002
266MHz
128KB/256KB
0.13 micron
Mobile Athlon XP
1700+
April 17, 2002
266MHz
128KB/256KB
0.13 micron
Mobile Athlon XP
1600+
April 17, 2002
266MHz
128KB/256KB
0.13 micron
Mobile Athlon XP
1500+
April 17, 2002
266MHz
128KB/256KB
0.13 micron
Mobile Athlon XP
1400+
April 17, 2002
266MHz
128KB/256KB
0.13 micron
Mobile Athlon 4
1600+
March 13, 2002
266MHz
128KB/256KB
0.18 micron
Mobile Athlon 4
1500+
January 28, 2002
266MHz
128KB/256KB
0.18 micron
Mobile Athlon 4
1200
November 12, 2001 200MHz
128KB/256KB
0.18 micron
Mobile Athlon 4
1100
August 20, 2001
200MHz
128KB/256KB
0.18 micron
Mobile Athlon 4
1000
May 14, 2001
200MHz
128KB/256KB
0.18 micron
Mobile Athlon 4
950
May 14, 2001
200MHz
128KB/256KB
0.18 micron
Mobile Athlon 4
900
May 14, 2001
200MHz
128KB/256KB
0.18 micron
Mobile Athlon 4
850
May 14, 2001
200MHz
128KB/256KB
0.18 micron
As with desktop chips, the chief difference between AMDs Mobile Athlon and Mobile
Duron is the size of the secondary cacheonly 64KB in the Duron chips. Table 5.22
summarizes the Mobile Duron line.
TA B L E 5 . 2 2
Summary of AMD Mobile Duron Microprocessor Features
Processor
MHz
Introduction Date
FSB Speed
Cache Size
(L1/L2)
Technology
Mobile Duron
1200
January 30, 2002
200MHz
128KB/64KB
0.18 micron
Mobile Duron
1100
January 30, 2002
200MHz
128KB/64KB
0.18 micron
Mobile Duron
1000
December 17, 2001
200MHz
128KB/64KB
0.18 micron
Mobile Duron
950
November 12, 2001
200MHz
128KB/64KB
0.18 micron
Mobile Duron
850
May 14, 2001
200MHz
128KB/64KB
0.18 micron
Mobile Duron
800
May 14, 2001
200MHz
128KB/64KB
0.18 micron
Mobile Duron
700
January 15, 2001
200MHz
128KB/64KB
0.18 micron
Mobile Duron
600
January 15, 2001
200MHz
128KB/64KB
0.18 micron
Current Products
Opteron
AMD chose the name Opteron for what it calls its eighth generation of microprocessors,
for which it has used the code-name Hammer during development. The Opteron represents the first 64-bit implementation of Intel architecture, something Intel has neglected
to develop.
The Opteron design extends the registers of Pentium-style computers to a full 64-bits
wide. Its a forthright extension of the current Intel architecture, and AMD makes the
transition the same way Intel extended the original 16-bit bus of the 8086-style chips to
32 bits for the 386 series. The new, wide registers are a superset of the 32-bit registers. In
the Opterons compatibility mode, 16-bit instructions simply use the least significant 16
bits of the wide registers, 32-bit instructions use the least significant 32 bits, and 64-bit
instructions use the entire register width. As a result, the Opteron can run any Intel code
at any time without the need for emulators or coprocessors. Taking advantage of the full
64-bit power of the Opteron will, of course, require new programs written with the new
64-bit instructions.
The Opteron design also changes the structure of the processor core, rearranging the
pipelines and processing units. The Opteron design uses three separate decode pipelines
that feed a packing stage that links all three pipelines to more efficiently divide operations
between them. The pipelines then feed into another stage of decoding, then eight stages
of scheduling. At that point, the pipelines route integer and floating-point operations to
individual processors. AMD quotes a total pipeline of 12 stages for integers and 17 for
floating-point operations. The floating-point unit understands everything from MMX
through 3DNow! to Intels latest SSE2.
As important as the core logic is, AMD has made vast improvements on the I/O of the
Opteron. Major changes come in two areas. AMD builds the memory controller into the
Opteron, so the chip requires no separate memory control hub. The interface uses DDR
memory through two 128-bit-wide channels. Each channel can handle four memory modules, initially those rated for PC1600 operation, although the Opteron design allows for
memory as fast as PC2700. According to AMD, building the memory interface into the
Opteron reduces latency (waiting time), the advantage of which is an increase with every
step up in clock speed.
The system bus of the Opteron uses the HyperTranport interface with a 16-bit channel
operating at an effective data rate of 1.6Gbps, giving the chip a peak I/O bandwidth of
3.2Gbps.
Transmeta Microprocessors
Strictly speaking, the Crusoe processors from Transmeta Corporation are not Intel
architecture chips. They use an entirely different instruction set from Intel chips, and by
149
150
themselves could run a Windows program on a dare. Transmetas not-so-secret weapon is

what it calls Code Morphing software, a program that runs on the Crusoe chip and translates
Intels instruction set into its own. In effect, the Crusoe chip is the core logic of a modern
Intel Architecture stripped of its hardware code translation.
The core is a very long instruction word processor, one that uses instructions that can be
either 64 or 128 bits long. The core has two pipelinesan integer pipeline with seven
stages and a floating-point pipeline with 10. Transmeta keeps the control logic for the
core logic simple. It does not allow out-of-order execution, and instruction scheduling is
handled by software.
Transmeta provides both a 64KB primary instruction cache and a 64KB primary data
cache. The Crusoe comes with either of two sizes of secondary cache. The TMS5500 uses
a 256KB secondary cache, and the TMS5800 has a 512KB secondary cache. At the time
this was written, the chips were available with speed ratings of 667, 700, 733, 800, 867,
and 900MHz.
To help the chip mimic Intel processors, the Crusoe family has a translation look-aside
buffer that uses the same protection bits and address-mapping as Intel processors. The
Crusoe hardware generates the same condition codes as Intel chips, and their floatingpoint units use the same 80-bit format as Intels basic FPU design (but not the 128-bit
registers used by SSE2 instructions).
The result of this design is a very compact microprocessor that does what it does very
quickly while using very little power. Transmeta has concentrated its marketing on the low
power needs of the Crusoe chips, and they are used almost exclusively in portable computers. A less charitable way of looking at the Crusoe is that its smaller silicon needs make
for a chip thats far less expensive to manufacturer and easier to design. Thats not quite
fair because developing the Code Morphing software is as expensive as designing silicon
logic. Moreover, the current Crusoe chips take advantage of small silicon needs of their
small logic cores, adding more features onto the same die. The current Crusoe models
include the north bridge circuits of a conventional chipset on the same silicon as the core
logic. The Crusoe chip includes the system bus, memory, and PCI bus interfaces, making
portable computer designs potentially more compact. Current Crusoe versions support
both SDR and DDR memory with system bus speeds up to 133MHz.
Another way to look at Code Morphing is to consider it as a software emulator, a program
that runs on a chip to mimic another. Emulators are often used at the system level to
allow programs meant for one computer to run on another. The chief distinctions
between Code Morphing and traditional emulation is that Code Morphing works at the
chip level, and the Crusoe chip keeps the necessary translation routines in firmware stored
in read-only memory (ROM) chips.
Current Products
According to Transmeta, Code Morphing also helps the Crusoe chip to be faster, enabling
it to keep up with modern superscalar chips. The Code Morphing software doesnt translate each Intel instruction on the fly. Instead, it translates a series of instructions, potentially even full subroutines. It retains the results as if in a cache so that if it encounters the
same set of Intel instructions again, it can look up the code to use rather than translating
it again. The effect doesnt become apparent, according to Transmeta, until the Intelbased routine has been executed several times. The tasks typically involved with running a
modern computerthe Windows graphic routines, software drivers, and so onshould
benefit greatly from this technology. In reality, Crusoe processors dont test well, but they
deliver adequate performance for the sub-notebook computers that are their primary
application.
151
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Chipsets
Buy a microprocessor, and its just an inert lump that sits in its static-free
box, super-powered silicon with extreme potential but no motivation. Its
sort of like a Hollywood hunk proto-star before he takes acting classes.
Even out of the box, the most powerful microprocessor is hardly impressiveor effective. It cant do anything on its own. Although it serves as
the centerpiece of a computer, it is not the computer itself. It needs help.
The electronics inside a personal computer that flesh out its circuitry
and make it work are called support circuits. At one time, support circuits
made up everything in a computer except for the microprocessor. Today,
most of the support functions needed to make a computer work take the
form of a few large integrated circuits called a chipset. Support circuits
handle only a few low-level functions to take care of the needs of the
chipset.
In current designs, the chipset is second in importance only to the
microprocessor in making your computer workand making your computer a computer. Your computers chipset is the foundation of its motherboard, and a computer-makers choice of chipsets determines the
overall capabilities of your computer. The microprocessor embodies its
potential. The chipset makes it a reality.
Although the chipset used by a computer is rarely mentioned in advertisements and specification sheets, its influence is pervasive. Among
other issues, the chipset determines the type and number of microprocessors that can be installed in the system; the maximum memory a
system can address; the memory technology used by the system (SDR,
DDR, or Rambus); whether the system can detect memory errors; the
154
speed of the memory bus; the size and kind of secondary memory cache; the type, speed,
and operating mode of the expansion bus; whether you can plug a video board into an
Accelerated Graphics Port in your computer; the mass-storage interfaces available on the
motherboard and the speed at which they operate; the ports available on your motherboard; and whether your computer has a built-in network adapter. Thats a lot of power
packed into a few chipsand a good reason for wondering what you have in your computer.
Background
The functions needed to make a microprocessor into a computer reads like a laundry list
for a three-ring circuslong and exotic, with a trace of the unexpected. A fully operational computer requires a clock or oscillator to generate the signals that lock the circuits
together; a memory mastermind that ensures each byte goes to the proper place and stays
there; a traffic cop for the expansion bus and other interconnecting circuitry to control
the flow of data in and out of the chip and the rest of the system; office-assistant circuits
for taking care of the daily routines and lifting the responsibility for minor chores from
the microprocessor; and, in most modern computers, communications, video, and audio
circuitry that let you and the computer deal with one another like rational beings.
Every computer manufactured since the first amphibians crawled from the primeval
swamp has required these functions, although in the earliest computers they did not take
the form of a chipset. The micro-miniaturization technology was nascent, and the need
for combining so many functions in so small a package was negligible. Instead, computer
engineers built the required functions from a variety of discrete circuitssmall, generalpurpose integrated circuits such as logic gatesand a few functional blocks, which are
larger integrated circuits designed to a particular purpose but not necessarily one that had
anything to do with computers. These garden-variety circuits, together termed support
chips, were combined to build all the necessary computer functions into the first computer.
The modern chipset not only combines these support functions onto a single silicon slice
but also adds features beyond the dreams of the first computers designersthings such as
USB ports, surround-sound systems, and power-management circuits. This all-in-one
design makes for simpler and far cheaper computers.
Trivial as it seems, encapsulating a piece of silicon in an epoxy plastic package is one of the
more expensive parts of making chips. For chipmakers, its much more expensive than
adding more functions to the silicon itself. The more functions the chipmaker puts on a
single chip, the less each function costs. Those savings dribble down to computer-makers
and, eventually, you. Moreover, designing a computer motherboard with discrete support
circuitry was a true engineering challenge because it required a deep understanding of the
electronic function of all the elements of a computer. Using a chipset, a computer engineer need only be concerned with the signals going in and out of a few components. The
CHAPTER 6 CHIPSETS
Functions
chipset might be a magical black box for all the designer cares. In fact, in many cases the
only skill required to design a computer from a chipset is the ability to navigate from a
roadmap. Most chipset manufacturers provide circuit designs for motherboards to aid in
the evaluation of their products. Many motherboard manufacturers (all too many, perhaps)
simply take the chipset-makers evaluation design and turn it into a commercial product.
You can trace the chipset genealogy all the way back to the first computer. This heritage
results from the simple need for compatibility. Todays computers mimic the function of
the earliest computersand IBMs original Personal Computer of 1981 in particularso
that they can all run the same software. Although seemingly anachronistic in an age when
those who can remember the first computer also harbor memories of Desotos and
dinosaurs, even the most current chipsets must precisely mimic the actions of early computers so that the oldest software will still operate properly in new computersprovided,
of course, all the other required support is also present. After all, some Neanderthal will
set switchboards glowing from I-95 to the Silicon Valley with threats of lawsuits and
aspersions about the parenthood of chipset designers when the DOS utilities he downloaded in 1982 wont run on his new Pentium 4.
Functions
No matter how simple or elaborate the chipset in a modern computer, it has three chief
functions. It must act as a system controller that holds together the entire computer, giving
all the support the microprocessor needs to be a true computer system. As a memory controller, it links the microprocessor to the memory system, establishes the main memory
and cache architectures, and ensures the reliability of the data stashed away in RAM chips.
And to extend the reach of the microprocessor to other system components, it must act as
a peripheral controller and operate input/output ports and disk interfaces.
How these functions get divided among silicon chips and the terminology describing the
chips has itself evolved.
When computers first took their current form, incorporating the PCI system for linking
components together, engineers lumped the system and memory functions together as a
unit they called the north bridge. Intel sometimes uses the term host bridge to indicate the
same function. The peripheral control functions make up the south bridge, sometimes
called the I/O bridge. The circuits that controlled the expansion bus often were grouped
together with the north bridge or as a separate chip termed the PCI bridge.
Intel has reworked its design and nomenclature to make the main liaison with the microprocessor into the memory controller hub. This chip includes the essential support functions
of the north bridge. It provides a wide, fast data pathway to memory and to the systems
video circuitry.
155
156
The I/O controller hub takes over all the other interface functions of the chipset, including
control of the expansion bus, all disks, all ports, and sound and power management. The
host controller provides a high-speed connection between the I/O controller hub and the
microprocessor.
The memory control hub links to the microprocessor through its system bus, sometimes
termed the front side bus, which provides the highest speed linkup available to peripheral
circuits. The logic inside the memory control hub translates the microprocessors address
requests, pumped down the system bus as address cycles, into the form used by the memory in the system. Because memory modules differ in their bus connections, the memory
hub controller circuitry determines what kind of memory can connect to the system.
Some memory hub controllers, for example, link only to Rambus memory, whereas others
may use double data-rate (DDR) modules.
Although the connection with graphics circuitry naturally is a part of the input/output system of a computer, the modern chipset design puts the link in the memory hub controller.
Video displays involve moving around large blocks of memory, and the memory host controller provides the highest-speed connection with the microprocessor, speeding video
access and screen updates. In addition, the design of the current video interconnection
standard, accelerated graphics port (AGP), uses a memory aperture to transfer bytes
between the microprocessor and the frame buffer (the memory that stores the image on
the display screen). Accessing this aperture is easiest and most direct through the memory
controller.
The I/O hub controller generates all the signals used for controlling the microprocessor
and the operation of the computer. It doesnt tell the microprocessor what instructions to
execute. Rather, it sets the speed of the system by governing the various clocks in the
computer, it manages the microprocessors power usage (as well as that of the rest of the
computer), and it tells the microprocessor when to interrupt one task and switch to
another or give a piece of hardware immediate attention. These are all system-control
functions, once the province of a dedicated system controller chip.
In addition, the I/O hub controller provides both the internal and external links to the
rest of the computer and its peripherals, its true input/output function. It controls the
computers expansion bus (which not only generates the signals for the expansion slots for
add-in cards but also provides the high-speed hark disk interface), all the ports (including
USB), and network connections. Audio circuitry is sometimes built in to the I/O hub controller, again linking through the expansion bus circuitry.
Memory Control
Memory speed is one of the most important factors in determining the speed of a computer. If a microprocessor cannot get the data and program code it needs, it has to wait.
CHAPTER 6 CHIPSETS
Memory Control
Using the right kind of memory and controlling it effectively are both issues governed by
the chipset.
Handling memory for todays chipsets is actually easier than a few years ago. When computer-makers first stated using memory caches to improve system performance, they
thrust the responsibility for managing the cache on the chipset. But with the current generation of microprocessors, cache management has been integrated with the microprocessor. The chipset only needs to provide the microprocessors access to main memory
through its system bus.
Addressing
To a microprocessor, accessing memory couldnt be simpler. It just needs to activate the
proper combination of address lines to indicate a storage location and then read its contents or write a new value. The address is a simple binary code that uses the available
address lines, essentially a 64-bit code with a Pentium-level microprocessor.
Unfortunately, the 64-bit code of the microprocessor is completely unintelligible to memory chips and modules. Semiconductor-makers design their memory chips so that they
will fit any application, regardless of the addressing of the microprocessor or even
whether they link to a microprocessor. Chips, with a few megabytes of storage at most,
have no need for gigabyte addressability. They need only a sufficient number of addresses
to put their entire contentsand nothing moreonline.
Translating between the addresses and storage format used by a microprocessor and the
format used by memory chips and modules is the job of the memory decoder. This chip (or,
in the case of chipsets, function) determines not only the logical arrangement of memory,
but also how much and what kind of memory a computer can use.
The memory functions of modern chipsets determine the basic timing of the memory system, controlling which signals in the memory system are active at each instant for any
given function. The ability to adjust these timing values determines the memory technology that a computer can use. For example, early chipsets timed their signals to match
Synchronous Dynamic Random Access Memory (SDRAM) memory chips. High-performance chipsets need to address Rambus memory and double-data rate memory, which use
not only a different addressing system but also an entirely different access technology.
In addition, the chipset determines how much memory a computer can possibly handle.
Although a current microprocessor such as the Pentium 4 can physically address up to
64GB of memory, modern chipsets do not support the full range of microprocessor
addresses. Most current Intel chipsets, for example, address only up to 2GB of physical
memory. Not that Intels engineers want to slight youthey have not purposely shortchanged you. The paltry memory capacity is a result of physical limits of the high-speed
157
158
memory buses. The distance that the signals can travel at todays memory speed is severely
limited, which constrains how many memory packages can fit near enough to the chipset.
Intel actually provides two channels on its Rambus chipsets, doubling the capacity that
otherwise would be available through a Rambus connection.
Refreshing
Most memory chips that use current technologies require periodic refreshing of their
storage. That is, the tiny charge they store tends to drain away. Refreshing recharges each
memory cell. Although memory chips handle the process of recharging themselves, they
need to coordinate the refresh operation with the rest of the computer. Your computer
cannot access the bits in storage while they are being refreshed.
The chipset is in charge of signaling memory when to refresh. Typically the chipset sends
a signal to your systems memory at a preset interval thats chosen by the computers
designer. With SDRAM memory and current chipsets, the refresh interval is usually 15.6
microseconds.
Memory can also self-refresh, triggering its own refresh when necessary. Self-refresh
becomes desirable when a system powers down to one of its sleep modes, during which
memory access is suspended along with microprocessor operation, and even the major
portions of the chipset switch off. Operating in this way, self-refresh preserves the contents of memory while reducing the power required for the chipset.
Error Handling
Computers, for which the integrity of their data is paramount, use memory systems with
error detection or error correction. The chipset manages the error-detection and errorcorrection processes.
Early personal computers relied on simple error detection, in which your computer
warned when an error changed the contents of memory (unfortunately warning usually
meant shutting down your computer). Current systems use an error-correction code (ECC),
extra information added to that sent into memory through which the change of a stored
bit can be precisely identified and repaired. The chipset adds the ECC to the data sent to
memory, an extra eight bits for every eight bytes of data stored. When the chipset later
reads back those eight bytes, it checks the ECC before passing the bytes on to your
microprocessor. If a single bit is in error, the ECC identifies it, and the chipset restores
the correct bit. If an error occurs in more than one bit, the offending bits cannot be
repaired, but the chipset can warn that they have changed.
ECC is a programmable feature of most chipsets. Depending on the design of your computer, it can be permanently set on or off at the factory, or you may be able to switch it on
and off through your systems advanced setup procedure. Simply switching ECC on
CHAPTER 6 CHIPSETS
System Control
doesnt automatically make your computers memory more secure. An ECC system
requires that your computer have special ECC memory modules installed.
System Control
The basic function of a chipset is to turn a chip into a computer, to add what a microprocessor needs to be an entire computer system. Not just any computer. You expect your
computer to match the prevailing industry standards so that it can run as wide a variety of
software as possible. You dont want surprises, such as your screen freezing when you try
to install an application recommended by a friend. No matter what program you buy, borrow, or steal, you want it to find all the features it needs in your computer.
To assure you this level of compatibility, nearly all the functions of the system controller
in a computer chipset are well defined and for the most part standardized. In nearly all
computers, the most basic of these functions use the same means of access and control
the ports and memory locations match those used by the very first computers. You might,
after all, want to load an old DOS program to balance your checkbook. Some people actually do it, dredging up programs that put roughshod white characters on a black screen,
flashing a white block cursor at them that demand they type a command. A few people
even attempt to balance their checkbooks.
Some features of the modern chipset exactly mimic the handful of discrete logic circuits
that were put together to make the very first personal computer, cobwebs and all.
Although many of these features are ignored by modern hardware and software, they continue as a vestigial part of all systems, sort of like an appendix.
Timing Circuits
Although anarchy has much to recommend should you believe in individual freedom or
sell firearms, its an anathema to computer circuits. Todays data-processing designs
depend on organization and controlled cooperationfactors that make timing critical.
The meaning of each pulse passing through a computer depends to a large degree on time
relationships. Signals must be passed between circuits at just the right moment for the
entire system to work properly.
Timing is critical in computers because of their clocked logic. The circuit element that
sends out timing pulses to keep everything synchronized is called a clock. Todays computers dont have just one clock, however. They have several, each for a specific purpose.
Among these are the system clock, the bus clock, the video clock, and even a real-time
clock.
The system clock is the conductor who beats the time that all the circuits follow, sending
out special timing pulses at precisely controlled intervals. The clock, however, must get its
159
160
cues from somewhere, either its own internal sense of timing or some kind of metronome.
Most clocks in electronic circuits derive their beats from oscillators.
Oscillators
An electronic circuit that accurately and continuously beats time is termed an oscillator.
Most oscillators work on a simple feedback principle. Like the microphone that picks up
its own sounds from public address speakers too near or turned up too high, the oscillator,
too, listens to what it says. As with the acoustic-feedback squeal that the public address
system complains with, the oscillator, too, generates it own howl. Because the feedback
circuit is much shorter, however, the signal need not travel as far and the resulting frequency is higher, perhaps by several thousand-fold.
The oscillator takes its output as its input, then amplifies the signal, sends it to its output,
where it goes back to the input again in an endless and out-of-control loop. The oscillator
is tamed by adding impediments to the feedback loop, through special electronic components added between the oscillators output and its input. The feedback and its frequency
can thus be brought under control.
In nearly all computers a carefully crafted crystal of quartz is used as this frequency-control element. Quartz is one of many piezoelectric compounds. Piezoelectric materials have
an interesting propertyif you bend a piezoelectric crystal, it generates a tiny voltage.
Whats more, if you apply a voltage to it in the right way, the piezoelectric material bends.
Quartz crystals do exactly that. But beyond this simple stimulus/response relationship,
quartz crystals offer another important property. By stringently controlling the size and
shape of a quartz crystal, it can be made to resonate at a specific frequency. The frequency
of this resonance is extremely stable and very reliableso much so that it can help an
electric watch keep time to within seconds a month. Although computers dont need the
absolute precision of a quartz watch to operate their logic circuits properly, the fundamental stability of the quartz oscillator guarantees that the computer operates at a clock frequency within the design limits always available to it.
The various clocks in a computer can operate separately, or they may be locked together.
Most computers link these frequencies together, synchronizing them. They may all originate in a single oscillator and use special circuits such as frequency dividers, which reduce
the oscillation rate by a selectable factor, or frequency multipliers, which increase it. For
example, a computer may have a 133MHz oscillator that directly controls the memory
system. A frequency divider may reduce that to one-quarter the rate to run the PCI bus,
and another divider may reduce it by 24 to produce the clock for the ISA legacy bus,
should the computer have one.
To go the other way, a frequency multiplier can boost the oscillator frequency to a higher
value, doubling the 133MHz clock to 266MHz or raising it to some other multiple.
CHAPTER 6 CHIPSETS
System Control
Because all the frequencies created through frequency division and multiplication originate in a single clock signal, they are automatically synchronized. Even the most minute
variations in the original clock are reflected in all those derived from it.
Some sections of computers operate asynchronously using their own clocks. For example,
the scan rate used by your computers video system usually is derived from a separate
oscillator on the video board. In fact, some video boards have multiple oscillators for different scan rates. Some computers may even have separate clocks for their basic system
board functions and run their buses asynchronously from their memory and microprocessor. For keeping track of the date, most computers have dedicated real-time clocks that
are essentially a digital Timex grafted into its chipset.
Although engineers can build the various oscillators needed in a computer into the
chipsetonly the crystal in its large metal can must reside outsidethe current design
trend is to generate the needed frequencies outside the chipset. Separate circuits provide
the needed oscillators and send their signals to the chipset (and elsewhere in the computer) to keep everything operating at the right rates.
System Clocks
Current Intel chipsets use eight different clock frequencies in their everyday operation.
They are used for everything from sampling sounds to passing bytes through your computers ports.
What Intel calls the oscillator clock operates at an odd-sounding 14.31818MHz rate. This
strange frequency is a carryover from the first personal computer, locked to four times the
3.58MHz subcarrier frequency used to put color in images in analog television signals.
The engineers who created the original computer thought compatibility with televisions
would be an important design element of the computerthey thought you might use one
instead of a dedicated monitor. Rather than anticipating multimedia, they were looking
for a cheap way of putting computer images onscreen. When the computer was released,
no inexpensive color computer monitors were available (or necessary because color
graphic software was almost nonexistent). Making computers work with color televisions
seemed an expedient way of creating computer graphics.
The TV-as-monitor never really caught on, but the 14.31818MHz frequency persists, but
not for motherboard television circuits. Instead, the chipset locks onto this frequency and
slices it down to 1.19MHz to serve as the timebase for the computers timer/counter circuit.
Because the chipset serves as the PCI bus controller, it also needs a precision reference
from which to generate the timing signals used by the PCI bus. The PCI clock provides
whats needed: a 33MHz signal. In addition, current Intel chipsets also need another clock
161
162
for the PCI bus, CLK66, which operates at 66MHz. These two signals are usually derived
from the same timebase, keeping them exactly synchronized. Typically a 66MHz may be
divided down to 33MHz.
New chipsets require a USB clock, an oscillator that provides the operating frequency for
the Universal Serial Bus (USB) port circuitry in the chipset, to develop the signals that let
you connect peripherals to your computer. This clock operates at 48MHz. It can be
divided down to 12MHz for USB 1.1 signals or multiplied up to 480MHz for USB 2.0.
Intels current chipsets have built-in audio capabilities, designed to follow the AC97 audio
standard. The chipsets use the AC97 Bit Clock as a timebase for digital audio sampling.
The required clock operating frequency is 12.288MHz. It is usually supplied by the audio
codec (coder/decoder), which performs the digital-to-analog and analog-to-digital conversion on the purely digital audio signals used by the chipset. A single source for this frequency ensures that the chipset and codec stay synchronized.
The Advanced Programmable Interrupt Controller (APIC) system used by the PCI bus
sends messages on its own bus conductors. The APIC clock provides a clocking signal (its
rising edge indicates when data on the APIC bus is valid). According to Intels specifications, its chipsets accept APIC clocks operating at up to 33.3MHz (for example, a signal
derived from the PCI bus clock).
The SMBus clock is one of the standard signals of the System Management Bus. All
SMBus interfaces use two signals for carrying data: a data signal and the clock signal. The
chipset provides this standard clock signal to properly pace the data and indicate when it is
valid.
Current Intel chipsets integrate a standard Ethernet local area network controller into
their circuitry, designed to work with an external module called LAN Connect, which provides both the physical interface for the network as well as a clock signal for the chipsets
LAN circuitry. The LAN I/F clock operates at one-half the frequency of the network carrier, either 5 or 50MHz under current standards.
Real-Time Clock
The first computers tracked time only when they were awake and running, simply counting the ticks of their timers. Every time you booted up, your computer asked you the time
to set its clock. To lift that heavy chore from your shoulders, all personal computers built
since 1984 have a built-in real-time clock among their support circuits. Its like your computer is wearing its own wristwatch (but, all too often, not as accurate).
All these built-in real-time clocks trace their heritage back to a specific clock circuit, the
Motorola MC146818, essentially an arbitrary choice that IBMs engineers made in designing the personal computer AT. No computer has used the MC146818 chip in over a
CHAPTER 6 CHIPSETS
System Control
decade, but chipsets usually include circuitry patterned after the MC146818 to achieve full
compatibility with older computers.
The real-time clocks have their own timebases, typically an oscillator built into the chip,
one thats independent from the other clocks in a computer. The oscillator (and even the
entire clock) may be external to the chipset. Intels current design puts all the real-timeclock circuitry, including the oscillator, inside the chipset. To be operational it requires
only a handful of discrete components and a crystal tuned to a frequency of 32.768 KHz.
A battery supplies current to this clock circuitry (and only this clock circuitry) when your
computer is switched offto keep accurate time, this clock must run constantly.
The battery often is not enough to ensure the accuracy of the real-time clock. Although it
should be as accurate as a digital watch (its circuitry is nothing more than a digital watch
in a different form), many tell time as imaginatively as a four-year old child. One reason is
that although a quartz oscillator is excellent at maintaining a constant frequency, that frequency may not be the one required to keep time correctly. Some computers have trimmers for adjusting the frequency of their real-time clocks, but their manufacturers
sometimes dont think to adjust them properly.
In many computers, a trimmer (an adjustable capacitor), in series with a quartz crystal,
allows the manufactureror anyone with a screwdriverto alter the resonate frequency
of the oscillator. Giving the trimmer a tweak can bring the real-time clock closer to realityor further into the twilight zone. (You can find the trimmer by looking for the short
cylinder with a slotted shaft in the center near the clock crystal, which is usually the only
one in a computer with a kilohertz rather than megahertz rating.) There are many other
solutions to clock inaccuracy. The best is to regularly run a program that synchronizes
your clock with that of the National Institute of Standards and Technologys atomic clock,
the official U.S. time standard. You can download this time-setting program from the
NIST Web site at tf.nist.gov/timefreq/service/its.htm.
The real-time clocks in many computers are also sensitive to the battery voltage. When
the CMOS backup battery is nearing failure, the real-time clock may go wildly awry. If
your computer keeps good time for a while (months or years) and then suddenly wakes up
hours or days off, try replacing your computers CMOS backup battery.
The real-time clock chip first installed in early computers also had a built-in alarm function, which has been carried into the real-time clocks of new computers. As with the rest
of the real-time clock, the alarm function runs from the battery supply so it can operate
while your computer is switched off. This alarm can thus generate an interrupt at the
appropriate time to start or stop a program or process. In some systems, it can even cause
your computer to switch at a preset time. Throw a WAV file of the national anthem in
your startup folder, and youll have a truly patriotic alarm clock.
163
164
The clock circuitry also includes 128 bytes of low-power CMOS memory that is used to
store configuration information. This is discussed in the next chapter.
Timers
In addition to the clock in your computerboth the oscillators and the real-time clock
youll find timers. Digitally speaking, computer timers are simply counters. They count
pulses received by an oscillator and then perform some action when they achieve a preset
value. These timers serve two important functions. They maintain the real time for your
computer separately from the dedicated real-time clock, and they signal the time to
refresh memory. Also, they generate the sounds that come from the tiny speakers of computers that lack multimedia pretensions.
The original IBM personal computer used a simple timer chip for these functions, specifically chip type 8253. Although this chip would be as foreign in a modern computer as a
smudge pot or buggy whip, modern chipsets emulate an updated version of that timer
chip, the 8254. The system timer counts pulses from the 1.19MHz clock (that the chipset
derives from the 14.31818MHz system clock) and divides them down to the actual frequencies it needs (for example, to make a beep of the right pitch).
The timer works simply. You load one of its registers with a number, and it counts to that
number. When it reaches the number, it outputs a pulse and starts all over again. Load a
8254 register with 2, and it sends out a pulse at half the frequency of the input. Load it
with 1000, and the output becomes 1/1000th the input. In this mode, the chip (real or
emulated) can generate an interrupt at any of a wide range of user-defined intervals.
Because the highest value you can load into its 16-bit register is 216 or 65,536, the longest
single interval it can count is about .055 seconds (that is, the 1.19MHz input signal
divided by 65,536).
The timer circuitry actually had six operating modes, all of which are carried through on
modern computers. Table 6.1 lists these modes.
TA B L E 6 . 1
Operating Modes of the 8253 Timer/Counter Chip
Name
Operation
Interrupt on Terminal Count
The timer is loaded with a value and counts down from

that value to zero, one counter per clock pulse.
Hardware Retriggerable One-Shot
A trigger pulse causes timer output to go low; when the

counter reaches zero, the output goes high and stays high
until reset. The process repeats every time its triggered.
The pulse length is set by writing a control word and initial count to chip before the first cycle.
Rate Generator
The timer divides incoming frequency by the value of the

initial count loaded into it.
CHAPTER 6 CHIPSETS
System Control
Name
Operation
Square Wave
Produces a series of square waves with a period (measured

in clock pulses) equal to the value loaded into the timer.
Software Retriggerable Strobe
The timer counts down the number of clock cycles loaded

into it; then it pulses its output. Software starts the next
cycle.
Hardware Retriggerable Strobe
The timer counts down the number of clock cycles loaded

into it; then it pulses its output. A hardware-generated
pulse initiates the next cycle.
The real-time signal of the timer counts out its longest possible increment, generating
pulses at a rate of 18.2 per second. The pulses cause the real-time interrupt, which the
computer counts to keep track of the time. These interrupts can also be used by programs
that need to regularly investigate what the computer is doing; for instance, checking the
hour to see whether its time to dial up a distant computer.
The speaker section of the timer system works the same way, only it generates a square
wave that is routed through an amplifier to the internal speaker of the computer to make
primitive sounds. Programs can modify any of its settings to change the pitch of the tone,
and with clever programming, its timbre.
In modern computers, the memory controller in the chipset rather than the system timer
handles refresh operations. Otherwise, the timer functions are much the same as they
were with the first computers, and the same signals are available to programmers.
Interrupt Control
Timers and strange oscillator frequencies arent the only place where computers show
their primitive heritage. Modern machines also carry over the old interrupt and direct
memory access systems of the first personal computers, and its the chipsets job to emulate these old functions. Most are left over from the design of the original PC expansion
bus that became enshrined as Industry Standard Architecture. Although the newest computers dont have any ISA slots, they need to incorporate the ISA functions for full compatibility with old software.
The hardware interrupts that signal microprocessors to shift their attention originate outside the chip. In the first personal computers, a programmable interrupt controller funneled
requests for interrupts from hardware devices to the microprocessor. The first personal
computer design used a type 8259A interrupt controller chip that accepted eight hardware
interrupt signals. The needs of even basic computers quickly swallowed up those eight
interrupts, so the engineers who designed the next generation of computers added a second 8259A chip. The interrupts from the new chip cascade into one interrupt input of the
165
166
old one. That is, the new chip consolidates the signals of eight interrupts and sends them
to one of the interrupt inputs of the chip used in the original design. The eight new interrupts, combined with the seven remaining on the old chip, yield a total of 15 available
hardware interrupts.
The PCI bus does not use this interrupt structure, substituting a newer design that uses a
four-wire system with serialized interrupts (discussed in Chapter 9, Expansion Buses).
The PCI system uses four interrupt lines, and the function of each is left up to the
designer of each individual expansion board. The PCI specification puts no limits on how
these interrupt signals are used (the software driver that services the board determines
that) but specifies level-sensitive interrupts so that the four signals can be reliably shared.
However, all computers that maintain an ISA legacy buswhich currently includes nearly
all computersmimic this design for compatibility reasons. Several legacy functions, such
as the keyboard controller, legacy ports, real-time clock, and IDE disk drives still use
these interrupts.
In this legacy scheme, each interrupt is assigned a priority, the more important one getting attention before those of lesser importance. In the original computer scheme of
things with eight interrupts, the lower the number assigned to an interrupt, the most priority it got. The two-stage design in current systems, however, abridges the rule.
In the current cascade design, the second interrupt controller links to the input of the first
controller that was assigned to interrupt 2 (IRQ2). Consequently, every interrupt on the
second controller (that is, numbers 8 through 15) has a higher priority than those on the
first controller that are numbered higher than 2. The hardware-based priorities of interrupts in current computers therefore ranges as listed in Table 6.2, along with the functions
that usually use those interrupts.
TA B L E 6 . 2
The Priorities of Interrupts
Interrupt Number
Priority
Function
IRQ0
Timer output 0
IRQ1
Keyboard (buffer full)
IRQ2
Unusable
Cascade from IRQ9
IRQ3
11
Serial port 2, serial port 4, SDLC communications, BSC

communications, cluster adapter, network (alternate),
3278/79 (alternate)
IRQ4
12
Serial port 1, serial port 3, SDLC communications, BSC

communications, voice communications adapter
IRQ5
13
Parallel port 2, audio
CHAPTER 6 CHIPSETS
System Control
Interrupt Number
Priority
Function
IRQ6
14
Floppy disk controller
IRQ7
15
Parallel port 1, cluster adapter (alternate)
IRQ8
Real-time clock
IRQ9
Software redirected to INT 0A(hex), video, network, 3278/79

adapter
IRQ10
Reserved
IRQ11
Reserved
IRQ12
Reserved, built-in mouse
IRQ13
Coprocessor
IRQ14
Primary hard disk controller (IDE)
IRQ15
10
Secondary hard disk controller
The original Plug-and-Play configuration system altered how computers assign interrupt
numbers, making the assignments automatically without regard to the traditional values.
In these systems, interrupt values and the resulting priorities get assigned dynamically
when the computer configures itself. The values depend on the combination of hardware
devices installed in the such systems. Although the standardized step-by-step nature of
Plug-and-Play and its Advanced Configuration and Power Interface (ACPI) successor usually assign the same interrupt value to the same devices in a given computer each time that
computer boots up, it doesnt have to. Interrupt values could (but usually do not) change
day to day.
A few interrupt assignments are inviolable. In all systems, four interrupts can never be
used for expansion devices. These include IRQ0, used by the timer/counter; IRQ1, used
by the keyboard controller; IRQ2, the cascade point for the upper interrupts (which is
redirected to IRQ9); and IRQ8, used by the real-time clock. In addition, all modern computers have microprocessors with integral floating-point units (FPUs) or accept external
FPUs, both of which use IRQ13.
If you want to know whats what, you can check interrupt assignment through Windows.
To view the actual interrupt usage of your computer, you need to consult the Device
Manager, available through the System icon in Control Panel. In Device Manager, select
View, Resources from the drop-down menu or (under Windows 95, 98, and Me) the View
Resources tab. Select your computer on the Device menu and click the Properties button.
Click the Interrupt Request menu selection or radio button (depending on your version of
Windows), and youll see a list of the actual interrupt assignments in use in your
computer, akin whats shown in Figure 6.1.
167
168
FIGURE 6.1
Windows Device
Manager showing
actual interrupt
assignments.
Although 15 interrupts seem like a lot, this number is nowhere near enough to give every
device in a modern computer its own interrupt. Fortunately, engineers found ways to let
devices share interrupts. Although interrupt sharing is possible even with the older computer designs, interrupt sharing works best when the expansion bus and its support circuitry are designed from the start for sharing. The PCI system was; ISA was not.
When an interrupt is shared, each device thats sharing a given interrupt uses the same
hardware interrupt request line to signal to the microprocessor. The interrupt-handling
software or firmware then directs the microprocessor for the action to take to service the
device making the interrupt. The interrupt handler distinguishes between the various
devices sharing the interrupt to ensure only the right routine gets carried out.
Although software or firmware actually sorts out the sharing of interrupts, a computer
expansion bus needs to be properly designed to make the sharing reliable. Old bus designs
such as the ISA legacy bus use edge-triggered interrupts, a technique that is particularly
prone to sharing problems. Modern systems use a different technology, level-sensitive
interrupts, which avoids some of the pitfalls of the older technique.
An edge-triggered interrupt signals the interrupt condition as the transition of the voltage
on the interrupt line from one state to another. Only the changethe edge of the waveformis significant. The interrupt signal needs only to be a pulse, after which the interrupt line returns to its idle state. A problem arises if a second device sends another pulse
CHAPTER 6 CHIPSETS
System Control
down the interrupt line before the first has been fully processed. The system may lose
track of the interrupts it is serving.
A level-sensitive interrupt signals the interrupt condition by shifting the voltage on the
interrupt request line (for example, from low to high). It then maintains that shifted condition throughout the processing of the interrupt. It effectively ties up the interrupt line
during the processing of the interrupt so that no other device can get attention through
that interrupt line. Although it would seem that one device would hog the interrupt line
and not share with anyone, the presence of the signal on the line effectively warns other
devices to wait before sending out their own interrupt requests. This way, interrupts are
less likely to be confused.
Direct Memory Access

The best way to speed up system performance is to relieve the host microprocessor of all
its housekeeping chores. Among the more time consuming is moving blocks of memory
around inside the computer; for instance, shifting bytes from a hard disk (where they are
stored) through its controller into main memory (where the microprocessor can use
them). Todays system designs allow computers to delegate this chore to support chips and
expansion boards (for example, through bus mastering). Before such innovations, when
the microprocessor was in total control of the computer, engineers developed Direct
Memory Access (DMA) technology to relieve the load on the microprocessor by allowing
a special device called the DMA controller to manage some device-to-memory transfers.
DMA was an integral part of the ISA bus design. Even though PCI lacks DMA features
(and soon computers will entirely lack ISA slots), DMA technology survives in motherboard circuitry and the ATA disk interface. Consequently, the DMA function has been
carried through into modern chipsets.
DMA functions through the DMA controller. This specialized chip only needs to know
the base location of where bytes are to be moved from, the address to where they should
go, and the number of bytes to move. Once it has received that information from the
microprocessor, the DMA controller takes command and does all the dirty work itself.
DMA operations can be used to move data between I/O devices and memory. Although
DMA operations could in theory also expedite the transfer of data between memory locations, this mode of operation was not implemented in computers.
The current DMA system offers seven channels, based on the design of the IBM personal
computer AT, which used a pair of type 8237A DMA controller chips. As with the interrupt system, the prototype DMA system emulated by modern chipsets used two fourchannel DMA controllers, one cascaded into the other. Each DMA channel is 16 bits
wide. Table 6.3 summarizes the functions commonly assigned to the seven DMA channels.
169
170
TA B L E 6 . 3
DMA Assignments of a Modern Motherboard
DMA Channel
Data Width
Function
8 or 16 bits
Audio
8 or 16 bits
Audio or LAN
8 or 16 bits
Floppy disk
8 or 16 bits
ECP/EPP port
16 bits
Cascade channel
16 bits
Not assigned
16 bits
Not assigned
16 bits
ISA IDE
You can check the DMA assignment of your computer much as you do interrupts,
through the Device Manager. Simply select DMA from the drop-down menu or radio
button instead of Interrupts. You should see your computers current DMA assignments,
listed much like Figure 6.2.
FIGURE 6.2
Windows Device
Manager showing
DMA usage.
In general, individual tasks use only one DMA channel at a time. When you run multiple
tasks at the same time, nothing changes under Windows 95, 98, or Me, because they serialize I/O. That means those operating systems handle only one I/O task at a time and
switch between them. Only a single path for data is available. In rare instances, however, a
path may have a DMA channel at both ends (which puts two channels into operation
simultaneously). Windows NT and 2000 know no such limits and can run multiple I/O
tasks and use multiple DMA channels.
CHAPTER 6 CHIPSETS
System Control
Bus Interface
In early computers, the expansion bus was nothing more than an extension of the connections on the microprocessor. In modern designs, the bus is a free-standing entity with its
own bus controller separate from the microprocessor. In fact, the microprocessor connects
to the bus as if it were just another expansion device. All modern chipsets use this latter
design to control the expansion bus. Not only does the chipset take on the management of
the PCI bus, but it also uses this decentralized control system for the legacy ISA bus,
should one be used in your computer. The chipset takes over the role of the microprocessor in the traditional ISA system. In fact, the entire ISA bus gets treated as if it were just
another PCI device.
The PCI bridge in the chipset handles all the functions of the PCI bus. Circuitry built in
to the PCI bridge, independent of the host microprocessor, operates as a DMA master.
When a program or the operating system requests a DMA transfer, the DMA master
intercepts the signal. It translates the commands bound for the 8237A DMA controller in
older computers into signals compatible with the PCI design. These signals are routed
through the PCI bus to a DMA slave, which may in turn be connected to the ISA compatibility bus, PCMCIA bus, disk drives, or I/O ports in the system. The DMA slave interprets the instructions from the DMA master and carries them out as if it were an ordinary
DMA controller.
In other words, to your software the DMA master looks like a pair of 8237A DMA controller chips. Your hardware sees the DMA slave as an 8237A thats in control. The signal
translations to make this mimicry possible are built in to the PCI bridges in the chipset.
Power Management
To economize on power to meet Energy Star standards or prolong the period of operation
on a single battery charge, most chipsetsespecially those designed for notebook computersincorporate power-management abilities. These include both manual power control,
which lets you slow your computer or suspend its operation, and automatic control, which
reduces power similarly when it detects system inactivity lasting for more than a predetermined period.
Typically, when a chipset switches to standby state, it will instruct the microprocessor to
shift to its low-power mode, spin down the hard disk, and switch off the monitor. In
standby mode, the chipset itself stays in operation to monitor for alarms, communications
requests (answering the phone or receiving a fax), and network accesses. In suspend mode,
the chipset itself shuts down until you reactivate it, and your computer goes into a vegetative state.
By design, most new chipsets conform to a variety of standards. They take advantage of
the System Management Mode of recent Intel microprocessors and use interrupts to
direct the power conservation.
171
172
Peripheral Control
If your computer didnt have a keyboard, it might be less intimidating, but you probably
wouldnt find it very useful. You might make do, drawing pictures with your mouse or trying to teach it to recognize your handwriting, but some devices are so integral to using a
computer that you naturally expect them to be a part of it. The list of these mandatory
components includes the keyboard (naturally), a mouse, a hard disk, a CD drive, ports,
and a display system. Each of these devices requires its own connection system or interface. In early computers, each interface used a separate expansion board that plugged into
a slot in the computer. Worse than that, you often had to do the plugging.
You probably dont like the sound of such a system, and computer manufacturers liked it
even less. Individual boards were expensive and, with most people likely not as capable as
you in plugging in boards, not very reliable. Clever engineers found a clever way to avoid
both the problems and expenses. They built the required interfaces into the chipset. At
one time, all these functions were built into a single chip called the peripheral controller, an
apt description of its purpose. Intels current chipset design scheme puts all the peripheral
control functions inside the I/O controller hub.
As with the system-control functions, most of the peripheral interface functions are welldefined standards that have been in use for years. The time-proven functions usually cause
few problems in the design, selection, or operation of a chipset. Some interfaces have a
more recent ancestry, and if anything will cause a problem with a chipset, they will. For
example, the PCI bus and ATA-133 disk interface (described in Chapter 10, Interfaces)
are newcomers when compared to the decade-old keyboard interface or even-older ISA
bus. When a chipset shows teething pains, they usually arise in the circuits relating to
these newer functions.
Hard Disk Interface

Most modern chipsets include all the circuitry needed to connect a hard disk to a computer, typically following an enhancement of the AT Attachment standard, such as
UDMA. Although this interface is a straightforward extension of the classic ISA bus, it
requires a host of control ports for its operation, all provided for in the chipset. Most
chipsets now connect the ISA bus through their PCI bridges, which may or may not be an
intrinsic part of the south bridge.
The chipset thus determines which modes of the AT Attachment interface your computer
supports. Intels current chipsets, for example, support all ATA modes up to and including
ATA/100. They do not support ATA/133, so computers based on Intel chipsets cannot take
full advantage of the highest speeds of the latest ATA (IDE) hard disk drives.
CHAPTER 6 CHIPSETS
Peripheral Control
Floppy Disk Controller

Chipsets often include a variety of other functions to make the computer designers life
easier. These can include everything from controls for indicator lights for the front panel
to floppy disk controllers. Nearly all chipsets incorporate circuitry that mimics the NEC
765 floppy disk controller used by nearly all adapter boards.
In that the basic floppy disk interface dates back to the dark days of the first computers, it
is a technological dinosaur. Computer-makers desperately want to eliminate it in favor of
more modern (and faster) interface alternatives. Some software, however, attempts to control the floppy disk by reaching deep into the hardware and manipulating the floppy controller directly through its powers. To maintain backward compatibility with new floppy
interfaces, the chipset must still mimic this original controller and translate commands
meant for it to the new interface.
Keyboard Controller
One additional support chip is necessary in every computer: a keyboard decoder. This
special-purpose chip (an Intel 8042 in most computers, an equivalent chip, or part of a
chipset that emulates an 8042) links the keyboard to the motherboard. The primary function of the keyboard decoder is to translate the serial data that the keyboard sends out into
the parallel form that can be used by your computer. As it receives each character from
the keyboard, the keyboard decoder generates an interrupt to make your computer aware
you have typed a character. The keyboard decoder also verifies that the character was correctly received (by performing a parity check) and translates the scan code of each character. The keyboard decoder automatically requests the keyboard to retransmit characters
that arrive with parity errors.
USB-based keyboards promise the elimination of the keyboard controller. These pass
packets of predigested keyboard data through the USB port of your computer instead of
through a dedicated keyboard controller. In modern system design, the keyboard controller is therefore a legacy device and may be eliminated from new systems over the next
few years.
Input/Output Ports
The peripheral control circuitry of the chipset also powers most peripheral ports. Todays
computers usually have legacy serial and parallel ports in addition to two or more USB
connections. As with much else of the chipset design, the legacy ports mimic old designs
made from discrete logic components. For example, most chipsets emulate 16550
universal asynchronous receiver/transmitters (UARTs) to operate their serial ports.
173
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
BIOS
Think of the Basic Input/Output System of your computer as its crazy
Aunt Maud, locked away from public eyes in a cobwebby garret. Except
for the occasional deranged laughter that rings through the halls, you
might never know she was there, and you dont think about her again
until the next reel of the B-movie that makes up your life.
As with your own old Aunt Maud, your computers BIOS is something
that you want to forget but is always there, lingering in the background,
popping into sight only at the least convenient times. Despite its idiosyncrasies and age, despite the embarrassment it causes, the BIOS is something your computer cant live without. It defines what your computer is
and keeps it in line, just as Aunt Maud defines what your family really is
and her antics keep you in line (or at least tangled up in obscure legal
proceedings). Youd really like to be rid of Aunt Maud, but only she
knows the secret of your familys jewels, a secret someday you hope
youll wrest from her.
The BIOS of your computer lingers around like that unwelcome relative,
but it also holds the secrets of your computer. You can lock it up, even
build a wall around it, but it will always be there. When you switch on
your computer, it laughs at you from the monitor screen, appearing in
the rags and tatters of text mode before your system jumps off into modern high-resolution color. Most of what you do on your computer now
sidesteps the BIOS, so youd think you could do without it. You might
never suspect that behind the scenes the BIOS of your computer tests
your system, assures you that everything is okay when you start your system, helps you set up your computer so that it runs at its best, and gracefully steps out of the way when you no longer need it.
176
Although modern operating systems such as Windows do their best to hide the BIOS and
take its place, it always lurks in the background like crazy old Aunt Maud locked in the
garret. Just as Aunt Maud may ruin a dinner party when she suddenly pops up babbling
about a talking moose, your BIOS bursts onto the scene when you least expect it and ruins
your fun (for example, hiding half of your new hard disk).
However, your Aunt Maud might just not be as crazy as she seems, watching over you
quietly from her garret, hiding in the background but working in her own mysterious
ways to make sure your life goes well.
Background
Although mostly invisible and oft forgotten, your computers BIOS is nevertheless one of
its most important and enabling parts. The BIOS is, in fact, the one essential constituent
that distinguishes one computer from another, even when they both share the same
microprocessor, motherboard, and support hardware.
Firmware
Strictly speaking, however, the BIOS isnt hardware at all, even though it is an essential
part of your computers hardware. The BIOS is special program codein a word,
softwarethats permanently (or nearly so) encapsulated in ROM chips or, as is most often
the case with newer computers, Flash memory. Because of the two-sided aspects of the
BIOS, existing in the netherworld between hardware and software, it and other pieces of
program code encapsulated in ROM or Flash memory are often termed firmware.
The importance of the BIOS arises from its function. The BIOS tests your computer
every time you turn it on. It may even allocate your systems resources for you automatically, making all the adjustments necessary to accommodate new hardware. The BIOS also
determines the compatibility of your computer with both hardware and software and can
even determine how flexible your computer is in setup and use.
Firmware can do more. If some enterprising computer-maker wanted to, it could put the
entire operating system and application software into firmware. Some portable computers
actually came closeearly Hewlett-Packard Omnibook sub-notebook computers packed
Windows and part of Microsoft Office into their ROMs. Of course, these machines are
not very useful any more. The Windows inside is version 3.0, which points out the disadvantage of putting too much into firmwareupdates are cumbersome and the code-storage requirements of todays software would fill more chips than you could comfortably
carry. The storage available in a computers ROM gives engineers kilobytes to work with,
not the gigabytes todays systems demand for storage of their software.
CHAPTER 7 BIOS
Background
Today, firmware plays only a subsidiary role in most computers. All modern operating systems enhance the basic BIOS firmware with additional instructions loaded from disk like
ordinary softwaretypically such enhancements totally replace the BIOS firmware. This
new code supplied by the operating system performs some of the same functions as the
traditional BIOS firmware, linking your computers hardware to the software programs
that you run. But every computer still requires at least a vestigial piece of BIOS firmware,
if just to enable enough of your system to run so that it can load the operating system.
Although the BIOS plays a less active role in every operation of your computer, it is
essential to getting your computer going, and it remains an essential part of every new
computer and future machines still waiting on the engineers drawing board.
Functions
The BIOS code of most computers has a number of separate and distinct functions. The
BIOS of a typical computer has routines that test the computer, blocks of data that give
the machine its personality, special program routines that allow software to take control of
the computers hardware so that it can more smoothly mesh with the electronics of the
system, and even a complete system (in some computers) for determining which expansion
boards and peripherals you have installed and ensuring that they do not conflict in their
requests for input/output ports and memory assignments. Although all these functions get
stored in the same memory chips, the program code of each function is essentially independent of the rest. Each function is a separate module, and the name BIOS refers to the
entire group of modules.
This list of functions is not an exhaustive list of what a BIOS could do. It represents only
what the makers of current computers use the BIOS for. In fact, theres little limit on what
BIOS code can do. Apple has long (since 1984) put most of the graphics of its Macintosh
systems in its BIOS, and IBM for years encapsulated a small programming language in the
BIOSs of its first computers.
The classic definition of the computer BIOS is the firmware that gives the computer its
personality. This definition refers only to one functional module of the computerthe one
thats invariably replaced by operating system code. The nebulous term personality
described how the computer performed its basic functions, those necessary to make it a
real computer. Although this definition included a number of different factors, including
how quickly and smoothly various operations were completed, the term personality mostly
distinguished computers from Apple Macintoshes.
Most of the personality of the computer has moved from its BIOS to the operating system, loaded entirely from disk storage into RAM. The BIOS now plays a subsidiary role.
Chiefly it is in charge of getting your computer going and running things until your hard
disk can disgorge the hundreds of megabytes of operating system code and software can
take over control of your computer.
177
178
In most computers, the first thing the BIOS tells the microprocessor to do is to run
through all the known components of the systemthe microprocessor, memory, keyboard, and so onand to test to determine whether they are operating properly. After the
system is sure of its own integrity, it checks to see whether you have installed any expansion boards that hold additional BIOS code. If you have, the microprocessor checks the
code and carries out any instructions it finds. A modern computer may even check to see
whether any new expansion boards are plugged in without being set up properly. The
BIOS code might then configure the expansion board so that it functions properly in your
computer.
When the microprocessor runs out of add-in peripherals, it begins the actual bootup
process, which engineers call the Initial Program Load (IPL). The BIOS code tells the
microprocessor to jump to a section of code that tells the chip how to read the first sector
of your floppy or hard disk. Program code then takes over from the BIOS and tells the
microprocessor how to load the operating system from the disk to start the computer running.
Exactly what the BIOS does after the operating system loads depends on the operating
system. The first computer operating system, DOS, worked in conjunction with the
BIOS. DOS relied on the BIOS firmware, which includes several sets of routines that programs can call to carry out everyday functions, such as typing characters on the screen or
to a printer, reading keystrokes, and timing events. Because of this basic library, programmers writing for DOS could create their grand designs without worrying about the tiny
details.
With Windows, however, the BIOS gets pushed out of the way. After the BIOS has
assured the operating integrity of your system, Windows takes over. First, the operating
system starts loading its own boot code. Then, it installs drivers that take over the various
interface functions of the BIOS, one by one.
The operating system replaces the BIOS interface routines for several reasons. Because
software drivers load into RAM, they are not limited in the amount of space available for
their code. Software drivers also extend the capabilities, whereas the BIOS limits them.
Using only the BIOS, your computer cannot do anything that the BIOS does not know
about. It enables you to use the wide variety of peripherals youre apt to connect to your
computer. Moreover, because the BIOS is designed to run at bootup when your computer
is in real mode, it uses only real-mode code, with all the limitations that implies (in particular, a total address space of 1MB). Software drivers can (and nowadays invariably do) run
in protected mode. Not only can they access more memory, but protected mode drivers
can be written in 32-bit code that executes faster on modern microprocessors.
CHAPTER 7 BIOS
Initialization Functions
The BIOS starts to work as soon as you switch your system on. When all modern Intel
microprocessors start to work, they immediately set themselves up in real mode and look
at a special memory location that is exactly 16 bytes short of the top of the 1MB realmode addressing rangeabsolute address 0FFFF0(hex). This location holds a special program instruction, a jump that points to another address where the BIOS code actually
begins.
The term cold boot describes the process of starting your computer and loading its operating system by turning the power on. If your computer is running, you cold boot by first
switching it off and then back on.
Warm boot describes the process of restarting your computer and loading its operating system anew after it has already been running and has booted up at least once before. You
start a warm boot by giving the infamous three-finger salute by pressing the Ctrl, Alt,
and Delete keys at the same time.
Finally, hot boot describes what you get when you slide a piece of footwear into your oven
in a mistaken attempt at the preparation of filet of sole. The term is not used to describe
the computer bootup process.
At the operating system level, a cold boot and a warm boot are essentially the same. Your
computer starts from the beginning and loads the operating system from scratch. A warm
boot or switching your computer off for a cold boot signals the microprocessor to reset
itself to its turn-on condition, erasing the contents of its registers. The microprocessor
then loads or reloads the operating system.
The important difference between a cold and warm boot is not what happens to your
operating system but the effect on your computers internal circuits. A cold boot automatically restores all the circuits in your computer to their original, default condition, whether
they are on the motherboard or the expansion boards, because it cuts off their electrical
supply. It also wipes away everything in its memory for a fresh start. A warm boot does
not affect the supply of electricity to your computers circuitry, so memory and the boards
installed in your computer are not wiped clean, although some of the contents get overwritten as your operating system reloads.
Because a warm boot does not automatically restore all the expansion boards in your computer to their initial conditions, it sometimes does not solve software problems. For example, your modem may not release the telephone line (hang up) at the end of your Internet
session. A warm boot may leave the modem connected, but a cold boot will ensure the
modem disconnects and releases your telephone line.
179
180
Unless the boards in your computer follow the Plug-and-Play standard that specifies a
standard reset condition, your computer has no way of telling the state of each board in
your computer after a warm boot. It makes no attempt to find out and just takes what it
gets. Ordinarily such blind acceptance is not a problem. If, however, some odd state of an
expansion board caused your computer to crash, a warm boot will not solve the problem.
For example, sometimes video boards will come up in odd states with strange screen displays after a crash thats followed by a warm boot. Cold-booting your computer again usually eliminates such problems.
Your computer also behaves differently during the cold- and warm-booting processes.
During a cold boot, your computer runs through its Power-On Self Test (POST) procedure to test all its circuitry. During a warm boot, your computer sidesteps POST under
the assumption that it has already booted up once, so its circuitry must be working properly.
Your computer must somehow distinguish between a cold and warm boot to decide
whether to run its POST diagnostics. To sort things out, your computer uses its normal
memory, which it does not wipe out during a warm boot. Each time your computer boots
up, it plants a special two-byte signature in memory. When your system boots, it looks for
the signature. If it find the signature, it knows it has been booted at least once since you
turned on the power, so it does not need to run through POST. When it fails to find the
signature, it runs its diagnostics as part of the cold-boot process. Note that if something in
your system changes the signature bytesas crashing programs sometimes doyour computer will run through a cold boot even though you havent turned it off.
The signature bytes have the value 1234(hex). Because they are stored in Intel little endian
format (that is, the least significant byte comes first), they appear in memory as the
sequence 34 12.
Programs can also initiate a warm or cold boot simply by jumping to the appropriate section of BIOS code. However, because some expansion boards dont automatically reset
when your computer runs through the cold-boot BIOS code, anomalies may persist after
such a program-initiated cold boot. For example, your video board may place itself into an
odd state, and a complete reset may not unlock it. For this reason, some programs instruct
you to turn your computer off and back on during the installation process to guarantee all
the hardware in your system properly resets.
Power-On Self Test

Every time your computer switches on, the BIOS immediately takes command. Its first
duty is to run through a series of diagnostic routines (system checks) called the Power-On
Self Test routine, or POST. This ensures every part of your computers hardware is functioning properly before you trust your time and data to it. One by one, the POST routine
CHAPTER 7 BIOS
checks the circuits of your system board and memory, the keyboard, your disks, and each
expansion board. After the BIOS makes sure the system is operating properly, it initializes
the electronics so that they are ready for the first program to load.
Error Codes
The BIOS tests are relatively simple. The BIOS sends data to a port or register and then
looks to see the results. If it receives expected results, the BIOS assumes all is well. If it
finds a problem, however, it reports the failure as well as it can. If the display system is
working, it posts an error-code number on your monitor screen. (The limited amount of
memory available prevents the BIOS from storing an elaboratethat is, understandable
message for all the hundreds of possible error conditions.) If your computer is so ill that
the display system will not even work, the BIOS sends out a coded series of beeps through
your systems loudspeaker.
Many BIOSs also write the code numbers of the ongoing tests to input/output port
80(hex). Special diagnostic boards that plug into a vacant expansion slot can monitor this
test procedure and show the test progress and failure. Repair technicians often use such
diagnostic boards when servicing systems.
Beep Codes
Displaying an error code on your monitor screen presupposes that at least part of your
computer (and a major part at that) is operating properlynamely, the display system. If
your video board or monitor is not functioning and you dont have a diagnostic board, the
error code display is useless. Consequently, most BIOSs also generate error identifications
that are independent of the display system.
Instead of visual displays, they use audible error codes played through the small speaker
system that is standard equipment in all computers. Because the speaker system links to
your computer at a rudimentary levelonly a single amplifier chip stands between the
speaker and its associated input/output portthe basic computer speaker will always function except during the most dire failures, such as a nonworking microprocessor. (Note
that this basic computer speaker is separate from the sound board or multimedia sound
system present in most new computers.) These audible error messages take the form of
beeps from the system speaker. Consequently, they are usually termed beep codes.
As computers have become more complex, the systems of beep codes have become correspondingly elaborate. Some sound more elaborate than a Mayday call from a sinking
ocean liner. Ever BIOS-maker uses its own coding scheme for these aural error warnings.
Usually a single short beep means your computer is okay; anything else warns of an error.
Should your computer start with a continuous beep that wont stop until you switch the
power off, your computer has a major problem, most likely with its power or logic system.
Oftentimes a defective expansion board or an improperly inserted expansion board will
181
182
elicit such an unending beep. It can also result from a bad microprocessor or other motherboard problem. If you encounter a beep that wont end, check your expansion boards
first. Ensure that all are seated properly and then remove them one by one or en masse
until the beep quiets down. (Be sure to switch off your computer before you add or
remove an expansion board.)
BIOS Extensions
After the actual testing of your computer and its associated circuitry is complete, its BIOS
begins to execute initialization routines that configure the various options inside your
computer. Exactly what happens next depends on the design of the BIOS. A conventional
BIOS merely looks for add-in BIOS routines, initializes devices using the routines already
in the BIOS, and starts the disk bootup.
Every computer assumes that many of its peripherals are loaded with specific data values
when it starts up. That is, all the default operating values are loaded by the BIOS, so your
timer knows what to time and the speaker knows the frequency at which to beep. The serial ports are set to their default speed. A Plug-and-Play system like those of nearly all new
computers runs through a more structured initialization process that amounts to completely setting up system resources each time the computer boots. The Plug-and-Play
section, later in this chapter, gives a step-by-step outline of this BIOS boot procedure.
By design, the basic computer BIOS knows how to search beyond the confines of the system board. It knows how to locate the extra instructions and data that get added in with
some expansion boards. Although some computers require you to replace the BIOS
firmware with new chips to add new features, the computer BIOS was designed to be
extendible. That is, the full extent of the BIOS is not forever cast in the silicon of the
firmware. The extendible BIOS is capable of accepting additional code as its own, creating
one integrated whole. Rather than replacing the BIOS chips, this extendibility means that
you can add more firmware containing it own BIOS routines to your computer. The
BIOS incorporates the new routines into itself.
The key to making the BIOS extendible is itself an extra firmware routine that enables the
BIOS to look for add-in code. During the initialization process, the BIOS code reads
through the address range looking for code stored on add-in boards. If it finds a valid section of code, it adds those instructions to the BIOS repertory. New interrupt routines may
be added, for instance, or the function of existing routines may be changed.
The routine to extend the BIOS works as follows: Following the actual testing portion of
the Power-On Self Test, after basic system board functions have been initialized (for
example, the interrupt vectors have been loaded into RAM), the resident BIOS code
instructs the computer to check through its ROM memory for the occurrence of the special preamble bytes that mark the beginning of add-in BIOS routines. The original IBM
CHAPTER 7 BIOS
BIOS searches for these preamble bytes in the absolute address range 0C8000(hex) to
0F4000(hex); newer BIOSs check the range from 0C0000(hex) to 0EFFFF(hex). Either of
these subsets of the full, reserved, high memory rangethat is, 0A0000(hex) to
0FFFFF(hex)exclude the areas used by video memory and the BIOS itself to prevent
confusing data with preamble bytes.
If the BIOS finds the special preamble bytes, it verifies that the subsequent section of code
is a legitimate BIOS extension by performing a form of cyclical redundancy check on the
specified number of 512 byte blocks. The values of each byte in the block are totaled
using Modulo 0100(hex) additionthe effect is the same as dividing the sum of all the
bytes by 4096. A remainder of zero indicates that the extension BIOS contains valid code.
The preamble bytes take a specific form. Two bytes indicate the beginning of an extension
code section055(hex) followed by 0AA(hex). Immediately following the two-byte preamble bytes is a third byte that quantifies the length of the additional BIOS. The number
represents the number of blocks, 512-bytes long, needed to hold the extra code. Plug-andPlay peripherals have a more structured header that follows the preamble bytes.
After a valid section of code is identified, system control (BIOS program execution) jumps
to the fourth byte in the extension BIOS and performs any functions specified there in
machine language. Typically, these instructions tell the BIOS how to install the extra code.
Finally, when the instructions in the extension BIOS have been completed, control returns
to the resident BIOS. The system then continues to search for additional blocks of extension BIOS. When it finally completes its search by reaching the absolute address
0F4000(hex), it starts the process of booting up your computer from disk.
The ROM or Flash memory chips containing this extra BIOS code do not have to be present on the system board. The memory locations used also are accessible on the expansion
bus. This feature allows new chips that add to the BIOS to be part of expansion boards
that slide into the computer. The code necessary to control the expansion accessory loads
automatically whenever the system boots up.
Multiple sections of this add-on code fit into any computer, limited only by the address
range available. One complication is that no two sections of code can occupy the same
memory area. As a result, the makers of conventional ISA expansion boards typically
incorporate jumpers, DIP switches, or EEPROM memory on their products to allow you
to reassign the addresses used by their BIOS extensions and avoid conflicts. Plug-and-Play
designs have made this conflict resolution automatic during the setup process.
Initial Program Load

Once the BIOS of your computer has completed all its tests, checked for additional BIOS
code, and set up all the devices it knows about, it yields control over to your operating
183
184
system. In older computers, the operating system simply takes over and brings your computer to life. In a Plug-and-Play system, the operating system continues the initialization
of the devices inside and connected to your computer. In effect, the Plug-and-Play system
completes the BIOS configuration process.
In either case, whether your BIOS completes the initialization of your system or transfers
control to the operating system to complete the process, somehow your operating system
must load. Your computer must know enough about your disksfloppy, hard, and CDROMto be able to find the operating system and get its code running. The process of
getting the operating system started is called the Initial Program Load (IPL).
The IPL procedure originally was a design expedient. When the first computer was created, ROM chips, like all other parts of computers, were expensive. The computers
designers opted to use as little ROM as possible to get the system started. They added a
short code routine at the end of the BIOS test procedure that told the BIOS to read the
first sector from your computers floppy disk (the only drive in the first personal computers). This first sector is sometimes called the boot sector because of its function.
In modern computers, the BIOS is smart enough to look to your hard disk as well as
floppy drive, and sometimes even to your CD-ROM drive, to find a boot sector. Many
BIOSs allow you to select which of these are searched for and in what order. If the BIOS
finds executable code there, it loads that code into memory and executes it. If it finds no
executable code or no available disk, the BIOS complains and does nothing because it
doesnt know how to do anything else.
The small amount of code used by the IPL routine is unforgiving. It searches only the
first physical sector on your disks. It looks for sectors rather than clusters or allocation
units because at this point it doesnt know what operating system you have and how it
organizes disk storage. The BIOS must know how to find the first physical sector on the
disk; consequently it requires a BIOS extension associated with the disk to tell it how to
make the required access. A disk drive that uses only a software device driver will be invisible to the BIOS and the IPL routines, and consequently such a disk cannot boot the system.
The code in the boot sector is also short and to the point because it has to fit into the
confines of a single sector. This code tells your computer how to find the rest of the operating system. Once it locates the main operating system, the boot sector code starts loading that code into memory and yields control to it. The operating system takes command
and completes the process of booting your computer.
Plug-and-Play
One of the gravest shortcomings of legacy computersold computers built in prePentium dayswas that their ISA expansion bus lacked automatic configuration
CHAPTER 7 BIOS
procedures. You were left to adjust settings and configure software yourself. And each of
the dozens of settings you had to make brought its own potential for error. Putting even
one jumper in the wrong place or throwing the wrong switch was often sufficient to turn
two thousand dollars of computer into a noisy desk ornament.
To eliminate such setup problems and shift the tedious job of system configuration to
someone (or something) better equipped to deal with it, the computer industry developed
the Plug-and-Play initiative. The idea was to make your computer responsible for configuring itself. After all, computers specialize in taking over jobs that you find tedious. The
goal was to create computers with no switches, no jumpers, and no headaches. So far,
theyve achieved two out of three, a pretty good average.
The straightforward goal belies the arduous journey in reaching todays reasonably successful automatic configuration systems. Making Plug-and-Play work required changes in
every computers BIOS, expansion bus, expansion boards, and operating system. The universal adoption of Plug-and-Play was a lengthy transition, one dragged on as compliant
products slowly coursed into the market.
Background
The idea behind the Plug-and-Play initiative was to shift responsibility for remembering
and assigning setup options from you to your computer system. After all, your computer
likely has a better memory than you do, and it doesnt mind running through a checkup
procedure however many times as is needed.
The first stab at a Plug-and-Play specification appeared with the original Intel/Microsoft
specification for ISA on May 28, 1993. That effort inspired other companies to join in.
Related standards are being developed to extend Plug-and-Play to other troubling configuration processes, particularly SCSI expansion. Compaq Computer Corporation and
Phoenix Technologies joined Intel to develop a BIOS specification for Plug-and-Play, first
released November 1, 1993.
Toshiba joined Intel and Microsoft to come up with a system to move power-management
responsibilities to the operating system. Because configuring the host system was an
important part of power management, they extended their specification to embrace setting
up all computer accessories, not just power, with one of its foundations being the Plugand-Play specification. They called the result the Advanced Configuration and Power
Interface (ACPI), first published in December 1996. The next major revision of the specification was Revision 2.0, published on July 27, 2000, as a 450-page tome covering the intimate details of the design of new computers. A minor revision on March 31, 2002,
brought the standard up to its current version, Revision 2.0a.
ACPI is neither a hardware nor software specification. Rather, it details the interface
between the two. That makes it not only function like the BIOS, but the specification also
185
186
details how the BIOS works with hardware and software to configure your computer.
Although it is not a BIOS in itself, the specification details how your computers BIOS
must operate. To wear a Windows logo, new computers must conform to the current
ACPI version.
ACPI doesnt replace Plug-and-Play but instead builds upon it. The basic Plug-and-Play
system still takes care of configuring your system, but ACPI gives it new commands and
versatility. The automatic configuration process is still termed Plug-and-Play (or, by those
who have less faith in its reliability, Plug-and-Pray).
For automatic configuration to work properly, it requires all three elements of the system
written to its standardsthe computer and its BIOS, the operating system, and the expansion boards and peripherals attached to the computer. Any older device that does not conform to the various Plug-and-Play standards is considered to be a legacy device. The
Plug-and-Play system attempts to accommodate legacy devices, but it cannot resolve all
conflicts between them. Developers see the ultimate solution as being attrition. As time
goes by, old noncompliant products will fade from the market. To accelerate the demise of
legacy products, Intel and Microsoft have written support for ISA out of their most recent
system standards, although many people still insist on at least one ISA slot in their newest
computers.
Expansion Board Support

The basic Plug-and-Play procedure is a three-step process that lends itself to automation:
First, the Plug-and-Play BIOS directs your system to check what resources each expansion
device needs. Next, the BIOS coordinates the assignments to avoid conflicts. Finally, it
tells your system and software which choices it has made.
The configuration process calls upon specific hardware features of ACPI-compliant
expansion boards. Most importantly, each board is able to deactivate itself so that it does
not respond to the normal control signals inside your computer. The board disconnects
itself from all system resources so that when it is inactive it cannot possibly cause conflicts.
In addition, each ACPI-compliant board has several new onboard registers that are
reached through a standardized set of three I/O port addresses so that the BIOS or operating system can control the configuration of the board. These ports are designated
Address, Write Data, and Read Data.
The Address port functions as a pointer that expands the number of control registers
directly accessible to your system without stealing more system resources. Loading a register number in the Address port makes that register available for reading or writing
through the Write Data and Read Data ports.
The ACPI specification explicitly defines eight card control registers and reserves two
large rangesone of 24 registers for future elaboration of the standard and the other a
CHAPTER 7 BIOS
16-port range that board-makers can assign to their own purposes. In addition, the specification allows cards to be configured as multiple logical devices, and it assigns ports for
their control. The Address port allows the Write Data port to select which of the logical
devices is active and the resources used by each.
ACPI-compliant expansion boards act in one of two ways, depending on whether they are
needed for booting the system. Boards that are required for bootup (that is, display
adapters and disk controllers) start up active. That is, they come online exactly like conventional expansion boards using the resources assigned them as power-on defaults. They
will grab the resources that they need, participate in the normal Power-On Self Test procedure, and let you operate your computer normally. They may also cause the same old
resource allocation problems, as will any conventional expansion boards that dont support
Plug-and-Play or ACPI. The other Plug-and-Play devices (those not needed in booting
your computer) automatically deactivate themselves when your system comes on, waiting
to be told what configuration to use by your operating system.
ACPI-compliant boards not required during bootup normally start up inactive. They do
nothing until specifically activated, typically by the ACPI-compliant operating system.
Every ACPI-compliant board has specific circuitry for managing its configuration. This
circuitry operates independently from the normal functions of the board. Unlike the functional circuits on the board that can be disconnected from the bus interface, the Plug-andPlay circuits always monitor the signals on the bus. However, the Plug-and-Play circuitry
operates in one of four statesWait for Key, Isolation, Configuration, and Sleepwithout regard for whether the functional circuitry is active or inactive.
Boot Operation
All ACPI-compliant boards, whether active or inactive, boot up in their Wait for Key
state, in which they refuse to respond until they receive an explicit command called an initiation key. This key is not a simple password but rather a precisely defined 32-step interaction between the host system and each board. All 32 steps must be completed
successfully before the board can be set up.
Once a board has been singled out with the initiation key, the BIOS can interrogate it
individually and determine whether it is required for booting up the system and what system resources it requires. The BIOS allocates the needed resources to boot devices. If a
board is not required to boot the system, however, the BIOS bows out, leaving resource
assignment to the operating system. The BIOS only needs to get the system going well
enough that the operating system can load and take over.
Plug-and-Play boards on the legacy ISA bus use a different process. They are isolated
using card select numbers, assigned by the BIOS. Once each board is uniquely identified, it
can be automatically configured by the BIOS (if it is required for booting the system) or
by the operating system.
187
188
Onboard ROM
The ACPI configuration process and the ability to isolate individual boards hold advantages beyond automatic resource allocation. The greatly enlarge the potential capacity of
storing firmware on each board.
In prePlug-and-Play days, the BIOS scanned each expansion board for additional BIOS
code. That code had to fit in a tight range of addresses, and the code contained on each
board could not overlap. Using Plug-and-Play technology, the BIOS scans each board in
isolation, so conflicts between boards are not a problem.
The Plug-and-Play BIOS scans for additional code exactly like a prePlug-and-Play
BIOS, looking for the special add-in ROM signaturethe two-byte code 055(hex) followed by 0AA(hex)that indicates a block of add-in BIOS code follows. The firmware on
Plug-and-Play boards adds a pointer that immediately follows this signature to indicate
the location of an expansion header in the BIOS code or a chain of several headers. Each
header is identified by a special four-byte preamble024(hex), 050(hex), 06E(hex), or
050(hex)that corresponds to the ASCII characters $PnP.
The ACPI scheme allows the ROM on each expansion board to hold interface and driver
code for multiple operating systems, each coded by its own expansion header. An operating system can identify the code that applies specifically to it by reading the headers.
The ACPI specification allows each board-maker a generous apportionment of the available memory and port addresses. Each board can use up to four noncontiguous ranges of
memory base addresses for BIOS code and up to eight noncontiguous base addresses for
input/output ports. In addition, a board can use from zero to two separate interrupt levels
and up to two DMA channels. Each manufacturer determines the number of resources
that a given board can use. Which it uses, however, is a matter left to the board-maker.
As this is written, the current ACPI specification is version 2.0a, and it runs over 500
pages long. You can download a copy from a Web site devoted to it, www.acpi.info.
Interface Functions
Although the primary role for the BIOS today is to get your computer started, it retains
the name it earned from its function as the software-to-hardware interface of your
machine. It is a control system that operates at the lowest possible level (the most basic) at
which the input and output of your computer are designed to link to programs. In effect,
it is like a remote control for a television set. The BIOS allows your software to press buttons at a distance from the hardware, just as you control the channel of your TV from
your easy chairand without prying open the set and ratcheting its tuning circuits. In
operation, the BIOS is a universal remote control. It lets you push the same buttons,
regardless of the model or brand of television at which you point it.
CHAPTER 7 BIOS
Interface Functions
The original purpose of this one-step-removed BIOS design was to allow computer hardware to be revised and updated without the need to change software correspondingly. It
helps guarantee the backward compatibility of computers. The extra BIOS step is needed,
because all computers have many hardware elements that are located at specific addresses
in memory or within the range of certain input/output ports. Other computer components may have registers of their own, used in their control, that also are addressed at specific locations. Because of the number of separate components inside any computer, the
potential number of possible variations in the location of these features is limitless.
Software that attempts to control any of this hardware must properly reach out to these
registers. As long as all computers are crafted exactly the same, with the same ports used
for exactly the same hardware with exactly the same registers, no problems should occur.
But if a computer designer wants to change the hardware to a technology that delivers
better performance or greater reliability, he may be stymied by the old addresses. Software
may expect to reach the old design at one set of ports, whereas the new-and-improved
design may be memory-mapped and not use ports at all. In this case, the old software
would not know how to reach the new hardware, and the new design simply would not
work.
The BIOS gives the software a link. The software reaches into the BIOS for the hardware
function it wants. Then the BIOS dips down into the hardware. If the design of a computer system is changed radically, only the BIOS needs to be changed to reflect the new
way of getting at the features of the computer. The changed BIOS still works the same
way as the software, so all older programs run exactly as they did before the change. In
effect, the new system design requires a new route to get to an old destinationa detour.
The new BIOS is an updated roadmap that shows only the detour.
Certainly todays driver software easily accomplishes this function of the BIOS. Back in
the days when the computer was first created, however, there were no drivers. The original computer BIOS didnt even have provisions for adding extra hardware, and the first
operating system had no intrinsic provisions for adding drivers. Not knowing what was to
come of the personal computer, IBM put its reliance on BIOS technology in its first computers and reserved the right to alter the hardware at will. The company made no guarantee that any of the ports or registers of the computer would be the same in any later
computer.
In effect, IBM created the computer in a well-ordered dream world in which programs
would never need to directly address hardware. Instead, each program would call up a
software routine in the BIOS, which would have the addressing part of the instruction
permanently set in its code. Later computers with different hardware arrangements would
use BIOS routines that worked like the old ones and were indistinguishable from the old
ones when used by application software. The addresses inside the routines would be
changed, however, to match the updated hardware. The same software could work, then,
189
190
with a wide variety of hardware designs, giving the designer and manufacturer the flexibility to upgrade the entirety of system hardware, if necessary.
Compatibility
In the days when your computers operating system worked cooperatively with the BIOS
instead of co-opting its function, BIOS compatibility was a major issue in buying a new
computer. A computer had to have a BIOS functionally equivalent to that of an IBM personal computer to work reliably with the software of the day. Consequently, after the IBM
BIOS made its debut, it became the most copied set of software routines in the world.
The computer BIOS laid out all the entry points used by subsequent IBM BIOSs and
most compatible BIOSs as well. It also defined the functions that couldand mustbe
expected in any BIOS as well as established the way that the BIOS works. The BIOS that
IBM developed for its 1984 Personal Computer AT still defines the minimum level of
compatibility for all computers.
Today, computer-makers can buy a compatible BIOS off the shelf from any of several
sources. It wasnt always that easy. Achieving off-the-shelf BIOS compatibility was one of
the great sagas of computer history.
Because the copyright laws forbade any copying of someone elses work, compatible
BIOSs had to be written clean. That is, the programmers were kept from ever viewing
the source code or having any knowledge of the routines it contained. Instead, they
worked from a list of instructions and the functions that the BIOS carries out when given
each specific instruction. In other words, they looked at the BIOS they wanted to copy as
a black box that takes an input and gives an output. The programmers then deduced the
instructions for the inside of the box that would give the desired results.
Working in this way was time consuming and expensive. Few computer companies had
the resources to do it all themselves. Compaq was the first to successfully tackle the job.
The vast majority of compatible computer manufacturers bought the necessary BIOS
firmware from specialist firms. The first to offer compatible BIOSs was Phoenix
Technologies. Now computer-makers also select BIOS chips from American Megatrends,
Inc., Award Software, and Mr. BIOS.
Because each BIOS vendor must develop its own product separately, the exact code used
by each BIOS version is different. Functionally, however, they all look the same to your
software (including your operating system).
At the other side of the interfacewhere the BIOS links to your computers hardware
BIOSs can be quite different. By design, every BIOS is created to match the specific hardware in the computer in which it operates. That is part of its jobuniting different
CHAPTER 7 BIOS
Interface Functions
hardware designs so that they work interchangeably with all software. Every BIOS is customized for the computer it controls. Typically, computer motherboard manufacturers
modify a generic BIOS from one of the BIOS-makers to suit their own purposes.
Because each BIOS is customized for a particular model of computer, no generic BIOS
can hope to work properly in a given computer. BIOSs are not interchangeable.
Moreover, should you want to change or upgrade your computers BIOS for any reason
either through buying new chips or downloading code for Flash memoryyou need to
get one that matches the exact model of computer you own.
Software Interrupts
The BIOS design created for the first computers does its linking through a system of software interrupts. To gain access to the underlying hardware, a program sends out an interrupt, which is a special instruction to the microprocessor. The software interrupt causes
the microprocessor to stop what it is doing and start a new routine. It does this by suspending the execution of the code on which it is working, saving its place, and then executing the program code of the BIOS routine.
The various functions the BIOS carries out are termed interrupt service routines or interrupt
handlers. To call a given interrupt handler, a calling program needs to be able to find the
program code that carries out the function. The design of the computer BIOS allows the
system to put these interrupt handlers at any convenient address. So that a calling program can find the proper routine, the BIOS reserves part of your computers memory for
a map called a BIOS interrupt vector table. This map consists of a long list of 32-bit
addresses, with one address corresponding to each interrupt handler. These addresses
point to the first byte of the program code for carrying out the interrupt and are called
interrupt vectors. The microprocessor reads the value of the vector and starts executing the
code located at the value stored in the vector.
The table of interrupt vectors begins at the very start of the microprocessors memory,
address 00000(hex). Each vector comprises four bytes, and all vectors are stored in
increasing order. Table 7.1 summarizes the principal BIOS interrupt vectors and the associated functions.
TA B L E 7 . 1
BIOS Interrupt Vector Table
Absolute Address
(Hex)
Interrupt
Value
Function
0000:0000
00H
Divide-by-zero interrupt header
0000:0004
01H
Single-step interrupt handler
Hardware
Interrupt
continues
191
192
TA B L E 7 . 1
Continued
Absolute Address
(Hex)
Interrupt
Value
Function
Hardware
Interrupt
0000:0008
02H
Non-maskable interrupt
0000:000C
03H
Breakpoint
0000:0010
04H
Arithmetic overflow handler
0000:0014
05H
Print screen
0000:0018
06H
Reserved
0000:001C
07H
Reserved
0000:0020
08H
Timer interrupt (18.21590/sec)
0000:0024
09H
Keyboard service
0000:0028
0AH
VGA retrace (AT slave)
IRQ2
0000:002C
0BH
Serial port 2
IRQ3
0000:0030
0CH
Serial port 1
IRQ4
0000:0034
0DH
Hard disk
IRQ5
0000:0038
0EH
Floppy disk
IRQ6
0000:003C
0FH
Parallel port
IRQ7
0000:0040
10H
Video services
0000:0044
11H
Equipment check
0000:0048
12H
Memory size check
0000:004C
13H
Floppy and hard disk I/O
0000:0050
14H
RS-232 service
0000:0054
15H
System services
0000:0058
16H
Keyboard
0000:005C
17H
Printer I/O
0000:0060
18H
Basic ROM entry point (startup)
0000:0064
19H
Initial Program Load (IPL)
0000:0068
1AH
Time of day
0000:006C
1BH
Keyboard break
0000:0070
1CH
User timer
0000:0074
1DH
Monitor ROM pointer
CHAPTER 7 BIOS
Interface Functions
Absolute Address
(Hex)
Interrupt
Value
Function
0000:0078
1EH
Disk control table pointer
0000:007C
1FH
Character generator pattern table pointer
0000:0080
20H
DOS terminate program
0000:0084
21H
DOS function calls
0000:0088
22H
DOS terminate address
0000:008C
23H
DOS Ctrl+Break exit address
0000:0090
24H
DOS fatal error exit address
0000:0094
25H
DOS absolute disk read
0000:0098
26H
DOS absolute disk write
0000:009C
27H
DOS Terminate and Stay Resident
0000:00A0
28H
DOS idle loop
0000:00A4
29H
DOS console device raw output handler
0000:00A8
2AH
DOS network communications
0000:00AC
2BH2DH
Reserved
0000:00B8
2EH
DOS execute command
0000:00BC
2FH
DOS print spool control
0000:00C0
30H-31H
DOS internal use
0000:00C8
32H
Reserved
0000:00CC
33H
Mouse driver calls
0000:00D0
34H3EH
Reserved
0000:00FC
3FH
LINK (internal use)
0000:0100
40H
Floppy and hard disk handler
0000:0104
41H
Pointer to hard disk parameters
0000:0108
42H
EGA video vector screen BIOS entry
0000:010C
43H
EGA initialization parameters
0000:0100
44H
EGA graphics character patterns
0000:0114
45H
Reserved
0000:0118
46H
Pointer to second fixed disk parameters
0000:011C
47H
Reserved
Hardware
Interrupt
continues
193
194
TA B L E 7 . 1
Continued
Absolute Address
(Hex)
Interrupt
Value
Function
Hardware
Interrupt
0000:0120
48H
PCjr cordless keyboard
0000:0124
49H
PCjr non-keyboard scan code table
0000:0128
4AH
Alarm routine
0000:012C
4BH4FH
Reserved
0000:0140
50H
Periodic alarm from timer
0000:0144
51H59H
Reserved
0000:0168
5AH
Cluster adapter BIOS-entry address
0000:016C
5BH
Cluster boot
0000:0170
5CH
NetBIOS entry point
0000:0174
5DH5FH
Reserved
0000:0180
60H66H
User program interrupts
0000:019C
67H
Expanded memory manager routines
0000:01A0
68H6BH
Unused
0000:01B0
6CH
System resume vector
0000:01B4
6DH6FH
Unused
0000:01C0
70H
Real-time clock
IRQ8
0000:01C4
71H
LAN adapter (IRQ2 replacement)
IRQ9
0000:01C8
72H
Reserved
IRQ10
0000:01CC
73H
Reserved
IRQ11
0000:01D0
74H
Mouse
IRQ12
0000:01D4
75H
80287 NMI error
IRQ13
0000:01D8
76H
Hard disk controller
IRQ14
0000:01DC
77H
Reserved
IRQ15
0000:01E0
78H7FH
Unused
0000:0200
80H85H
BASIC
0000:0218
86H
NetBIOS
0000:021C
87HF0H
BASIC
0000:03C4
F1HFFH
Reserved for program interrupts
CHAPTER 7 BIOS
Interface Functions
The interrupt vectors are stored in the RAM of your computer so that the values in the
table can be changed. For example, a program or software driver may want to alter or
update a BIOS routine to take advantage of a special feature of new hardware you install
in your computer. The BIOS code itself loads default values for many of these interrupt
vectors into the appropriate RAM locations with the vectors pointing at the routines
stored in your computers ROM.
Sometimes programs add extra routines to a given BIOS function, a process called
chaining interrupts. To chain an interrupt, a program looks at the interrupt vector it wants
to chain to and remembers its value. It then substitutes the starting location of the program code that it wants to add to the old routine. It then modifies its own code to make
the execution of its routines jump to the old interrupt when the processing of the new
code finishes. In this way, the new routine added in the interrupt chain executes first and
then starts the original interrupt handler.
Parameter Passing
Because fewer interrupts are available than tasks you might want the BIOS to handle, different functions are available for many of the interrupts. These separate functions are
identified through a technique called parameter passing. An interrupt needing service from
the BIOSthe calling programpasses a number identifying the command to be carried
out to the BIOS as a parameter, a value held in one or more of the registers at the time the
software interrupt is issued.
The calling program decides what function it wants the BIOS to carry out, loads the parameter value into the proper register, and issues the interrupt. The BIOS code examines
the register to determine what function the programs wants.
This same technique may be used to pass information between the calling program and
the BIOS. The data to be passed along is simply loaded into one register and the parameter identifying the command into another. To pass new data back to the calling program,
the BIOS loads the data into a register, which the calling program can then read.
Using registers to move data into a BIOS routine has its shortcomings. The functions and
parameters are limited to those that are common among computers. Theres little room
for programs to add their own interrupts. Moreover, the scant number of registers in Intel
microprocessors limits the number of parameters that can be passed to and from a function. Most routines use a single byte of data.
To move larger blocks of data, some BIOS routines and most programs use memory. A
calling program stores the values it wants to pass to a subroutine in a block of RAM
addresses. It then passes to the subroutine the location of the data, and the subroutine digs
195
196
into the data, changing what it requires, and leaves the block for later access by the program. Although programs and subroutines identify the block of memory by its address,
programming languages let you use names for the data. The program chooses the block of
addresses to associate with the name and looks to those addresses whenever the programmer uses that name. Similarly, program subroutines may be named, and the program uses
the names to find the block of memory holding the microprocessor instructions that comprise the subroutine.
Entry Points
The various code routines in each BIOS start and end at addresses assigned to the BIOS
function in the computer memory map. The address at which each routine starts is called
that routines entry point. The entry point of a BIOS function is completely different from
the interrupt that calls that function. When the BIOS in your computer sets itself up
either before or during the actual bootupit loads the addresses of the entry points into a
table in memory that becomes the interrupt vectors. In theory, an entry point can be at
any location for any BIOS functionit needs only to be loaded into the slot in the BIOS
interrupt vector table to be properly recognized. Unfortunately, some program writers
decided to call BIOS routines by their entry points instead of using interrupts, because the
direct approach is faster. Consequently, a few applications require that some entry points
be at specific physical BIOS addresses. If the entry point of a BIOS differs from what the
program expects, the result is likely to be a system crash.
IBM has maintained the same entry points with all its BIOSs, and many compatible
BIOSs use exactly the same addresses. A few do not, however. In general, the BIOSs with
varying entry points have been written as programming modules that can be combined in
various ways to suit the needs of a computer designer. What these modular BIOSs add in
flexibility, they lose in compatibility.
Unlike programs that write directly to system hardware, however, programs that require
specific BIOS entry points are rare. With the popularity of compatible computers and
modular BIOSs, they are sure to become more rare. Modern software, nevertheless, is
getting away from a dependency on complete compatibility down to the level of specific
entry points. In fact, many programs avoid the BIOS entirely.
The chief beneficiaries of the BIOS are individual programmers who need to create code
quickly. In many cases, using BIOS routines can simplify the writing of a program.
Certain system operations always are available and can be accessed easily through software. They are reasonably well documented and understood, removing many of the programmers concerns.
CHAPTER 7 BIOS
Interface Functions
Linking to Hardware
Controlling the hardware of your computer requires that your system be able to pass
commands and data to that hardware. For example, if you have a modem, you need to
send the modem commands to dial numbers and change speeds. Once youre communicating, you have to send bytes, even megabytes, of data to the modem and receive more
bytes in return.
The BIOS routines may use memory or I/O ports to pass the commands and data to
hardware devices. When a device uses a block of memory to exchange information, it is
termed memory mapped. When it uses I/O ports, it is termed I/O mapped.
Memory mapping works exactly as described in the previous section on parameter passing.
The BIOS routine and hardware device share a common range of memory addresses they
use for passing bytes back and forth. Programs use BIOS routines instead of physical
addresses to reach memory-mapped devices, giving the hardware designer the freedom to
move the memory window used by a given device. The designer needs only to change the
interrupt handler to match the address he chooses.
I/O mapping uses input/output ports to exchange data. An input/output port of a microprocessor is a special range of addresses that is isolated from main memory and has special
access instructions.
In the Intel scheme of things, microprocessors have a range of 65,536 input/output ports
(compared to over four billion discrete memory addresses). However, due to an aberration
in the design of the first expansion boards for computers, only 1024 are commonly
assigned to devices that plug into standard computer expansion slots. To cut costs of the
first expansion boards, engineers designed boards to decode only 10 of the 16 address
lines.
Although newer buses can take advantage of the full range of input/output ports, expansion boards can quickly consume the available ports, or two boards may attempt to use the
same port. Such conflicts between boards is the most common reason why expansion
boards fail to operate properly. Sometimes installing an expansion board creates a port
conflict that prevents the rest of the system from working at all. Table 7.2 lists some of the
common conflicts that arise in I/O port assignments.
TA B L E 7 . 2
Common I/O Port Assignments and Potential Conflicts
Device
Range Used
Conflict
Color video boards
3D03DF
3C0
COM1
3F83FF
3E0
continues
197
198
TA B L E 7 . 2
Continued
Device
Range Used
Conflict
COM2
2F82FF
2E0
COM3
3E83EF
3E0
COM4
2E82EF
2E0
Expansion unit
210217
200
Floppy controller
3F03F7
3E0
Game controller
20020F
200
LPT1
3BC3BF
3A0
LPT2
37837F
360
LPT3
27827F
260
MDA, monochrome video
3B03BF
3A0
Microsoft bus mouse, primary
23C23F
Microsoft bus mouse, alternate
23823B
Primary bi-sync card
3A03A9
3A0
Secondary bi-sync card
38038C
380
SoundBlaster speaker default
220223
220
SoundBlaster control
388389
VGA
3C03CF
3C0
If youre running Windows on your computer, you can view the actual I/O port assignments used by your computer. To view these assignments, consult the Device Manager by
selecting the System icon in Control Panel. Select your computer on the Device menu
and click the Properties button. From the View Resources menu that pops on your screen,
select the Input/Output (I/O) radio button. Youll see a screen akin Figure 7.1, listing all
the input/output ports in use and the hardware using them.
Through Plug-and-Play and the manual options of the Add Hardware process, Windows
generally will make the proper assignments of I/O ports. If you want to reserve one or
more I/O ports so that devices you install later cannot steal them, click the Reserve
Resources tab and enter the ports you want to reserve.
New expansion buses such as PCI provide access to the full range of I/O ports, thus
reducing the chance of conflicts. Plug-and-Play expansion boards are supposed to automatically resolve port conflicts entirely. Eventually, worries about I/O conflicts should disappear. For most computers, however, eventually has not yet arrived.
CHAPTER 7 BIOS
Storage Functions
FIGURE 7.1
The Windows Device
Manager showing I/O
port usage.
Storage Functions
Because the BIOS comes in the physical form of ROM chips, its storage functions should
not be surprising. For example, the BIOS stores its own name and the date it was written
inside its own code. But the BIOS also incorporates storage functions that go beyond the
bytes encapsulated in its ROM silicon. Although physically separate from the BIOS chips
themselves, the setup memory of your computer is controlled by BIOS functions and is
often considered withand even as part ofthe BIOS. This memory records vital details
about your computer hardware so that you dont have to set up each expansion board and
disk drive every time you switch on your computer. In addition, the BIOS tells your computer to reserve several small blocks of memory for dynamic data that your computer and
its operating system use for tracking several system functions.
Legacy Data
As with any computer program, the BIOS code uses data for its operations and produces
data as its result. The BIOS stores some of this data to be used as a reference by your programs. The BIOS acquires some of this information as it sets up the system. By storing it
all in one place, the BIOS eliminates the need for every program to waste its time looking
up common features. Other data is encoded as part of the BIOS itself so that programs
can determine what kind of computerand what kind of BIOSthey have to work with.
BIOS Data Area

During the initialization process, the BIOS searches through the system for specific features. It checks the number and kinds of ports installed, the type of display adapter
(monochrome or color), and other options. Included among the data that it stores are
199
200
equipment flags, the base addresses of input/output adapters, keyboard characters, and
operating modes. The BIOS then stores this information in specific memory locations so
that your programs can check to see which features are available or being used. For example, the BIOS looks for serial ports at specific addresses (see Chapter 11, Ports). As it
finds them, the BIOS records the base address that each serial port uses.
All this self-descriptive information about your system is stored in a special part of RAM
called the BIOS data area. Located just above the interrupt vectors, the BIOS data area
comprises 256 bytes of memory, starting at absolute memory location 000400(hex) and
running to 0004FF(hex).
Date
Nearly every BIOS identifies itself with a copyright message and (usually) a version number so that you can determine who made it. (More specifically, the copyright message protects the writers of the BIOS from other people copying their work.) This identification
information may appear anywhere within the BIOS code range.
In addition, the revision date of your BIOS helps you identify how recently its code was
updated. As computers have expanded their capabilities, BIOSs have been revised to
enable new operations. Sometimes older BIOSs will not work with new peripherals. For
example, many BIOSs from before 1998 dont let you use all the space on hard disks
larger than 8GB. Older BIOSs may impose even more severe limits.
You can determine the date of your BIOS in several ways. The most convenient include
catching the bootup screen of your computer, which usually displays the date and revision
number of the BIOS, and using a diagnostic utility. Microsoft Diagnostics, supplied with
most versions of Windows, will show you the detail of your computers BIOSits date
and manufacturer. This program is located in the Command subfolder of your Windows
folder as MSD.EXE. You can either click its icon or run it from the MS-DOS prompt by
typing MSD and then pressing Enter. Click the Computer button, and it will list some of
the BIOS details of your computer, as shown in Figure 7.2.
Note that Microsoft has not updated the diagnostic program for years, so the most recent
microprocessor it will recognize is a 486DX! The System Information program found
under System Tools in Accessories from the Programs option at the Start button is
Microsofts preferred tool. It shows your system in exquisite, even painful, detail, but it
offers no direct display of your systems BIOS date.
System Identification Bytes

The original computer BIOS included a byte of data to identify the kind of computer in
which the BIOS was installed. Programs that might react differently depending on the
processor or other aspects of the computer could use this byte to identify the hardware
CHAPTER 7 BIOS
Storage Functions
and adjust themselves accordingly. At least that was the idea of the IBM engineers who
came up with it in 1981.
FIGURE 7.2
Microsoft Diagnostics
showing system
details.
The scheme had its problems, however. Allowing for a single byte soon seemed shortsighted, so this memory area was expanded to two bytes, starting with the IBM XT Model
286, introduced in 1985. The overall storage area was termed the system identification bytes,
and in IBM nomenclature, it was divided into the model byte and the submodel byte. A third
byte was reserved to indicate any major revision to the system.
IBM created an elaborate table of designations for the various computer models it offered.
Unfortunately, the company made no provision to coordinate the use of this storage area
for other manufacturers. Consequently, the rest of the industry used designations marking
their IBM equivalent. Because all systems since 1984 follow the architecture of the IBM
AT system, nearly all computers now define themselves as using the AT code. In other
words, the system identification bytes store essentially meaningless data.
The model byte is located at absolute memory address 0FFFFE(hex) and is FC(hex) in
nearly all current computers. The submodel byte follows it with a value of 01(hex). The
revision byte follows next and is usually zero.
The BIOS also has its own individual identification encoded into its bytes. Although it
may be nothing more than a copyright notice, each BIOS has some code identifying its
origins. Most BIOS makes have their own codes to identify the version of the BIOS,
which revision it is, and even the model of computer for which it is meant.
Disk Parameter Tables

One of the more important entries in the BIOS data area is the disk parameter table. Most
computers store the parameters used by user-defined hard disks, typically reflected as hard
disk type 47 in the setup program, in the BIOS data area. The most common memory
201
202
location for this information is absolute address 0000:0300. Some BIOSs give you the
option of locating the disk parameter table in an alternate location, typically DOS memory. In any case, interrupt vector 41(hex) points to the location of the disk parameter table
for the first hard disk in a computer. Interrupt vector 46(hex) points to the location of the
parameters of the second hard disk.
In general, the best location for your disk parameter table is absolute address 0000:0300.
However, some peripherals, such as older sound cards and network adapters, use this
address for their own purposes. Conflicts between the peripheral and your hard disk may
render either or both inoperable. Relocating the disk parameter table through a BIOS setting solves the problem.
You may encounter two kinds of disk parameter tables. A standard fixed disk parameter table
(FDPT) contains a single description of the vital parameters of the associated hard disk,
listing the number of cylinders, heads, and sectors used by the disk. An enhanced fixed disk
parameter table (EDPT) has two entries: the logical cylinder-head-sector values and the
physical cylinder-head-sector values. The logical values reflect the way your computer
sees and accesses the disk. The physical values reflect the actual number of cylinders,
heads, and sectors used by the disk.
The standard FDPT is sufficient only for smaller disks, those with both fewer than 1024
cylinders and a capacity less than 504MB. Due to the design of the standard FDPT, only
ten bits are allowed for storing the number of cylinders, so numbers larger than 1024 cannot be encoded. The EDPT defaults to 16 heads for all disks, allowing four more bits for
storing the number of cylinders.
Your computer deals with any disk by the logical values in the EDPT. Programmers check
the value stored here using software interrupt 13(hex).
DOS Tables
DOS stores many of the parameters it uses for managing your computer and its programs
in a series of tables. To achieve the utmost in compatibility, DOS locates these tables at
the lowest available reaches of conventional memory. Among the data stored here are file
handle identifications (FILES), the last referenced disk data (BUFFERS), file control
information (FCBS), drive tables (LASTDRIVE), and system operating information
(STACK).
Most memory managers allow you to relocate this data into XMS memory to free up
more space in the conventional area. Although this technique is commonplace and well
tested, you will sometimes encounter compatibility problems, such as a crash of your system. For example, early incarnations of most network operating systems expect to find
this data in conventional memory (along with COMMAND.COM itself). The only way to avoid
this problem is to avoid the network, avoid DOS, or refrain from relocating this data with
CHAPTER 7 BIOS
Storage Functions
your memory managerthe last usually being the most viable option. Youll want to avoid
relocating any of these DOS tables if your system exhibits compatibility problems or its
network redirector behaves oddly.
ACPI Tables
One of the chief responsibilities of an ACPI-compliant BIOS is to set up the ACPI tables
that describe the interfaces to the hardware installed in the system. The operating system
can then read the tables to determine not only the resources required by a specific board
but also the code with initialization routines or arbitrary operation sequences needed to
make the hardware function. The ACPI tables may contain not only data but also program code in p-code, the programming code or the language of the ACPI system.
The ACPI specification describes a programming language, called the ACPI Source
Language (ASL), in which engineers can write their code for their device interfaces. ASL is
then compiled into ACPI Machine Language (AML), the object code that is actually stored
in the ACPI tables. The ACPI-compatible operating system reads the AML as a set of
program instructions.
The BIOS sets up the tables during the boot process. It reads the data for each table from
the devices it scans and then stores each table separately in system memory. It makes one
master table, called the root system description table, which holds the location of each of the
separate data tables.
The BIOS itself does nothing with the data it puts in these tables. It is only an agent that
moves the data from device to memory. The ACPI-compliant operating system uses the
data or p-code contained in the tables.
To enable the operating system to find the root system description table, the BIOS plants
a pointer in the Extended BIOS data area of memory. The operating system identifies this
root system description pointer by looking for the character sequence RSD PTR (including
the trailing space) in the Extended BIOS data area. The pointer stores the base address of
the root system description table from which the operating system can find the rest of the
tables.
System Configuration
In order to properly test a computer, the BIOS needs to know exactly what it is testing
what peripherals are installed, how much memory it must look through, and whether you
have installed a coprocessor. In order to boot a computer, the BIOS needs to know exactly
what kind of disk is connected to it. In order for you to see what is going on when your
system boots up, the BIOS needs to know what kind of display system you have.
In some cases, the BIOS code itself can be written to search out and find the vital information it needs to get your system going. Such a search is not always accurate, nor is it
203
204
easy to write the proper search method into a few kilobytes of BIOS code. Even if the
proper program could be written and packed into ROM, you probably do not want to sit
around and waitand waitwhile the BIOS probes into every nook and cranny to see if
something is there.
To let the BIOS (and the rest of the computer) know what options are installed in a given
system, all computers record vital setup information that can be referenced quickly. The
storage system for this datathat is, system setup memoryhas one requirement: It must
be nonvolatile. Other than that, the storage format and method are flexible because the
BIOS isolates them from the rest of your software. The BIOS looks up this system data
and transfers it to the BIOS data area for reference by your programs.
This flexibility has given designers freedom to use a number of storage technologies for
this setup information. Among the most popular have been physical memory (switches),
nonvolatile electronic memory, and magnetic (disk) memory. Since the AT was introduced
a decade and a half ago, however, the basic form of this setup memory used by most computers has been the samea few bytes of CMOS memory kept fresh (that is, continuously
operating) by battery power.
CMOS
When the world and computers were young, all the differences between computers could
be coded by one or two banks of DIP switches. But as the options began to pile up, the
switch proved to be more a problem than a panacea. A reasonable number of switches
couldnt allow for the number of options possible in a modern computer. Another problem with switches is that they are prone to mechanical problemsboth of their own making and otherwise. Switch contacts naturally go bad, and they can be helped along the
path of their own destruction by people who attempt to adjust them with pencils (the
graphite that scrapes off the point is conductive and can short out the switches). People
often set switches wrong and wonder what is awry.
Background
IBM developed a better scheme for the AT. Vital system parameters would be stored in a
special, small block of battery-backed CMOS memory, a total of 64 bytes, in the form of a
special chip, a Motorola MC146818, which also held a real-time clock. The lower 14 of
those bytes are used by the real-time clock to hold the current time and an alarm time,
leaving 40 bytes for storage of setup information. Locations were assigned for storing
information about floppy and hard disks, the presence of a coprocessor, the amount of
installed memory, and the type of display system (monochrome or color). Because CMOS
memory is volatile, every system has a battery of some kind to keep this memory fresh and
the real-time clock running. (See Chapter 31, Power.)
CHAPTER 7 BIOS
Storage Functions
Most computers followed the basic IBM scheme, minimally augmenting it to reflect the
increasing capabilities of their systems. Although the MC146818 chip rarely appears in
computers anymore, its functions are carried over into new support chips. For example,
clock modules from Dallas Semiconductor incorporate all the functions of the Motorola
chip, a built-in lithium battery, and more. The only major change in CMOS storage that
has been made is the almost universal adoption (under the aegis of Microsoft) of the Plugand-Play system.
Basic Memory Assignments
The basic CMOS data remains the same regardless of the manufacturer of the computer,
and the various bytes of data are stored in the same rigidly structured format. Table 7.3
summarizes the basic storage assignments that are common across nearly all modern computers.
TA B L E 7 . 3
CMOS Memory Byte Assignments
Byte
Assignment
Purpose
00
Seconds
Real-time clock
01
Second alarm
Real-time clock
02
Minutes
Real-time clock
03
Minute Alarm
Real-time clock
04
Hours
Real-time clock
05
Hours Alarm
Real-time clock
06
Day of the week
Real-time clock
07
Date of the month
Real-time clock
08
Month
Real-time clock
09
Year
Real-time clock
10
Status Register A
Real-time clock
11
Status Register B
Real-time clock
12
Status Register C
Real-time clock
13
Status Register D
Real-time clock
0A
Status Register A
0B
Status Register B
0C
Status Register C
continues
205
206
TA B L E 7 . 3
Continued
Byte
Assignment
0D
Status Register D
0F
Shutdown Status
10
Diskette Drive Type
11
Reserved
12
Hard Disk Type
13
Reserved
14
Equipment Type
15
Low Base Memory
16
High Base Memory
17
Low Expansion Memory
18
High Expansion Memory
19
Reserved
1A
Reserved
1B
Reserved
1C
Reserved
1D
Reserved
1E
Reserved
20
Reserved
21
Reserved
22
Reserved
23
Reserved
24
Reserved
25
Reserved
26
Reserved
27
Reserved
28
Reserved
29
Reserved
2A
Reserved
2B
Reserved
Purpose
CHAPTER 7 BIOS
Storage Functions
Byte
Assignment
2C
Reserved
2D
CMOS checksum
2E
CMOS checksum
30
Low expansion memory
31
High expansion memory
32
Date century
33
Information flags
34
Reserved
35
Reserved
36
Reserved
37
Reserved
38
Reserved
39
Reserved
3A
Reserved
3B
Reserved
3C
Reserved
3D
Reserved
3E
Reserved
3F
Reserved
Purpose
Although the CMOS that holds this data is conventional memory, it is not in the direct
reach of your computers microprocessor. Unlike normal system memory, this CMOS
setup memory was I/O mapped. That is, its contents were accessed through two
input/output ports. Port 070(hex) indicates the memory byte you want to access, and port
071(hex) provides the pathway to the indicated byte. Reading or writing a byte of CMOS
setup memory requires two steps. First, you write to port 070(hex) with the byte location
in the CMOS range you want to read or write. Reading port 071(hex) tells you the value
stored at the location you have chosen. Writing to port 071(hex) changes the byte value at
the appointed location.
The contents of most of the storage locations in this CMOS setup memory are monitored
by storing a checksum in bytes 02E and 02F(hex). If the checksum does not agree with the
modular total of the monitored bytes, your system reports a CMOS memory error. The
207
208
diagnostic status byte, 00E(hex), indicates the gross nature of the error and additionally
reports whether battery power has failed. If all systems are go, this byte has a zero value;
any bit set indicates a specific error (with two bits reserved).
Newer computers have elaborated on this CMOS storage scheme, adding more bytes to
hold the status of other system features. In function and operation, however, all follow the
pattern set by the AT.
Resetting CMOS
The chief reason youre advised not to tinker with the advanced settings of your computers BIOS is that the control they afford allows you to set your computer up so that it
does not work. For example, if you err by setting too few wait states for the memory you
have installed in your computer, you may encounter a memory error as soon as your computer switches on and checks its BIOS settings. You wont have an opportunity to jump
back into setup to fix the problem. Youll be stuck with a dead computer and have no way
of bringing it back to life.
Well, not quite. Most computer BIOSs have factory defaults that are set conservatively
enough that the machine will operate with whatever youve installed. The only trick is to
restore the factory defaults.
Switch or Jumper Reset
Many computers have jumpers or a DIP switch that forces the reset. Check the documentation of your computer or motherboard to see if this option is available to you. Typically
youll find this information using the index of your instruction manual and looking under
factory defaults or BIOS defaults.
The exact procedure usually takes one of two forms, depending on the design of your
computer. The easiest requires only that you move a jumper or slide a DIP switch and
then turn your computer back on. The position of the switch or jumper doesnt matter.
All that counts is that you move it. The most complicated procedure isnt much more difficult. You move the jumper or slide the switch, wait a few minutes, and then move it
back. The delay allows the tiny amount of power thats locked in the CMOS circuitry to
drain away.
Power Deprivation
The alternate procedure for resetting your CMOS works for nearly every computer and is
based on the same power-depravation principle. You only need to deprive your computers
CMOS of its battery power so that the contents of its memory evaporate. Exactly how to
deprive your CMOS of its lifeblood electricity depends on the design of your system.
If your computers motherboard uses an external battery, simply unplug it from the motherboard. If your computer uses a replaceable disc-style battery in a matching battery
CHAPTER 7 BIOS
Storage Functions
holder, pop the battery out. In either case, allow ten minutes for the residual power in the
CMOS to drain away before reconnecting or reinstalling the battery.
Some motherboards use permanently installed rechargeable nickel-cadmium batteries
instead of replaceable cells. Usually these computers have some provision for electrically
disconnecting the battery power from the CMOS (a jumper or switch, as noted in the preceding section). A few computer motherboards make no provision for disconnecting their
nickel-cadmium batteries (also known as ni-cads). If you have such a computer and have
put your BIOS in a nonfunctional configuration, as a last resort you can sometimes force a
reset to the factory defaults by discharging the battery. The battery will recharge the next
time you operate your computer.
Never short out a nickel-cadmium battery to discharge it. The low resistance of ni-cad
cells produces high currents (even with small batteries), sufficient to melt circuit board
traces and even set your computer on fire. Instead of shorting out the battery, discharge it
through a resistor. A half-watt, 39-ohm resistor will safely discharge a ni-cad cell in about
half an hour without danger to you or your computer. Alternately, you can use a six-volt
lamp, such as one designed for battery-powered lanterns, as a battery load. The lamp will
show you the progress of the discharge, glowing brightly at first and dimming as the batterys charge gets drained away. Connect either the resistor or the lamp directly between
the terminals of the battery using clip leads.
Advanced Setup
After you press the right combination of keys to bring up your computers setup menu,
you are usually allowed to jump to Advanced Setup. This extension of Setup allows you to
alter vital system operating parameters that are controlled by the motherboard chipset.
The Advanced Setup procedure alters additional, often proprietary, storage in CMOS and,
sometimes, in other parts of your computer. Typically, Advanced Setup is a second menu
(or series of menus) accessed through the initial Setup menu.
Altering the options can have a dramatic effect on the performance of your computer.
Importune tinkering often can lead to having a machine that will not work. Performance
becomes zero. On the other hand, you can sometimes optimize your system to take best
advantage of the options youve installed (for example, to coax the best possible performance from the memory system or highest transfer rate from your expansion bus).
The range of features controlled by advanced setup varies with both the BIOS and the
chipset inside your computer. Most chipsets give you a wide range of options to wade
through. However, one of the most popular, Intels Triton chipset, provides very limited
control range. For example, the only way to alter the ISA bus speed is to change jumper
or DIP switch settings, and you cannot alter cache timing at all.
209
210
Some of the options available with different chipsets and BIOSs are detailed in the following subsections.
Parity Check
Some systems permit you to switch off memory parity checking, thus disabling error
detection. Taking this option prevents your system from halting when memory parity
errors are detected. You are well advised to leave parity checking enabled except when
attempting to find elusive memory problems. Because this option is meaningful only for
computers that have parity-checked memory, it offers you no savings as would a system
that used 8-bit SIMMs instead of parity-checked 9-bit SIMMs. Your system is likely to
halt when a parity-error occurs in program code, but the crash is less controlled than an
error-warning message. Files may be destroyed. If the error occurs within data, you may
never know when your information is inaccurate.
Memory Testing
Testing the prodigious amounts of RAM in a modern computer can take a long time,
indeed, adding a minute or more to the cold boot sequence. To minimize the wear and
tear on your patience, many BIOSs allow you to defeat the entire memory test sequence
or specific portions of it. With the claims of extreme reliability for todays memory systems, you may want to consider switching off your computers memory test.
Commonly youll find an option called Above 1MB Memory Check that allows you to
switch off memory testing beyond the base megabyte in your computer. Selecting this
option will cut your bootup time from a few minutes to a few dozen seconds. Checking
the base megabyte of memory in your computer usually is not an option because your
computer needs error-free real-mode memory for running the setup program.
Some systems give you the option of adding an audible click or beep each time a memory
block of a megabyte or so is tested. The sound is simply reassurance that your computer is
successfully coursing through its lengthy memory test. Some people find this sound
obnoxious rather than reassuring, and the Memory Test Click Sound setup option allows
them to silence their systems.
Numeric Processor Testing
The BIOSs of some older computers offer the option of defeating the numeric processor
test. Although you might be tempted to forgo this test to accelerate the boot process, you
may be surprised at its effect. Many BIOSs assume that if you tell them not to test the
numeric processor, you dont have one. They set a flag that tells your applications you
dont have a coprocessor installed, so your software wont use the coprocessor even if one
is installedeven as part of your microprocessor.
CHAPTER 7 BIOS
Storage Functions
Computers based on the 386 and 486 microprocessors may also give you the option of
selecting a Weitek coprocessor. You must select this option to enable your software to take
advantage of the Weitek chip.
Cache Operation
Some systems require that you set the size of secondary (external) memory cache you have
installed. Youll want to change this setting only if you install additional SRAM chips to
increase the size of your computers memory cache.
Some BIOSs allow you to switch on or off the cache. Some allow the individual control of
the internal (primary) cache inside your microprocessor and the external (secondary)
cache. The only time you should switch off your systems caches is when you want to pin
down software problems or diagnose hardware errors.
Wait States
This setting controls the number of wait states injected during memory accesses.
Typically, you have a choice of zero, one, two, or possibly three wait states. Some systems
allow the separate setting of read and write wait states. Choosing fewer wait states makes
your computer faster, but choosing too few to accommodate the speed of your systems
memory leads to memory errors. Set this value too low and your computer may not boot
at all. If you are a die-hard tinkerer and want to explore the potentials of this setting,
adjust it downward one step at a time. Then run your computer for a while to check its
reliability. Toying with this may also familiarize you with the location of the CMOS reset
jumper.
Bus Clock
Many ISA systems allow you to adjust the clock speed of their expansion buses so that you
can eke the most performance from your old expansion boards. Some systems give you a
choice of clock speeds in megahertz; others express the speed in terms of the microprocessor clock (for example, CLOCKIN/4 implies one-quarter the microprocessor clock
speed8.25MHz with a 33MHz system). Higher speeds (lower divisors) can deliver more
performance, but rates above about 12MHz may sacrifice reliability. In VL Bus and PCI
systems, the bus clock setting has less effect on overall performance because the devices
that need the most bus bandwidth (that is, a faster bus speed) will be connected to the
higher-speed local bus. The clock speed of the local bus is not affected by the bus clock
setting.
ROM Shadowing
Manufacturers may provide any of a number of options to enable or disable the shadowing of ROM code in fast RAM memory. Some merely allow the simple enable/disable
choice. Others allow you to control ROM elements separately. One variety of BIOS lets
you individually enable/disable system ROM and expansion ROM. System ROM means
211
212
the BIOS code on your motherboard; expansion ROM is the code on expansion boards,
which typically includes your video card and hard disk host adapter. Another variety of
BIOS gives the choice of system, video, or adapter. A fourth BIOS type lets you choose
whether to shadow ROM memory by address range, letting you select enabling or disabling by 16K, 32K, or 64K blocks.
In general, you should shadow only those memory ranges actually in use by ROM.
Shadowing unused ranges wastes memory. Shadowing memory ranges used for buffers by
network adapters or ancient expanded memory (EMS) boards may actually degrade system
performance or cause compatibility problems.
Remember, shadowing only helps the performance of applications that use BIOS code in
ROM, which for the most part means only DOS. If you primarily run any version of
Windows or OS/2 that makes little use of the BIOS, shadowing yields no performance
improvement and may cause compatibility problems. You will probably want to switch off
shadowing when using any modern operating system.
Experimenting with shadowing can also lead to a system that does not boot. It is best to
enable one block or shadowing feature at a time and then run your system to observe the
result.
Concurrent Refresh
Most early computers must devote 10 percent or more of their active time to refreshing
memory. Newer systems are able to refresh memory concurrently (that is, while they are
performing normal tasks). Concurrent refreshing ekes more performance from your system, so it is the preferred operating mode. Some BIOS-makers call this kind of memory
refresh hidden refresh. Normally your computer will deliver better performance with concurrent refresh. Some older RAM designs (used mostly in 386 and older computers) cannot handle concurrent refreshing and may cause memory errors when this feature is
enabled.
Page Interleave
Some systems with two or four identical banks of RAM chips or memory modules can
operate in an interleaved fashion (that is, alternating banks in back-to-back memory
requests). Statistically, interleaving can cut wait states by nearly half. Although the effect is
less pronounced in modern cached systems, enabling page interleave (when supported
with sufficient memory) can improve performance.
Page Mode
Page mode memory chips and modules also can trim wait states. Page mode SIMMs can
make repeated accesses to one memory page without wait states. If you equip your system
with page mode memory and enable this option, you should get better performance.
CHAPTER 7 BIOS
Storage Functions
Virus Protection
With all the concern in the industry about computer viruses, some BIOS-makers have
added their own form of protection, a warning message that appears when software
attempts to write to the boot sector of your hard disk. When the protection is switched
on, youre given the option of canceling the write operation to prevent infection. Because
some operating systems, such as the OS/2 Boot Manager, rewrite boot sector data when
you switch between boot modes, you cant simply obstruct all boot sector write operations.
The protection afforded by this system is limited in two ways. It potentially stops only
boot sector viruses. It does not stop those that afflict the active operating system files.
Also, it protects only hard disks controlled through the computer BIOS. SCSI disks that
rely on separate BIOS code in your SCSI host adapter are not protected.
Typematic Rate
Many BIOSs allow you to alter the way the keyboard reacts to your keystrokes, repeating
them if you hold a key down. This feature, called typematic, is fully discussed in Chapter
20, Keyboards.
Many BIOSs have an overall toggle, typically termed Typematic Programming, that enables
or disables the settings for the typematic feature. You must enable typematic programming to alter other typematic settings. In addition, your keyboard must allow its typematic feature to be programmed. You can control the typematic feature through Windows
95 and 98 no matter what kind of keyboard you have or the BIOS typematic settings.
If a BIOS and keyboard support typematic programming, you can generally alter two
aspects of this feature. Typematic Rate Delay specifies how long you must hold down a
key before the typematic feature kicks in and characters start repeating. This setting is
generally specified in milliseconds. Typematic Rate controls how quickly the characters
repeat after the initial delay. This value is usually specified as a number of characters per
second that will appear on your screen and in your text.
Num Lock
Some BIOSs let you configure the toggle state of some keyboard keys. The most common
option is the setting of the Num Lock key. Most systems default to booting up with Num
Lock switched off. This option lets you have your system boot with Num Lock on so you
can go right to work crunching numbers. This option is an aid primarily for systems with
old 84-key keyboards that lack a separate cursor pad. MS-DOS 6.0 and later enable you to
accomplish the same effect as this option by adding the line NUMLOCK=ON to your
CONFIG.SYS file.
213
214
Boot Device or Sequence

By design, computers normally try to boot from the floppy disk first, and only if they find
no disk in the floppy drive do they attempt to boot from the hard disk. This strategy helps
you get going when something happens to your hard disk so that its no longer bootable.
On the downside, it slows the boot process and forces you to wait for the buzz of the first
floppy drive access. It also requires you keep disks out of your floppy drive until after your
computer boots up.
Although neither of these shortcomings is fatal, they can be bothersome. Consequently,
some BIOS-makers give you the option of setting the order that the BIOS searches for a
bootable disk drive. Depending on the computer and BIOS-maker, several options are
possible.
Some BIOS-makers enable you to directly specify the boot order of your computers disk
drives. Typically your options will be listed as A:, C: or C:, A:. Sometimes only the
first drive letter will be listed. Specify A: to make your system attempt first to boot from
floppy or C: to skip the floppy and try your hard disk first.
Other BIOSs have the option Floppy Seek at Bootup, which lets you toggle between Yes
or No. If you select No, the BIOS will ignore the floppy disk drive when attempting to
boot your computer, even if it has a valid system disk in it. You can take advantage of this
option to prevent inexperienced users from booting your computer from their own, possibly virus-infected, floppy disks.
Recent systems also give you the option Auto Boot Sequence, or Smart Boot. This is an
attempt at the best of both worlds. Most of the time the intelligence of this technique is
limited to sampling the floppy disk drive to determine whether it contains a disk, booting
from it if the disk is a system disk, and ignoring the drive and booting from hard disk if
the disk in the floppy drive is not a system disk or no disk is there at all.
Passwords
Many BIOSs have built-in password protection so that you can limit who uses your computer. The setup program typically gives you the option to enable or disable password
protection. When you enable protection, the system stores your password in its CMOS
memory. If the battery that backs up your CMOS gets low, youll lose password protection
until you replace the battery.
Tuning for Performance

Extra speed is hidden in each of the major components of your computerthe microprocessor, the memory, and your chipset. It takes the form of component tolerances and
safety factors, a little bit more than the ratings that every electronic part is good for. By
pushing a little harder, you can gain more power from your computer. Although you
CHAPTER 7 BIOS
Storage Functions
wont double your systems overall speed, you could gain several percent, enough to give
you an edge in game-playing or quicken your daily grind.
Fine-tuning your BIOS is the key to unlocking the extra speed potential of your computer. A suitable BIOS will allow you to tweak the configuration of your motherboard
chipset and alter microprocessor and memory timing. Each adjustment promises an extra
dollop of speed.
Unfortunately, the tenability of BIOSs varies widely. Large computer companies
Compaq, Dell, Hewlett-Packard, IBM, Sony, and Toshibagive you control only over a
handful of setup parameters that enable you to customize security, add a few peripherals,
and configure ports. They dont let you try to eke more performance from your computer
because, well, they figure youre likely to screw up. Set the wrong value for a vital parameter, and your computer will stop working and the manufacturer will get a support call
and may have to send out a technician to get your machine running again. By not giving
you such intimate control, the manufacturer saves on support.
Smaller computer-makerssay, the one operating in the garage next doorget a competitive edge by using motherboards with tunable BIOSs. They can tweak in more performance than the bigger computer-makers and give you a faster system. Most makers of
motherboards not affiliated with one of the big computer companies have learned that
they gain a competitive edge with power users by giving them complete setup flexibility.
The pioneer among these companies was Abit, which developed its extremely tweakable
SoftBIOS in 1994. Through it and its successors, you can adjust nearly every controllable
parameter of your motherboards chipset. Not only can you adjust the clock speed of the
microprocessor to edge up performance by overclocking, you can shave cycles off the
inevitable waits for memory refresh and access. With some systems, you can even adjust
the voltage supplied to the microprocessor, monitor the environment inside your computer, and set limits on its operation.
Most computer-makersand the major brands in particularare conservative in the
demands they make on computer components, such as the microprocessor and memory.
They prefer to err on the cautious side because it minimizes the need for support.
Moreover, they make no allowances for the differences between individual components.
You may be lucky and get memory thats rated at 133MHz but will function fine at
150MHz. No motherboard-maker wants to take on the burden of testing the limits of
each component when preparing a product. But you can do the testing yourself and set
your computer to take advantage of the full potential of its major components.
Realistically speaking, the major manufacturers caution is justifiable. The gains from finetuning your BIOS setup are modest, typically a few percentage points. If youre the slightest bit greedy and push your chipset too hard, you can render your computer completely
unusable and even destroy the data stored on your hard disk. Go too far, and you can even
215
216
kiss off an investment in $1000 silicon when your microprocessor melts down into worthless sand.
Play it safe. Backup anything thats important on your system before you begin.
Familiarize yourself with your motherboard and locate its CMOS reset (if it has one).
Also, check for jumper or switch settings for the basic microprocessor and bus operating
parameters. Some BIOSs, although they allege to give you complete control through the
setup, actually require hardware settings of some vital parameters. Your BIOS may or may
not be able to override these hardware settings. If in the process of trying to set the chip
speed through your BIOS your system does not cooperate, double-check your motherboard manual for hardware adjustments.
Clock Speeds
The obvious way to eke more speed from a system is to up the microprocessor speed, an
enhancement usually termed overclocking because it requires operating the chip over and
beyond the clock speed rating set by its manufacturer. Kick up the microprocessors clock,
and you can expect a corresponding linear increase in performance.
Raising memory speed also influences the overall performance of your system. The effect
can be dramatic because memory performance has not kept in step with microprocessor
speed. All modern computer memory systems operate at a fraction of microprocessor
speed, requiring the microprocessor to slow down or wait for memory access. On-chip
memory caching reduces, but does not eliminate, the waiting.
Optimizing the performance of your computer requires considering both the speed at
which you operate the microprocessor and the speed at which you operate memory. Not
only do you need to consider adjusting both, but in most systems you will also be forced
to adjust both. The two speeds interact, and their control is often linked.
Most BIOS setup systems do not give you the option of simply changing the microprocessor clock speed because of the design of the chips themselves. Most microprocessors
accept a single external clock frequency, which controls the operation of the front side bus
or memory bus. The microprocessor speed is derived from this value through a multiplier.
Consequently, if you do not vary the multiplier, microprocessor and memory speed settings interact. Up the memory speed and youll raise the microprocessor speed.
The multiplier guarantees that the jump in microprocessor speed will be substantially
higher than the memory speed increase (it multiplies the effectwhat did you expect?),
and you may soon run past the microprocessors inherent speed limit.
Depending on the actual memory modules installed in your computer, your systems
memory system may be more or less tolerant of speed adjustments, as is your microprocessor, even after allowing for the effect of the multiplier. Consequently, you may need
CHAPTER 7 BIOS
Storage Functions
to adjust both the clock speed setting and multiplier setting to achieve optimum performance from your system.
A further clocking issue is that the design of some motherboards links the PCI expansion
bus clock to the front side bus (memory) clock. Although the PCI bus is designed to allow
asynchronous operation (it need not be linked to memory speed), many systems do forge
such a connection. This linkup can be problematic because PCI devices can be the most
speed sensitive in your system. Typical expansion boardsmodems, network interface
cards (NICs), and SCSI disk controllersoften develop problems when you raise the PCI
bus from its nominal 33MHz speed past about 37MHz.
Cycle Shaving
In addition to adjusting the clock speed of memory, many BIOSs allow you to adjust intimate aspects of memory timing (for example, the number of clock cycles required for the
operation of the memory system). Most of the available settings control the number of
cycles of delay between memory control signals. The opportunities for fine-tuning are
amazingyou could spend weeks testing the potential combinations of adjustments.
Such attempts will likely be fraught with problems, however. Most adjustments will yield
nonworking systems. Moreover, many of them interact, and most make no discernable difference in system performance.
The most worthwhile memory parameter to experiment with is the Column Address
Strobe latency, or CAS latency. This setting adjusts the number of clock cycles the system
waits in between reading bits in the same memory column. A lower value means less waiting and thus a faster computer.
Typical CAS latency values in Pentium III-class and Athlon systems range from two to
four cycles. Some BIOSs may even allow half-cycle values (for example, 2.5 cycles).
From a technical standpoint, adjusting CAS latency rather than clock speed holds an
advantage of a different type. Changing the CAS latency does not raise the operating
speed of any part of the computer. Hence, it should not add to the heat stress of any computer component and should not affect the life of your computers circuits.
System Integrity
Pushing any component beyond its rating runs the risk that the component will no longer
operate. A single malfunctioning component can lead to your computer no longer booting. But just because your computer boots after you make a BIOS adjustment that affects
performance does not mean all is well. You still run the risk that one or more of the parts
inside your computer are on the edge of failing. As a result, your computer may boot, but
may not operate reliably.
217
218
Memory errors are among the most likely problems to arise from pushing memory speed
up. Although server memory often has built-in error correction to ward against memory
problems, most modern workstations, desktops, and notebook computers lack even paritychecking memory that warns of memory errors (by halting the system). These unprotected systems can develop memory errors without warning, and these errors can ripple
though your system until it crashes. In other words, adjust memory speed cautiously, and
dont depend on a souped-up system until youve given it time to prove itself.
Microprocessors, too, become unreliable if operated at speeds that are too high. Some
BIOSs, however, allow you to help your microprocessor cope with high clock speeds by
increasing the core voltage. Higher voltages allow the chip to operate faster without sacrificing data integrity. If you increase your microprocessor speed significantly, you may want
to increase its core voltage to ensure its reliable operation. The corollary also applies. If
you increase the core voltage of your microprocessor, you should be able to increase the
chips clock frequency more than would have otherwise been possible.
You wont find guidance as to the relationship between voltage and potential speed
increases anywhere. Chip-makers test only at design voltages and frequencies.
Motherboard-makers defer to the chip-makers, warning only that you should keep within
the range specified by the chip-maker. Your only choice is to try and see.
Before adjusting microprocessor core voltage, be sure that you have adequate cooling for
your chip. Microprocessor power useand therefore heat outputincreases to the square
of the increase in voltage. Consequently, a small voltage increase can quickly lead to the
overheating of your microprocessor and, potentially, its self-immolation.
BIOS-makers provide some help with this issue, however. Among the settings for most
BIOSs affording voltage control are temperature sensors that allow you to monitor both
chip temperature and the temperature inside your computers case. Some allow for warning when temperatures exceed the limits you set. You may be able to set a shutdown temperature for your microprocessor.
Even these shutdown temperature settings dont guarantee the safety of your expensive
silicon, however. The temperature probes measure the microprocessors external temperature, not the temperature at the semiconductor junctions where damage actually occurs. If
you grossly increase the voltage to the microprocessor, the junction temperature can skyrocket into the meltdown range well before any external chip sensor detects a significant
temperature change. In other words, step cautiously. If you are tempted to try increasing
chip core voltage, do so in small incrementsand carefully monitor chip temperature
after each change.
CHAPTER 7 BIOS
Storage Functions
Other Options
Speeding up clocks and shaving cycles arent the only ways you can potentially pluck more
performance from your system through BIOS adjustments. Many BIOSs afford you some
opportunity to optimize the waits inherent in bus access. In addition, nearly all modern
BIOSs allow you to shadow the BIOS code itself.
Although these adjustments sound promising, in reality they do little to improve the overall speed of your system. For example, the real purpose of the bus recovery time adjustments is to help achieve compatibility with older expansion boards. Both the 8-bit I/O
recovery time and 16-bit I/O recovery time settings present in the BIOSs of systems with
both PCI and ISA expansion buses let you slow down the system to accommodate marginal
expansion boards connected through the legacy or ISA bus. Although increasing the
recovery time, usually measured in bus cycles, may allow you to use older ISA expansion
boards, decreasing it will not affect normal system performance.
BIOS shadowing copies the BIOS firmware from ROM to RAM. In theory, it enables your
system to read the BIOS code from RAM (32-bit or 64-bit) memory instead of 8- or 16bit ROM memory. This should speed up BIOS-mediated operations substantially.
However, modern operating systems, including Linux and all varieties of Windows from
95 through XP, sidestep the BIOS by using drivers that take direct control of system hardware. Shadowing will not affect the performance of programs run under any of these
modern operating systems. In other words, shadowing is a carryover from the days of
DOS thats essentially irrelevant to modern computers. Leaving BIOS shadowing off will
help ensure system stability and slightly increase available memory.
Video shadowing is similarly ineffectual in accelerating todays software. Windows uses its
own drivers that directly access video hardware, so video shadowing does not affect normal system performance.
Some BIOSs also offer adjustments for your systems memory cache. Depending on the
BIOS, you may have individual control of both Level 1 (L1) and Level 2 (L2) caches,
which are sometimes termed primary and secondary, or more confusingly, internal and
external (confusing because both the L1 and L2 caches are internal with many microprocessors, such as current Athlon, Duron, Pentium III, and Pentium 4 chips). Your BIOS
may also allow you to selectively switch off the microprocessor read cache or write cache.
No matter. In any case, the only potential here is for slowing down your system by turning the cache off. These cache controls are not designed to adjust performance. Rather,
they are useful only for troubleshooting or preventing problems with software. But programs that have problems with the cache are probably programs you shouldnt be running.
219
220
Tuning your system with its BIOS is inevitably a trial-and-error experience. You are
investing your time to make tests more rigorous than those made by chip manufacturers
so that you can run all of your computers components as close to the edge as possible.
Upgrades
At one time, engineers viewed the BIOS as a great sculpture or other great work of art,
fixed forever in time, perfect and immutable. The BIOS code would be written in readonly memory chips that could not be altered, at least by mere mortals without electrical
engineering degrees. But human frailties and rapidly changing technologies forced them
to change their minds. They made mistakes in writing some BIOSs that only became
apparent after they were shipped off to consumers in finished computers. Or, a few
months after they had finished work on their BIOSsand after computers using them
found their ways into consumers handssomeone decided it would be great if hard disks
could store twice as much as the BIOS made allowances for. The only way to make up for
the shortcoming of the permanent BIOS was to replace the memory chips, a task most
people face with the same calm ease as disarming a tactical nuclear weapon.
Consequently, today most computers come with their BIOS code written in Flash ROM
chips, which can be rewritten with relative safety and without even removing the lid from
your computer system. If your BIOS needs an upgrade, either to fix an error or to accommodate a new technology, you can usually download a new BIOS from the Web site of the
computers manufacturer.
Updating a Flash BIOS usually involves two files: a binary file that contains the actual
BIOS code, and a special program called a loader. The loader activates the motherboard
circuitry required to write to the BIOS chips and then transfers the contents of the binary
BIOS file to the chips. The loaders job is one of the most delicate in the operation of a
computer because if it makes a mistake, your system may be rendered inoperable. It might
not be able to boot up and rerun the loader. Consequently, for your security, some system-makers require you take elaborate precautions before you try making a BIOS
upgrade. For example, the manufacturer may require you to use both the battery and AC
power supplies of a notebook computer to make a BIOS update.
To prevent a BIOS upgrade from accidentally rendering your computer inoperable, many
computers include boot block protection. This feature simply protects or prevents the block
of the Flash ROM used for the essential bootup codethe code necessary to let your system read a floppy diskfrom being erased during the upgrade. The protected code is sufficient to get your floppy going so that you can try upgrading your BIOS once again (or if
the new BIOS itself is the problem, restore your old BIOS).
CHAPTER 7 BIOS
Upgrades
Because motherboard circuitry varies among different computer brands and models, the
BIOS loader made for one system may not work with another. Unless youre advised otherwise by your computers manufacturer or a BIOS-maker, you should assume that you
need a matched pair of filesthe BIOS code and loaderto make a BIOS upgrade. In
general a loader will warn if it is not meant for your particular system, but dont depend
on such a warning to avoid incompatibilities. A loader that makes a mistake can mean
your computer must go back to the factory for new BIOS chips before it will boot again.
Along with fixing problems, upgrading a flash BIOS can sometimes cause them, too.
BIOS-makers usually design their flash BIOS loaders to reset your computers setup parameters to the factory defaults so your system gets a fresh start with its new BIOS. If
youve customized your system before making the upgrade (for example, adjusting the
energy conservation options of a notebook machine), youll have to re-establish your settings after the upgrade. Sometimes you may find that the load process has altered the date
set in CMOS. Operating your system without checking and resetting the date can cause
problems with programs sensitive to the relationship between the current time and file
date codes. For example, your backup system or disk maintenance program may not work
properly after youve upgraded your BIOS if you dont properly restore your system clock.
221
PART 3
Communications
Lets say I have a box containing all the knowledge in the world. How much would you pay
me for it? Oh, one more thing. Its a sealed box and nothing you can do will open it.
Nothing, not now or ever. It has no hinges or even a lid. You cant saw the top off or even
bore a hole through one of the sides. It resists all acids, nuclear explosions, even Supermans
X-ray eyes. It is just an unopenable, sealed box that I absolutely guarantee you is filled with
the answers to all your questions and all those of everyone else in the world. And its a nifty
box, oak-oiled a nice, rich brown.
Id guess you might offer a few dollars, whatever its decorative value. You might even eye it
for seating if you didnt have a chair (though I have to admit that the box is not all that big
and it sounds pretty empty at that). Basically, the value of all that knowledge is nothing if you
cant get at it and use it.
What would your computer be worth to you if you couldnt get your answers out? If you
couldnt connect with the Internet? If you couldnt plug in a CD writer or even a printer? Id
wager not much. Its even an uncomfortable flop as a chair.
Fortunately, your computer is not resistant to a handsaw or electric drill. You can even open
it up with a screwdriver. Better still, you can simply plug in a cable or two and connect up
with another computer, a network, the Internet, or an Erector set. Your computer is a
superlative communications machine. Thats what gives it its valueunless you really are
using yours as a chair.
A computer communicates in many ways, both outside and inside its box. Moving information around electronically takes a lot of ingenuity and some pretty amazing technologies.
Your systems microprocessor must organize its thoughts, make them into messages that it
can send out through its ports or down the wires that make up its expansion bus. It needs to
know how to talk and, more importantly, how to listen. It needs to put the signals in a form
that other computers and add-on devices (peripherals) can understanda form that can travel
a few inches across a circuit board, tens of thousands of miles through thin copper wires, or
as blips of light in a fiber-optic cable. And it has to do all that without making a mistake.
The next seven chapters outline how your computer communicates. Well start out with the
concepts that underlie all computer (and, for that matter, electronic) communications. From
there, well step into the real world and examine how your system talks to itself, sending messages inside its own case to other devices you might install there. Well look at the interfaces
it uses to send messagesthe computers equivalent of a telephone handset that it speaks and
listens through. Then well examine how it connects to external devices, all the accessories
and peripherals you plug in. From there, well examine the technologies that allow you to
connect to the most important peripherals of allother computersboth in your home or
office on a network, and elsewhere in the world through the Internet.
Theres a lot to explore, so you might as well put your feet up. After all, a computer may not
work well as a chair, but it can be a pretty good footstool.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Channels and Coding

Communications is all about language and, unfortunately, understanding
computer communications requires learning another language, one that
explains the technology that allows us to send messages, music, and even
pictures electronically.
Were going to talk about channels, which are nothing more than the
paths your electronic messages take. A communication channel is like a
television channel. More than that, a television channel is a communication channela designated pathway for the television signal to get from
the broadcasting station to your antenna and television set.
Different channels require technologies for moving information. Paper
mail is a communication channel that requires you to use ink and paper
(maybe a pencil in a pinch). Electronic communication requires that you
translate your message into electronic form.
You cannot directly read an electronic message, no more than you could
read it were it written in a secret code. In fact, when your message
becomes electronic, it is written in a nonsecret code. Although human
beings cannot read it, computers can, because some aspects of the signal
correspond to the information in your message, exactly as code letters
correspond to the text of a secret message.
As background for understanding how computer communications work,
were first going to look at the characteristics and limitations of the
channels that your computer might use. Then well take a look at some
of the ways that signals get coded so they can be passed through those
channelsfirst as expansion buses inside your computer, then as serial and
226
parallel communication channels between computers. Note that this coding of data is an
important technology that has wide applications. Its useful not only in communications
but also data storage.
Signal Characteristics
Communications simply means getting informationbe it your dulcet voice, a sheaf of
your most intimate thoughts, or digital datafrom one place to another. It could involve
yelling into the wind, smoke signals, express riders on horseback, telegraph keys, laser
beams, or whatever technology currently holds favor. Even before people learned to
speak, they gestured and grunted to communicate with one another. Although today many
people dont do much more than that, we do have a modern array of technologies for
moving our thoughts and feelings from one point to another to share with others or for
posterity. Computers change neither the need for, nor purpose of, communications.
Instead, they offer alternatives and challenges.
Thanks to computer technology, you can move not only your words but also images
both still and movingat the speed of light. Anything that can be reduced to digital form
moves along with computer signals. But the challenge is getting what you want to communicate into digital form, a challenge we wont face until we delve into Part 5, Human
Interface.
Even in digital form, communications can be a challenge.
Channels
If you were Emperor of the Universe, you could solve one of the biggest challenges in
communications simply by forbidding anyone else to send messages. You would have
every frequency in the airwaves and all the signal capacity of every wire and optic cable
available to you. You wouldnt have to worry about interference because should someone
interfere with your communications, you could just play the Red Queen and shout Off
with their heads! Alas, if your not Emperor of the Universe, someone may take a dim
view of your hogging the airwaves and severing the heads of friends and family. Worse
yet, youre forced to do the most unnatural of actsactually sharing your means of communication.
Our somewhat civilized societies have created a way to make sharing more orderly if not
fairdividing the space available for communications into channels. Each television and
radio station, for example, gets a range of frequencies in which to locate its signals.
But a communications channel is something more general. It is any path through which
you can send your signals. For example, when you make a telephone connection, youve
established a channel.
CHAPTER 8 CHANNELS AND CODING
Although there are regulatory systems to ensure that channels get fairly shared, all channels are not created equal. Some have room for more signals or more information than
others. The channel used by a telephone allows only a small trickle of information to flow
(which is why telephone calls sound so tinny). A fiber optic cable, on the other hand, has
an immense space for information, so much that it is usually sliced into several hundred or
thousand channels.
Bandwidth
The primary limit on any communications channel is its bandwidth, the chief constraint on
the data rate through that channel. Bandwidth merely specifies a range of frequencies
from the lowest to the highest that the channel can carry or present in the signal. Its like
your stereo system. Turn the bass (the low frequencies) and the treble (the high frequencies) all the way down, and youve limited the bandwidth to the middle frequencies. It
doesnt sound very good because youre missing part of the signalyouve reduced its
bandwidth.
Bandwidth is one way of describing the maximum amount of information that the channel
can carry. Bandwidth is expressed differently for analog and digital circuits. In analog
technology, the bandwidth of a circuit is the difference between the lowest and highest
frequencies that can pass through the channel. Engineers measure analog bandwidth in
kilohertz or megahertz. In a digital circuit, the bandwidth is the amount of information
that can pass through the channel. Engineers measure digital bandwidth in bits, kilobits,
or megabits per second. The kilohertz of an analog bandwidth and the kilobits per second
of digital bandwidth for the same circuit are not necessarily the same and often differ
greatly.
Nature conspires with the laws of physics to limit the bandwidth of most communications
channels. Although an ideal wire should be able to carry any and every frequency, real
wires cannot. Electrical signals are social busybodies of the physical world. They interact
with themselves, other signals, and nearly everything else in the worldand all these
interactions limit bandwidth. The longer the wire, the more chance there is for interaction, and the lower the bandwidth the wire allows. Certain physical characteristics of wires
cause degradations in their high frequency transmission capabilities. The capacitance
between conductors in a cable pair, for instance, increasingly degrades signals as their frequencies rise, finally reaching a point that a high frequency signal might not be able to
traverse more than a few centimeters of wire. Amplifiers or repeaters, which boost signals
so that they can travel longer distances, often cannot handle very low or very high frequencies, thus imposing more limits.
Even fiber optic cables, which have bandwidths dozens of times that of copper wires, also
degrade signals with distancepulses of light blur a bit as they travel through the fiber,
until equipment can no longer pick digital bits from the blur.
227
228
Emissions Limitations
Even the wires inside computers have bandwidth limitations, although they are infrequently explicitly discussed as such. Every wire, even the copper traces on circuit boards
(which act as wires), not only carries electricity but also acts as a miniature radio antenna.
Its unavoidable physicsmoving electricity always creates a magnetic field that can radiate
away as radio waves.
The closer the length of a wire matches the wavelength of an electrical signal, the better it
acts as an antenna. Because radio waves are really long at low frequencies (at one megahertz, the wavelength is about 1000 feet long), radio emissions are not much of a problem
at low frequencies. But todays computers push frequencies into the gigahertz range, and
their short wavelengths make ordinary wires and circuit traces into pretty good antennas.
Engineers designing modern computers have to be radio engineers as well as electrical
engineers.
Good designs cause fewer emissions. For example, engineers can arrange circuits so the
signals they emit tend to cancel out. But the best way to minimize problems is to keep frequencies lower.
Lower frequencies, of course, limit bandwidth. But when youre working inside a computer, theres an easy way around thatincrease the number of channels by using several
in parallel. That expedient is the basis for the bus, a collection of several signals traveling
in parallel, both electrically and physically. A computer expansion bus uses 8, 16, 32, or 64
separate signal pathways, thus lowering the operating frequency to a fraction of that
required by putting all the data into a single, serial channel.
Standards-Based Limitations
The airwaves also have a wide bandwidth, and they are not bothered by radiation issues.
After all, the signals in the airwaves are radiation. Nor are they bothered by signal degradation (except at certain frequencies where physical objects can block signals). But because
radio signals tend to go everywhere, bandwidth through the airwaves is limited by the old
bugaboo of sharing. To permit as many people to share the airwaves as possible, channels
are assigned to arbitrarily defined frequency bands.
Sharing also applies to wire- and cable-based communications. Most telephone channels,
for example, have an artificial bandwidth limitation imposed by the telephone company.
To get the greatest financial potential from the capacity of their transmissions cables,
microwave systems, and satellites, telephone carriers normally limit the bandwidth of telephone signals to a narrow slice of the full human voice range. Thats why phone calls have
their unmistakable sound. By restricting the bandwidth of each call, the telephone company can stack many calls on top of one another using an electronic technique called multiplexing. This strategy allows a single pair of wires to carry hundreds of simultaneous
conversations.
The real issue is not how signals get constrained by bandwidth, but rather the effect of
bandwidth on the signal. Bandwidth constrains the ability of a channel to carry information. The narrower the bandwidth, the less information that gets through. It doesnt matter whether the signal is analog or digital, your voice, a television picture, or computer
data. The bandwidth of the channel will constrain the actual information of any signal
passing through the channel.
Bandwidth limits appear wherever signals must go. Even the signals flowing from one part
of a circuit to another inside a computer are constrained by bandwidth limits.
Modulation
Computers and the standard international communications systems grew up independently, each with its own technological basis. Computers began their commercial lives as
digital devices because of the wonderful ability of digital signals to resist errors.
Communications systems started on a foundation of analog technology. Their primary
focus began as the transmission of the human voice, which was most easily communicated
in analog form with the primitive technologies then available. Since the early days of computers and communications, the two technologies have become intimately intertwined:
Computers have become communications engines, and most communications are now
routed by computers. Despite this convergence, however, the two technologies remain
stuck with their incompatible foundationscomputers are digital and human communications are analog.
The technology that allows modems to send digital signals through the analog communications circuits is called modulation. The modulation process creates analog signals that
contain all the digital information of the original digital computer code but which can be
transmitted through the analog circuits, such as the voice-only channels of the telephone
system.
Someday the incompatibility between the computer and communications worlds will end,
perhaps sooner than you might think. In the year 2000, computer data overtook the
human voice as the leading use of long-distance communications. Even before that, nearly
all long distance voice circuits had been converted to use digital signals. But these changes
havent eliminated the need for using modulation technology. Dial-up telephone lines
remain the way most people connect their computers to the Internet, and local loops (the
wires from the telephone company central office to your home) remain stubbornly analog.
Moreover, many types of communications media (for example, radio waves) require modulation technology to carry digital (and even voice and music) signals.
The modulation process begins with a constant signal called the carrier, which carries or
bears the load of the digital information. The digital data in some ways modifies the carrier, changing some aspect of it in such a way the change can be later detected in receiving
229
230
equipment. The changes to the carrierthe information content of the signalare the
modulation, and the data is said to modulate the carrier.
Demodulation is the signal-recovery process, the opposite of modulation. During demodulation, the carrier is stripped away and the encoded information is returned to its original
form. Although logically just the compliment of modulation, demodulation usually
involves entirely different circuits and operating principles.
In most modulation systems, the carrier is a steady-state signal of constant amplitude
(strength), frequency, and coherent phasethe electrical equivalent of a pure tone.
Because it is unchanging, the carrier itself is devoid of content and information. Its the
equivalent of one, unchanging digital bit. The carrier is simply a package, like a milk carton for holding information. Although you need both the carrier and milk carton to move
its contents from one place to another, the package doesnt affect the contents and is
essentially irrelevant to the product. You throw it away once youve got the product where
you want it.
The most common piece of equipment you connect to your computer that uses modulation is the modem. In fact, the very name modem is derived from this term and the reciprocal circuitthe demodulatorthats used in reception. Modem is a foreshortening of
the words modulator and demodulator.
A modem modulates an analog signal that can travel through the telephone system with
the digital direct current signals produced by your computer. The modem also demodulates incoming signals, stripping off the analog component and passing the digital information to your computer. The resulting modulated carrier wave remains an analog signal
that, usually, whisks easily through the telephone system.
Radio systems also rely on modulation. They apply voice, video, or data information to a
carrier wave. Radio works because engineers created receivers, which can tune to the frequency of the carrier and exclude other frequencies, so you can tune in to only the signal
(and modulation) you want to receive. Note that most commonly, radio systems use analog signals to modulate their carriers. The modulation need not be digital.
The modulation process has two requirements. The first is the continued compatibility
with the communications medium so that the signal is still useful. The second is the ability to separate the modulation from the carrier so that you can recover the original signal.
Just as AM and FM radio stations use different modulation methods to achieve the same
end, designers of communications systems can select from several modulation technologies to encode digital data in a form compatible with analog transmission systems. The
different forms of modulation are distinguished by the characteristics of the carrier wave
that are changed in response to changes in data to encode information. The three primary
characteristics of a carrier wave that designers might elect to vary for modulation are its
amplitude, its frequency, and its phase. Pulse modulation is actually a digital signal that
emulates an analog signal and can be handled by communications systems as it if it were
one.
Carrier Wave Modulation

The easiest way to understand the technique of modulation is to look at its simplest form,
carrier wave modulation, which is often abbreviated as CW, particularly by radio broadcasters.
As noted previously, a digital signal, when stripped to its essential quality, is nothing more
than a series of bits of information that can be encoded in any of a variety of forms. We
use zeros and ones to express digital values on paper. In digital circuits, the same bits take
the form of the high and low direct current voltages, the same ones that are incompatible
with the telephone system. However, we can just as easily convert digital bits into the
presence or absence of a signal that can travel through the telephone system (or that can
be broadcast as a radio wave). The compatible signal is, of course, the carrier wave. By
switching the carrier wave off and on, we can encode digital zeroes and ones with it.
The resulting CW signal looks like interrupted bursts of round sine waves, as shown in
Figure 8.1.
FIGURE 8.1
Carrier wave modulation.
The figure shows the most straightforward way to visualize the conversion between digital
and analog, assigning one full wave of the carrier to represent a digital one (1) and the
absence of a wave to represent a zero (0). In most practical simple carrier waves systems,
however, each bit occupies the space of several waves. The system codes the digital information not as pulses per se, but as time. A bit lasts a given period regardless of the number of cycles occurring within that period, making the frequency of the carrier wave
irrelevant to the information content.
Although CW modulation has its shortcomings, particularly in wire-based communications, it retains a practical application in radio transmission. It is used in the simplest radio
transmission methods, typically for sending messages in Morse code.
231
232
One of the biggest drawbacks of carrier wave modulation is ambiguity. Any interruption
in the signal may be misinterpreted as a digital zero. In telephone systems, the problem is
particularly pernicious. Equipment has no way of discerning whether a long gap between
bursts of carrier wave is actually significant data or a break in or end of the message.
Frequency Shift Keying

A more reliable way of signaling digital information is to use separate and distinct frequencies for each digital state. For example, a digital 1 would cause the carrier wave to
change to a higher frequency, much as it causes a higher voltage, and a digital 0 would
shift the signal to a lower frequency. Two different frequencies and the shifts between
them could then encode binary data for transmission across an analog system. This form
of modulation is called frequency shift keying, or FSK, because information is encoded in
(think of it being keyed to) the shifting of frequency.
The keying part of the name is actually left over from the days of the telegraph when
this form of modulation was used for transmitting Morse code. The frequency shift came
with the banging of the telegraph key.
In practical FSK systems, the two shifting frequencies are the modulation that is applied
to a separate (and usually much higher) carrier wave. When no modulation is present,
only the carrier wave at its fundamental frequency appears. With modulation, the overall
signal jumps between two different frequencies. Figure 8.2 shows what an FSK modulation looks like electrically.
FIGURE 8.2
Frequency shift keying.
Frequency shift keying unambiguously distinguishes between proper code and a loss of
signal. In FSK, a carrier remains present even when no code is being actively transmitted.
FSK is thus more reliable than CW and more widely used in communications systems.
The very earliest of computer modems (those following the Bell 103 standard) used ordinary FSK to put digital signals on analog telephone lines.
Amplitude Modulation
Carrier wave modulation is actually a special case of amplitude modulation. Amplitude is the
strength of the signal or the loudness of a tone carried through a transmission medium,
such as the telephone wire. Varying the strength of the carrier in response to modulation
to transmit information is called amplitude modulation. Instead of simply being switched on
and off, as with carrier wave modulation, in amplitude modulation the carrier tone gets
louder or softer in response to the modulating signal. Figure 8.3 shows what an amplitude
modulated signal looks like electrically.
FIGURE 8.3
Amplitude modulationsignal strength
(vertical) versus time
(horizontal).
Amplitude modulation is most commonly used by radio and television broadcasters to

transmit analog signals because the carrier can be modulated in a continuous manner,
matching the raw analog signal. Although amplitude modulation carries talk and music to
your AM radio and the picture to your (nondigital) television set, engineers also exploit
amplitude modulation for digital transmissions. They can, for example, assign one amplitude to indicate a logical 1 and another amplitude to indicate a logical 0.
Pure amplitude modulation has one big weakness. The loudness of a signal is the characteristic most likely to vary during transmission. Extraneous signals add to the carrier and
mimic modulation. Noise in the communication channel line mimics amplitude modulation and might be confused with data.
Frequency Modulation
Just as amplitude modulation is a refinement of CW, frequency shift keying is a special
case of the more general technology called frequency modulation. In the classic frequency
modulation system used by FM radio, variations in the loudness of sound modulate a carrier wave by changing its frequency. When music on an FM station gets louder, for example, the radio stations carrier frequency shifts its frequency more. In effect, FM translates
changes in modulation amplitude into changes in carrier frequency. The modulation does
not alter the level of the carrier wave. As a result, an FM signal electrically looks like a
train of wide and narrow waves of constant height, as shown in Figure 8.4.
233
234
FIGURE 8.4
Frequency modulationsignal strength
(vertical) versus time
(horizontal).
In a pure FM system, the strength or amplitude of the signal is irrelevant. This characteristic makes FM almost immune to noise. Interference and noise signals add into the
desired signal and alter its strength, but FM demodulators ignore these amplitude
changes. Thats why lightning storms and motors dont interfere with FM radios. This
same immunity from noise and variations in amplitude makes frequency modulation a
more reliable, if more complex, transmission method for digital data.
Phase Modulation
Another variation on the theme of frequency modulation is phase modulation. This technology works by altering the phase relationship between the waves in the signal. An
unmodulated carrier is a train wave in a constant phase relationship. That is, the waves
follow one after another precisely in step. The peaks and troughs of the train of waves
flow in constant intervals. If one wave were delayed for exactly one wavelength, it would
fit exactly atop the next one.
By delaying the peak of one wave so that it occurs later than it should, you can break the
constant phase relationship between the waves in the train without altering the overall
amplitude or frequency of the wave train. In other words, you shift the onset of a subsequent wave compared to those that precede it. At the same time, you will create a
detectable state change called a phase shift. You can then code digital bits as the presence
or absence of a phase shift.
Signals are said to be in phase when the peaks and troughs of one align exactly with
another. When signals are 180 degrees out of phase, the peaks of one signal align with
the troughs of the other. Figure 8.5 shows a 180-degree phase shift of a carrier wave.
Note that the two waveforms shown start in phase and then, after a phase shift, end up
being 180 degrees out of phase.
FIGURE 8.5
Phase modulation
showing a 180-degree
phase shift.
If you examine the shapes of waves that result from a phase shift, youll see that phase
modulation is a special case of FM. Delaying a wave lengthens the time between its peak
and that of the preceding wave. As a result, the frequency of the signal shifts downward
during the change, although over the long term the frequency of the signal appears to
remain constant.
One particular type of phase modulation. called quadrature modulation. alters the phase of
the signal solely in increments of 90 degrees. That is, the shift between waves occurs at a
phase angle of 0, 90, 180, or 270 degrees. The quad in the name of this modulation
method refers to the four possible phase delays.
Pulse Modulation
Pulse modulation is actually a digital signal made up from discrete pulses of energy. The
pulses have an underlying characteristic frequencyessentially their data clockthat is
the equivalent of a carrier wave. In fact, to analog circuits, some forms of pulse modulation are demodulated as if they were analog signals. For example, add in some extra
pulses, and you change the apparent frequency of the clock/carrier, thus creating frequency modulation. Change the duty cycle of the pulses (make the on part of the pulse
last longer or shorter), and the analog strength of the carrier varies, thus creating analog
modulation.
But you can also alter the pulses in a strictly digital manner, changing the bit-pattern to
convey information. The signal then is completely digital, yet it still acts like it is an ordinary modulated carrier wave because it retains its periodic clock signal.
Pulse frequency modulation (PFM) is the simplest form of most pulse modulation. Using
PFM, the modulator changes the number of pulses in a given period to indicate the
strength of the underlying analog signal. More pulses means a higher frequency, so the
frequency changes in correspondence with the amplitude of the analog signal. The result
can be detected as ordinary frequency modulation.
235
236
Pulse width modulation (PWM) varies the width of each pulse to correspond to the strength
of the underlying analog signal. This is just another way of saying that PWM alters the
duty cycle of the pulses. The changes in signal strength with modulation translate into
ordinary amplitude modulation.
Pulse code modulation (PCM) is the most complex form of pulse modulation, and it is
purely digital. PCM uses a digital number or code to indicate the strength of the underlying analog signal. For example, the voltage of the signal is translated into a number, and
the digital signal represents that number in binary code. This form of digital modulation
is familiar from its wide application in digital audio equipment such as CD players.
Pulse position modulation (PPM) uses the temporal position of a pulse within a clock period
to indicate a discrete value. For example, the length of one clock period may be divided
into four equal segments. A pulse can occur in one, and only one, of these chip segments,
and which chip the pulse appears in (its position inside the symbol duration or clock
period) encodes its value. For example, the four segments may be numbered 0, 1, 2, and 3.
If the pulse appears in segment 2, that clock period carries a value of 2.
Figure 8.6 shows the pulse position for the four valid code values using the PPM scheme
developed for the IrDA interface.
FIGURE 8.6
Pulse positions for the
four valid code values
of 4PPM under IrDA.
Complex Modulation
The various modulation techniques are not mutually exclusive. Engineers have found ways
of pushing more data through a given communication channel by combining two or more
techniques to create complex modulation schemes.
In broadcasting, complex modulation is not widely used. Theres not much point to it. In
data communications, however, it allows engineers to develop systems in which a signal
can have any of several states in a clock period. Each state, called a symbol by engineers,
can encode one or more bits of data. The more states, the more bits that each symbol can
encode using a technique called group-coding, which is discussed later in the section with
the same name.
Parallel Channels
A parallel connection is one with multiple signals traveling down separate wires, side by side.
For computers, this is the natural way of dealing with communications. Everything about
a microprocessor is parallel. Its registers look at data bits in parallel32 or more at a
time. The chip sucks information into its circuitry the same way32, 64, or even 128 bits
at a time. Its only natural. The chip works on multiple bits with every tick of the system
clock, so you want to fill it with data the same way (and at the same rate).
When it comes time for the chip to communicateboth with the rest of the computer
and with the outside worldits only natural to use the same kind of connection, one that
can pass multiple data bits in parallel. Moreover, by providing a number of data paths, a
parallel channel has the potential for moving more information in a given time. Parallel
channels can be faster.
With this design philosophy in mind, engineers have developed parallel communication
channels of two types. Expansion buses provide a wide, parallel path for data inside the
computer. The familiar Peripheral Component Interconnect (PCI) bus is such a design.
Parallel communications ports serve to connect the computer with other equipment. The
parallel printer port is the exemplar of this technology.
Of the two, ports are simpler. They just have less to do. They connect external devices to
the computer, and they only need to move information back and forth. They sacrifice the
last bit of versatility for practicality. Expansion buses must do more. In addition to providing a route for a stream of data, they also must provide random access to that data. They
must operate at higher speeds, as well, as close to the clock speed of the microprocessor as
can be successfully engineered and operated. They need more signals and a more complex
design to carry out all their functions.
Some designs, however, straddle the difference. SCSI offers external parallel connections
like a port but also works a bit like an expansion bus. The common hard disk interface,
AT Attachment (or IDE), is part of an expansion bus, but it doesnt allow direct addressing. All, however, use parallel channels to move their information.
Ports
The concept behind a parallel port is simply to extend the signals of the computer to
another device. When microprocessors had data buses only eight bits wide, providing an
external connection with the same number of bits was the logical thing to do. Of course,
when microprocessors had data buses only eight bits wide, the circuitry needed in order to
convert signals to another form was prohibitively expensive, especially when you need to
convert the signal when it leaves the computer and convert it back to parallel form at the
receiving device, even a device as simple as a printer. Of course, when microprocessors
had eight-bit data buses, Mammoths still strode the earth (or so it seems).
Many people use the term port to refer to external connections on their computers. A port
is a jack into which you plug a cable. By extension, the name refers to the connection system (and standard) used to make the port work. Take a similar design and restrict it to use
237
238
inside the computer, and it becomes an interface, such as the common disk interfaces. By
these definitions, the Small Computer System Interface (SCSI) is schizophrenic. It has a
split personality and works either way. In truth, the terms port and interface have many
overlapping meanings.
In the modern world, parallel connections have fallen by the wayside for use as ports. One
reason is that it has become cheaper to design circuits to convert signal formats. In addition, the cost of equipping signal paths to ward off interference increases with every added
path, so cables for high-speed parallel connections become prohibitively expensive.
Consequently, most new port designs use serial rather than parallel technology. But parallel designs remain a good starting point in any discussion of communications because they
are inherently simpler, working with computer data in its native form.
Even the most basic parallel connection raises several design issues. Just having a parallel
connection is not sufficient. Besides moving data from point to point, a parallel channel
must be able to control the flow of data, both pacing the flow of bits and ensuring that
information wont be dumped where it cannot be used. These issues will arise in the
design of every communications system.
Data Lines
The connections that carry data across a parallel channel are termed data lines. They are
nothing more than wires that conduct electrical signals from one end of the connection to
the other. In most parallel channels, the data lines carry ordinary digital signals, on-or-off
pulses, without modulation.
The fundamental factor in describing a parallel connection is therefore the number of
data lines it provides. More is always better, but more also means increased complexity
and cost. As a practical matter, most parallel channels use eight data lines. More usually is
unwieldy, and fewer would be pointless. The only standard connection system to use more
than eight is Wide SCSI, which doubles the channel width to 16 bits to correspondingly
double its speed potential.
Although todays most popular standards use only enough data lines to convey data, sometimes engineers add an extra line for each eight called a parity-check line. Parity-checking
provides a means to detect errors that might occur during communication.
Parity-checking uses a simple algorithm for finding errors. The sending equipment adds
together the bits on the data lines without regard to their significance. That is, it checks
which lines carry a digital one (1) and which carry a digital zero (0). It adds all the ones
together. If the sum is an odd number, it makes the parity-check line a one. If the sum is
even, it makes the parity-check line a zero. In that way, the sum of all the linesdata lines
and parity-check linesis always an even number. The receiving equipment repeats the
summing. If it comes up with an odd number, it knows the value of one of the data lines
(or the parity-check line) has changed, indicating an error has occurred.
Timing
The signals on a parallel channel use the same clocked logic as the signals inside a computer. Accordingly, there are times when the signals on the data lines are not validfor
example, when they are changing or have changed but are not yet stable. As with the signals inside the computer system, the parallel channel uses a clock to indicate when the signals on the data lines represent valid data and when they do not.
The expedient design, in keeping with the rest of the design of the parallel channel, is to
include an extra signal (and a wire to carry it) that provides the necessary clock pulses.
Most parallel channels use a dedicated clock line to carry the pulses of the required timing
signal.
Flow Control
The metaphor often used for a communications channel is that of a pipe. With nothing
more to the design than data lines and a clock, that pipe would be open ended, like a
waterline with the spigot stuck on. Data would gush out of it until the equipment
drowned in a sea of bits. Left too long, a whole room might fill up with errant data; perhaps buildings would explode and scatter bits all over creation.
To prevent that kind of disaster, a parallel channel needs some means of moderating the
flow of data, a giant hand to operate a digital spigot. With a parallel connection, the natural means of controlling data flow is yet another signal.
Some parallel channels have several such flow-control signals. In a two-way communications system, both ends of the connection may need to signal when they can no longer
accept data. Additional flow-control signals might indicate specific reasons why no more
data can be accepted. For example, one line in a parallel port might signal that the printer
has run out of paper.
Buses
The computer expansion bus shows digital communications at its most basicif you want
to move digital signals away from a microprocessor, just make the wires longer. The first
expansion bus shows this principle in action. The IBM Personal Computer expansion bus,
which evolved into the Industry Standard Architecture (a defining characteristic of personal computers for nearly 15 years), was really nothing more than the connections to the
microprocessor extended to reach further inside the computer.
Modern expansion buses go furtherin technology rather than distance. Although they
still stay within the confines of a computers case, they have evolved into elaborate
239
240
switching and control systems to help move data faster. One of the reasons for this change
in design arises from limitations imposed by the communications channel of the bus itself.
In addition, the modern bus has evolved as a communications system independent of the
microprocessor in the computer. By looking at the bus as a communications system unto
itself, engineers have added both versatility and speed. The next generation of expansion
buses will be almost indistinguishable from long-distance data communications systems,
sharing nearly all their technologies, except for being shorter and faster.
Background
The concept of a bus is simple: If you have to move more information than a single wire
can hold, use several. A bus is comprised of several electrical connections linking circuits
carrying related signals in parallel. The similarity in the name with the vehicle that carries
passengers across and between cities is hardly coincidental. Just as many riders fit in one
vehicle, the bus, many signals travel together in an electrical bus. As with a highway bus,
an electrical bus often makes multiple stops on its way to its destination. Because all the
signals on a bus are related, a bus comprises a single communication channel.
The most familiar bus in computer systems is the expansion bus. As the name implies, it is
an electrical bus that gives you a way to expand the potential of your computer by attaching additional electrical devices to it. The additional devices are often termed peripherals
or, because many people prefer to reserve that term for external accessories, simply expansion boards.
A simple bus links circuits, but todays expansion bus is complicated by many issues. Each
expansion bus design represents a complex set of design choices confined by practical constraints. Designers make many of the choices by necessity; others they pick pragmatically.
The evolution of the modern expansion bus has been mostly a matter of building upon
the ideas that came before (or stealing the best ideas of the predecessors). The result in
todays computers is a high-performance Peripheral Component Interconnect, or PCI,
upgrade system that minimizes (but has yet to eliminate) setup woes.
The range of bus functions and resources is wide. The most important of these is providing a data pathway that links the various components of your computer together. It must
provide both the channel through which the data moves and a means of ensuring the data
gets where it is meant to go. In addition, the expansion bus must provide special signals to
synchronize the thoughts of the add-in circuitry with those of the rest of the computer.
Newer bus designs (for example, PCI, but not the old Industry Standard Architecture, or
ISA, of the last generation of personal computers) also include a means of delegating system control to add-in products and tricks for squeezing extra speed from data transfers.
Although not part of the electrical operation of the expansion bus, the physical dimensions of the boards and other components are also governed by agreed-on expansion bus
specifications, which we will look at in Chapter 30, Expansion Boards. Moreover,
todays bus standards go far beyond merely specifying signals. They also dictate transfer
protocols and integrated, often automated, configuration systems.
How these various bus features workand how well they workdepends on the specific
features of each bus design. In turn, the expansion bus design exerts a major influence on
the overall performance of the computer it serves. Bus characteristics also determine what
you can add to your computerhow many expansion boards, how much memory, and
what other system componentsas well as how easy your system is to set up. The following sections examine the features and functions of all expansion buses, the differences
among them, and the effects of those differences in the performance and expandability of
your computer.
Data Lines
Although the information could be transferred inside a computer either by serial or parallel means, the expansion buses inside computers use parallel data transfers. The choice of
the parallel design is a natural. All commercial microprocessors have parallel connections
for their data transfers. The reason for this choice in microprocessors and buses is exactly
the same: Parallel transfers are faster. Having multiple connections means that the system
can move multiple bits every clock cycle. The more bits the bus can pass in a single cycle,
the faster it will move information. Wider busesthose with more parallel connections
are faster, all else being equal.
Ideally, the expansion bus should provide a data path that matches that of the microprocessor. That way, an entire digital quad-word used by todays Pentium-class microprocessors (a full 64 bits) can stream across the bus without the need for data conversions.
When the bus is narrower than a device that sends or receives the signals, the data must
be repackaged for transmissionthe computer might have to break a quad-word into four
sequential words or eight bytes to fit an old 16- or 8-bit bus. Besides the obvious penalty
of requiring multiple bus cycles to move the data the microprocessor needs, a narrow bus
also makes system design more complex. Circuitry to handle the required data repackaging complicates both motherboards and expansion boards.
On the other hand, increasing the number of data connections also complicates the design
of a computer, if just because of the extra space and materials required. Finding the optimal number of connections for a given application consequently is a tradeoff between
speed (which requires more connections) and complexity (which increases with connection
count and thus speed).
In modern computers with multiple expansion buses, each bus usually has a different
number of data connections, reflecting the speed requirements of each bus. The bus with
the highest performance demandsthe memory bushas the most connections. The bus
with the least performance demandsthe compatibility bushas the fewest.
241
242
Address Lines
As long as a program knows what to do with data, a bus can transfer information without
reference to memory addresses. Having address information available, however, increases
the flexibility of the bus. For example, making addressing available on the bus enables you
to add normal system memory on expansion boards. Addressing allows memory-mapped
information transfers and random access to information. It allows data bytes to be routed
to the exact location at which they will be used or stored. Imagine the delight of the dead
letter office of the Post Office and the ever-increasing stack of stationery accumulating
there if everyone decided that appending addresses to envelopes was superfluous.
Similarly, expansion boards must be able to address origins and destinations for data. The
easiest method of doing this is to provide address lines on the expansion bus corresponding to those of the microprocessor.
The number of address lines used by a bus determines the maximum memory range
addressable by the bus. Its a purely digital issuetwo to the power of the number of
address lines yields the maximum discrete address values that the bus can indicate. A bus
with eight address lines can address 256 locations (two to the eighth power); with 16
address lines, it can identify 65,536 locations (that is, two to the sixteenth power.)
Engineers have a few tricks to trim the number of address lines they need. For example,
modern microprocessors have no need to address every individual byte. They swallow data
in 32- or 64-bit chunks (that is, double-words or quad-words). They do not have to be
able to address memory any more specifically than every fourth or eighth byte. When
they do need to retrieve a single byte, they just grab a whole double-word or quad-word
at once and sort out what they need internally. Consequently, most 32-bit expansion buses
designed to accommodate these chips commonly delete the two least-significant address
bits.
Some bus designs eliminate the need for address lines entirely by using multiplexing.
Instead of separate address lines for addresses and data, they combine the functions on the
same connections. Some multiplexed designs use a special signal that, when present, indicates the bits appearing on the combined address and data lines are data. In the absence of
the signal, the bits on the lines represent an address. The system alternates between
addresses and data, first giving the location to put information and then sending the information itself. In wide-bus systems, this multiplexing technique can drastically cut the
number of connections required. A full 64-bit system might have 64 address lines and 64
data linesa total of 128 connections for moving information. When multiplexed, the
same system could get by with 65 connections64 combined address/data lines and one
signal to discriminate which is which.
Most systems require an address for every transfer of data across the bus. That is, for each
32-bit transfer through a 32-bit bus, the system requires an address to indicate where that
information goes. Some buses allow for a burst mode, in which a block of data going to
sequential addresses travels without indicating each address. Typically, a burst-mode transfer requires a starting address, the transfer of data words one after another, and some indication the burst has ended. A burst mode can nearly double the speed of data transfer
across the bus.
Timing
Buses need timing or clock signals for two reasons. The bus may be called upon to distribute timing systems throughout a computer to keep all the circuits synchronized. In addition, the data traveling on the bus itself requires clockingthe clock signal indicates when
the data (or addresses, or both) on the bus is valid.
The same signal can be used for clocking the bus data and synchronizing other circuits.
Such a bus is termed synchronous because it is synchronized. Alternatively, the bus may
have two (or more) separate clock signals for different purposes. Buses in which the clock
for data on the bus is not synchronized with the system clock are termed asynchronous
buses.
Flow Control
Buses require flow control for the same reason as portsto prevent peripherals from
being swamped with a tidal wave of data or to keep a peripheral from flooding the rest of
the system. To avoid data losses when a speed disparity arises between peripherals and the
host, most buses include flow-control signals. In the simplest form, the peripheral sends a
special not ready signal across the bus, warning the host to wait until the peripheral
catches up. Switching off the not ready signal indicates to the host to dump more data
on the peripheral.
System Control
Buses often carry signals for purposes other than moving data. For example, computer
expansion buses may carry separate lines to indicate each interrupt or request for a DMA
transfer. Alternatively, buses can use signals like those of parallel ports to indicate system
faults or warnings.
Most buses have a reset signal that tells the peripherals to initialize themselves and reset
their circuits to the same state they would have when initially powering up. Buses may
also have special signals that indicate the number of data lines containing valid data during
a transfer. Another signal may indicate whether the address lines hold a memory address
or an I/O port address.
Bus-Mastering
In the early days of computers, the entire operation of the expansion bus was controlled
by the microprocessor in the host computer. The bus was directly connected to the
243
244
microprocessor; in fact, the bus was little more than an extension of the connections on
the chip itself. A bus connected directly to the microprocessor data and address lines is
termed a local bus because it is meant to service only the components in the close vicinity
of the microprocessor itself. Most modern buses now use a separate bus controller that
manages the transfers across the bus.
Many bus designs allow for decentralized control, where any device connected to the bus
(or any device in a particular class of devices) may take command. This design allows each
device to manage its own transfers, thus eliminating any bottlenecks that can occur with a
single, centralized controller. The device that takes control of the bus is usually termed a
bus master. The device that it transfers data to is the bus slave. The bus master is dynamic.
Almost any device may be in control of the bus at any given moment. Only the device in
control is the master of the bus. However, in some systems where not every device can
control the bus, the devices that are capable of mastering the bus are termed bus masters,
even when they are not in controlor even installed on the system, for that matter.
If multiple devices try to take control of the bus at the same time, the result is confusion
(or worse, a loss of data). All buses that allow mastering consequently must have some
scheme that determines when and which master can take control of the bus, a process
called bus arbitration. Each bus design uses its own protocol for controlling the arbitration
process. This protocol can be software, typically using a set of commands set across the
bus like data, or hardware, using special dedicated control lines in the bus. Most practical
bus designs use a hardware signaling scheme.
In the typical hardware-based arbitration system, a potential master indicates its need to
take control of the bus by sending an attention signal on a special bus line. If several
devices want to take control at the same time, an algorithm determines which competing
master is granted control, how long that control persists, and when after losing control it
can be granted the same master again.
In software-based systems, the potential bus master device sends a request to its hosts
bus-control logic. The host grants control by sending a special software command in
return.
The design of the arbitration system usually has two competing goals: fairness and priority. Fairness ensures that one device cannot monopolize the bus, and that within a given
period every device is given a crack at bus control. For example, some designs require that
once a master has gotten control of the bus, it will not get a grant of control again until all
other masters also seeking control have been served. Priority allows the system to act
somewhat unfairly, to give more of the bus to devices that need it more. Some bus designs
forever fix the priorities of certain devices; others allow priorities of some or all masters to
be programmed.
Power Distribution
Although all electrical devices need some source of electricity to operate, no immutable
law requires an expansion bus to provide that power. For example, when you plug a
printer into your computer, you also usually plug the printer into a wall outlet. Similarly,
peripherals could have their own sources of needed electricity. Providing that power on
the bus makes the lives of engineers much easier. They dont have to worry about the
design requirements, metaphysics, or costs of adding power sources such as solar panels,
magnetohydrodynamic generators, or cold fusion reactors to their products.
Non-Bussed Signals
Buses are so named because their signals are bused together; but sometimes engineers
design buses with separate, dedicated signals to individual devices. Most systems use only
a single device-specific signal on a dedicated pin (the same pin at each device location, but
a different device-specific signal on the pin for each peripheral). This device-specific signaling is used chiefly during power-on testing and setup. By activating this signal, one
device can be singled out for individual control. For example, in computer expansion
buses, a single, non-bussed signal allows each peripheral to be individually activated for
testing, and the peripheral (or its separately controllable parts) can be switched off if
defective. Device-specific signals also are used to poll peripherals individually about their
resource usage so that port and memory conflicts can be automatically managed by the
host system.
Bridges
The modern computer with multiple busesoften operating with different widths and
speedsposes a problem for the system designer: how to link those buses together. Make
a poor link-up, and performance will suffer on all the buses.
A device that connects two buses together is called a bridge. The bridge takes care of all
the details of the transfer, converting data formats and protocols automatically without the
need for special programming or additional hardware.
A bridge can span buses that are similar or different. A PCI-based computer can readily
incorporate any bus for which bridge hardware is available. For example, in a typical PCI
system, bridges will link three dissimilar busesthe microprocessor bus, the high-speed
PCI bus, and an ISA compatibility busand may additionally tie in another, similar PCI
bus to yield a greater number of expansion opportunities.
The interconnected buses logically link together in the form of a family tree, spreading
out as it radiates downward. The buses closer to the single original bus at the top are
termed upstream. Those in the lower generations more distant from the progenitor bus are
termed downstream. Bridge design allows data to flow from any location on the family tree
to any other. However, because each bridge imposes a delay for its internal processing,
245
246
called latency, a transfer that must pass across many bridges suffers performance penalties.
The penalties increase the farther downstream the bus is located.
To avoid latency problems, when computers have multiple PCI buses, they are usually
arranged in a peer bus configuration. That is, instead of one bridge being connected to
another, the two or three bridges powering the expansion buses in the computer are connected to a common pointeither the PCI controller or the most upstream bridge.
Serial Channels
The one great strength of a busits wealth of signal pathwayscan be its greatest disadvantage. For every digital signal you want to move, youve got to arrange for 8, 16, even
64 separate connections, and each one must be protected from interference. Simple cables
become wide, complex, and expensive. Plugs and jacks sprout connections like a
Tyrannosaurus Rexs mouth does teethand can be just as vicious. You also need to work
out every day at the gym just to plug one in. Moreover, most communications channels
provide a single data path, much a matter of history as convenience. For conventional
communications such as the telephone, a single pathway (a single pair of wires) suffices for
each channel.
The simplicity of a single data pathway has another allure for engineers. Its easier for
them to design a channel with a single pathway for high speeds than it is to design a channel with multiple pathways. In terms of the amount of information they can squeeze
through a channel, a single fast pathway can move more information than multiple slower
paths.
The problem faced by engineers is how to convert a bus signal into one compatible with a
single pathway (say, a telephone connection). As with most communications problems, the
issue is one of coding information. A parallel connection like a bus encodes information as
the pattern of signals on the bus when the clock cycles. Information is a snapshot of the
bits on the bus, a pattern stretched across space at one instant of time. The bits form a
line in physical spacea row of digital ones and zeroes.
The solution is simple: Reverse the dimensions. Let the data stretch out into a long row,
not in the physical dimension but instead across time. Instead of the code being a pattern
that stretches across space, it becomes a code that spans time. Instead of the parallel position of the digital bits coding the information, their serial order in a time sequence codes
the data. This earns its name serial because the individual bits of information are transferred in a long series. Encoding and moving information in which the position of bits in
time conveys importance is called serial technology, and the pathway is a called a serial
channel.
Clocking
The most straightforward way of communicating (or storing) data is to assign a digital
code so that the presence of a signal (for instance, turning on a five-volt signal) indicates a
digital value of one, and another signal (say, zero volts) indicates the other digital value,
zero. Thats what digital is all about, after all.
However, such a signaling system doesnt work very well to communicate or store digital
values, especially when the values are arranged in a serial string. The problem is timing.
With serial data, timing is important in sorting out bits, and a straight one-for-one coding
system doesnt help with timing.
The problem becomes apparent when you string a dozen or so bits of the same value in a
row. Electrically, such a signal looks like a prolonged period of high or low voltage,
depending on whether the signal is a digital one or zero. Some circuit has to sort out how
many bits on one value this prolonged voltage shift represents. If the prolonged voltage
lasts only a bit or two, figuring out how many bits is pretty easy. The length of a two-bit
string differs from a three-bit string by 50 percent. If you time the duration of the shift,
the accuracy of your timing needs only to be accurate to a bit less than 50 percent for you
to properly sort out the bits. But string 100 bits in a row, and a mistake of 2 percent in
timing can mean the loss or, worse, misinterpretation of two bits of information.
An easy way to sort out the bits is to provide a separate timing signal, a clock, that defines
when each bit begins and ends. A separate clock is standard operating procedure for
clocked logic circuits, anyway. Thats why they are called clock logic. You can similarly provide a clock signal in communication circuits. Just add a separate clock signal.
However, a separate clock would take a separate channel, another communication circuit,
with all its added complication. Consequently, engineers have developed ways of eliminating the need for a separate clock.
Synchronous communications require the sending and receiving systemfor our purposes,
the computer and printerto synchronize their actions. They share a common timebase,
a serial clock. This clock signal is passed between the two systems either as a separate signal or by using the pulses of data in the data stream to define it. The serial transmitter
and receiver can unambiguously identify each bit in the data stream by its relationship to
the shared clock. Because each uses exactly the same clock, they can make the match based
on timing alone.
In asynchronous communications, the transmitter and receiver use separate clocks. Although
the two clocks are supposed to be running at the same speed, they dont necessarily tell
the same time. They are like your wristwatch and the clock on the town square. One or
the other may be a few minutes faster, even though both operate at essentially the same
speed: a day has 24 hours for both.
247
248
An asynchronous communications system also relies on the timing of pulses to define the
digital code. But they cannot look to their clocks as infallible guidance. A small error in
timing can shift a bit a few positions (say, from the least-significant place to the most significant), which can drastically affect the meaning of the digital message.
If youve ever had a clock that kept bad timefor example, the CMOS clock inside your
computeryou probably noticed that time errors are cumulative. They add up. If your
clock is a minute off today, it will be two minutes off tomorrow. The more time that
elapses, the more the difference between two clocks will be apparent. The corollary is also
true: If you make a comparison over a short enough period, you wont notice a shift
between two clocks even if they are running at quite different speeds.
Asynchronous communications banks on this fine slicing of time. By keeping intervals
short, they can make two unsynchronized clocks act as if they were synchronized. The
otherwise unsynchronized signals can identify the time relationships in the bits of a serial
code.
Isochronous communications involve time-critical data. Your computer uses information that
is transferred isochronously in real time. That is, the data is meant for immediate display,
typically in a continuous stream. The most common examples are video image data that
must be displayed at the proper rate for smooth full-motion video and digital audio data
that produces sound. Isochronous transmissions may be made using any signaling
schemebe it synchronous or asynchronous. They usually differ from ordinary data
transfers in that the system tolerates data errors. It compromises accuracy for the proper
timing of information. Whereas error-correction in a conventional data transfer may
require the retransmission of packets containing errors, an isochronous transmission lets
the errors pass through uncorrected. The underlying philosophy is that a bad pixel in an
image is less objectionable than image frames that jerk because the flow of the data stream
stops for the retransmission of bad packets.
Bit-Coding
In putting bits on a data line, a simple question arises: What is a bit, anyway? Sure, we
know what of bit of information isits the smallest possible piece of information.
However, in a communications system, the issue is what represents each bit. It could be a
pulse of electricity, but it could equally be a piece of copper in the wire or a wildebeest or
a green-and-purple flying saucer with twin chrome air horns. Although some of these
codes might not be practical, they would be interesting.
Certainly the code will have to be an electrical signal of some sort. After all, thats what we
can move through wires. But whats a bit? What makes a digital one and what makes a
digital zero?
The obvious thing to do is make the two states of the voltage on the communication wire
correspond to the two digital states. In that way, the system signals a digital one as the
presence of a voltage and a zero as its absence.
To fit with the requirements of a serial data stream, each bit gets assigned a timed position
in the data stream called a bit-cell. In most basic form, if a bit appears in the cell, it indicates a digital one; if not, its a zero.
Easy enough. But such a design is prone to problems. A long string of digital ones would
result in a constant voltage. In asynchronous systems, thats a problem because equipment
can easily lose track of individual bits. For example, when two ones travel one after
another, the clock only needs to be accurate to 50 percent of the bit-length to properly
decipher both. With a hundred ones in a series, even a 1 percent inaccuracy will lose a
bit or more of data.
RTZ
To help keep the data properly clocked, most digital signals work by switching each bit
both on and off in each bit-cell. To indicate a digital one, the most basic of such systems
will switch on a voltage and then switch it off before the end of the bit-cell. Because the
signaling voltage rises and then falls back to zero in each data cell, this form of bit-coding
is called Return-to-Zero coding, abbreviated RTZ.
RZI
Related to RTZ is Return-to-Zero Inverted coding, abbreviated RZI. The only difference is
a result of the I in the designation, the inversion, which simply means that a pulse of electricity encodes a digital zero instead of a digital one. The voltage on the line is constant
except when interrupted by a pulse of data.
NRZ
Although RTZ and RZI bit-coding are simple, they are not efficient. They effectively
double the frequency of the data signal, thus increasing the required bandwidth for a
given rate of data transmission by a factor of two. Thats plain wasteful.
Engineers found a better way. Instead of encoding bits as a voltage level, they opted to
code data as the change in voltage. When a bit appears in the data stream, NRZ coding
shifts the voltage from whatever it is to its opposite. A bit could be a transition from zero
to high voltage or the transition back the other way, from high voltage to zero. The signal
then remains at that level until the next bit comes along. Because each bit does not cause
the signal to fall back to zero voltage, this system is called No Return-to-Zero bit-coding.
NRZI
A related form of signaling is No Return-to-Zero Inverted bit-coding. The inversion is
before the coding. That is, a lack of a data bit in a bit-cell causes the coded signal voltage
249
250
to change. If data appears in the bit-cell, no change occurs in the encoded signal. Note
that a zero in the code stream triggers each transition in the resulting NRZI code. A long
period of no data on an NRZI-encoded channel (a continuous stream of logical zeros)
results in a on-off pattern of voltages on the signal wires. This signal is essentially a square
wave, which looks exactly like a clock signal. Figure 8.7 shows how the transitions in an
NRZI signal encode data bits.
FIGURE 8.7
NRZI coding scheme
used by USB.
The NRZI signal is useful because it is self-clocking. That is, it allows the receiving system to regenerate the clock directly from the signal. For example, the square wave of a
stream of zeros acts as the clock signal. The receiver adjusts its timing to fit this interval.
It keeps timing even when a logical one in the signal results in no transition. When a new
transition occurs, the timer resets itself, making whatever small adjustment might be necessary to compensate for timing differences at the sending and receiving ends.
Bit-Stuffing
No matter the coding scheme, including NRZI, long periods of an unchanging signal can
still occur. With an NRZ signal, a string of zeros results in no signal. In NRZI, it takes a
string of digital ones, but it can happen nevertheless. And if it does, the result can be clock
problems and data errors.
For example, two back-to-back bit-cells of constant voltage can be properly decoded even
if the data clock is 50 percent off. But with 100 back-to-back data cells of constant voltage,
even being 1 percent off will result in an error. It all depends on the data.
Thats not a good situation to be in. Sometimes the data you are sending will automatically cause errors. An engineer designing such a system might as well hang up his hat and
jump from the locomotive.
But engineers are more clever than that. They have developed a novel way to cope with
any signal and ensure that constant voltage will never occur: break up the stream of data
that causes the uninterrupted voltage by adding extra bits that can be stripped out of the
signal later. Adding these removable extraneous bits results in a technology called bitstuffing.
Bit-stuffing can use any of a variety of algorithms to add extra bits. For example, bit-stuffing as used by the Universal Serial Port standard injects a zero after every continuous
stream of six logical ones. Consequently, a transition is guaranteed to occur at least every
seven clock cycles. When the receiver in a USB system detects a lack of transitions for six
cycles and then receives the transition of the seventh, it knows it has received a stuffed bit.
It resets its timer and discards the stuffed bit and then counts the voltage transition (or
lack of it) occurring at the next clock cycle to be the next data bit.
Group-Coding
Bit-coding is a true cipher in that it uses one symbol to code anothera bit of data is one
symbol, and the voltage on the communication line (or its change) is a different one.
Nothing about enciphering says, however, that one bit must equal one symbol. In fact,
encoding groups of bits as individual symbols is not only possible but also often a better
strategy. For obvious reasons, this technology is called group-coding.
Cant think of a situation where one symbol equals multiple bits would work? Think
about the alphabet. The ASCII codeASCII is short for the American Standard Code for
Information Interchangethats used for encoding letters of the alphabet in binary form
is exactly that. The letter H, for example, corresponds to the binary code 00101000 in
ASCII.
Certain properties of data communications systems and data storage systems make groupcoding the most efficient way of moving and storing information. Group-coding can
increase the speed of a communications system, and it can increase the capacity of a storage system.
One way that group-coding can increase data communication speed is when the communications medium allows you to use more than the two binary states to encode bits. For
example, if a telephone line is stable, predictable, and reliable enough, you might be able
to distinguish four different voltage levels on the line instead of just two (on and off). You
could use the four states each as an individual symbol to encode bit-patterns. This example allows for an easy code in which numbers encode the binary value, as shown in
Table 8.1.
TA B L E 8 . 1
A Simple Example of Symbol Encoding Using Binary Bit Patterns
Symbol
Bit-Pattern
00
01
10
11
251
252
Although this encoding process seems trivial, it actually doubles the speed of the communication channel. Send the symbols down the line at the highest speed it allows. When
you decode the symbols, you get twice as many bits. This kind of group-coding makes
high-speed modems possible.
In the language of telecommunications, the speed at which symbols move through a communications medium is measured in units of baud. One baud is a communications rate of
one symbol per second. The term baud was named after J.M.E. Baudot, a French telegraphy expert. His full name (Baudot) is used to describe a five-bit digital code used in teletype systems.
The relationship between the number of bits that can be coded for each baud and the
number of signal symbols required is geometric. The number of required symbols skyrockets as you try to code more data in every baud, as shown in Table 8.2.
TA B L E 8 . 2
Signals States Required to Encode Bits
Number of States
Bits per Baud
16
64
256
16
Although it can be more efficient, this form of group-coding has a big disadvantage. It
makes the transmitted data more vulnerable to error. An error in one bad baud causes
multiple bits of raw data to be wrong. Moreover, the more symbols you use in any communication medium, the more similar they are to one another. The smaller the differences
between symbols, the more likely errors will result from smaller disruptions to the signal.
With modems, this drawback translates into the need for the best possible connection.
Engineers use a similar group-coding technique when they want to optimize the storage
capacity of a medium. As with communications systems, storage systems suffer bandwidth
limitations. The electronics of a rotating storage system such as a hard disk drive (see
Chapter 17, Magnetic Storage) are tuned to operate at a particular frequency, which is
set by the speed the disk spins and how tightly bits can be packed on the disk. The packing of bits is a physical characteristic of the storage medium, which is made from particles
of a finite size. Instead of electrical signals, hard disks use magnetic fields to encode information. More specifically, the basic storage unit is the flux transition, the change in direction of the magnetic field. Flux transitions occur only between particles (or
agglomerations of particles). A medium can store no more data than the number of flux
transitions its magnetic particles can make.
The first disk systems recorded a series of regularly spaced flux transitions to form a clock.
An extra flux transition between two clock bits indicated a one in digital data. No added
flux transition between bits indicated a zero. This system essentially imposed an NRZ signal on the clock and was called frequency modulation because it detected data as the change
in frequency of the flux transitions. It was also called single-density recording.
The FM system wastes half of the flux transitions on the disk because of the need for the
clock signal. By eliminating the clock and using a coding system even more like NRZ,
disk capacities could be doubled. This technique, called Modified Frequency Modulation
recording (MFM) or double-density recording was once the most widely used coding system for computer hard disks and is still used by many computer floppy disk drives. Instead
of clock bits, digital ones are stored as a flux transition and zeros as the lack of a transition
within a given period. To prevent flux reversals from occurring too far apart, MFM differs
from pure NRZ coding by adding an extra flux reversal between consecutive zeros.
Modern disks use a variety of group-codings called Run Length Limited (RLL). The symbols used for coding data in RLL are bit-patterns, but patterns that are different from the
data to be recorded. The RLL patterns actually use more bits than the original data. RLL
increases data density by the careful choice of which bit-patterns to use.
RLL works by using a flux transition to indicate not individual bits but the change from
zero to one or from one to zero. If two ones occur in a row, theres no flux transition. If
eight ones occur in a row, there is still no flux transition. Similarly, two or eight zeros in a
row result in no flux transition. The trick in RLL is to choose bit-patterns that only have
long series of ones or zeros in them, resulting in fewer flux transitions.
The first common RLL system was called 2,7 RLL because it used a set of symbols, each
16-bits long, chosen so that no symbol in the code had fewer than two identical bits and
more than seven. Of the 65,564 possible 16-bit symbols, the 256 with the bit-patterns best
fitting the 2,7 RLL requirements encode the 256 possible byte values of the original data.
In other words, the RLL coding system picks only the bit-patterns that allow the highest
storage density from a wide-ranging group of potential symbols. The 16-bit code patterns
that do not enforce the 2,7 rule are made illegal and never appear in the data stream that
goes to the magnetic storage device. Although each symbol involves twice as many bits as
the original data, the flux transitions used by the code occur closer together, resulting in a
50 percent increase in the capacity of the storage system.
Other RLL systems, sometimes called Advanced RLL, work the same way but use different
code symbols. One system uses a different code that changes the bit-pattern so that the
number of sequential zeros is between three and nine. This system, known for obvious
253
254
reasons as 3,9 RLL or Advanced RLL, still uses an 8-to-16 bit code translation, but it
ensures that digital ones will never be closer than every four bits. As a result, it allows data
to be packed into flux transitions four times denser. The net gain is that information can
be stored about twice as densely with 3,9 RLL as ordinary double-density recording techniques.
Frames
The basic element of digital information in a serial communications system is the data
frame. Think of this as a time frame, the frame bracketing the information like a frame
surrounds a window. The bits of the digital code are assigned their value in accord with
their position in the frame. In a synchronous serial communications system, the frame
contains the bits of a digital code word. In asynchronous serial communications, the frame
also contains a word of data, but it has a greater significance. It is also the time interval in
which the clocks of the sending and receiving systems are assumed to be synchronized.
When an asynchronous receiver detects the start of a frame, it resets its clock and then
uses its clock to define the significance of each bit in the digital code within the frame. At
the start of the next frame, it resets its clock and starts timing the bits again.
The only problem with this system is that an asynchronous receiver needs to know when a
frame begins and ends. Synchronous receivers can always look to the clock to know, but
the asynchronous system has no such luxury. The trick to making asynchronous communications work is unambiguously defining the frame. Todays asynchronous systems use
start bits to mark the beginning of a frame and stop bits to mark its end. In the middle are a
group of data bits.
The start bit helps the asynchronous receiver find data in a sea of noise. In some systems,
the start bit is given a special identity. In most asynchronous systems, it is twice the length
of the other bits inside the frame. In others, the appearance of the bit itself is sufficient.
After all, without data, you would expect no pulses. When any pulse pops up, you might
expect it to be a start bit.
Each frame ends with one or more stop bits. They assure the receiver that the data in the
frame is complete. Most asynchronous communication systems allow for one, one and a
half, or two stop bits. Most systems use one because that length makes each frame shorter
(which, in turn, means that it takes a shorter time to transmit).
The number of data bits in a frame varies widely. In most asynchronous systems, there
will be from five to eight bits of data in each frame. If you plan to use a serial port to connect a modern serial device to your computer, your choices will usually be to use either
seven bits or eight bits, the latter being the most popular.
In addition, the data bits in the frame may be augmented by error-correction information
called a parity bit, which fits between the last bit of data and the stop bit. In modern serial
systems, any of five varieties of parity bits are sometimes used: odd, even, space, mark, and
none.
The value of the parity bit is keyed to the data bits. The serial transmitter counts the
number of digital ones in the data bits and determines whether this total is odd or even.
In the odd parity scheme, the transmitter will turn on the parity bit (making it a digital
one) only if the total number of digital ones in the data bits is odd. In even priority systems, the parity bit is set as one only if the data bits contain an even number of digital
ones. In mark parity, the parity bit is always a mark, a digital one. In space parity, the parity bit is always a space, a digital zero. With no parity, no parity bit is included in the digital frames, and the stop bits immediately follow the data bits.
By convention, the bits of serial data in each frame are sent least-significant bit first.
Subsequent bits follow in order of increasing significance. Figure 8.8 illustrates the contents of a single data frame that uses eight data bits and a single stop bit.
FIGURE 8.8
A serial data frame
with eight data bits
and one stop bit.
A frame corresponds to a single character. Taken alone, thats not a whole lot of information. A single character rarely suffices for anything except answering multiple-choice tests.
To make something meaningful, you combine a sequence of characters together to form
words and sentences.
The serial communications equivalent of a sentence is a packet. A packet is a standardized
group of characters, or frame, that makes up the smallest unit that conveys information
through the communications system.
As the name implies, a packet is a container for a message, such as a diplomatic packet or
envelope. The packet holds the data. In addition, in most packetized systems, the packet
also includes an address and, often, a description of its contents. Packets may also include
extra data to ensure the integrity of their contents (for example, an error-detection or
error-correction scheme of some sort). Figure 8.9 shows the constituents of a typical data
packet.
255
256
FIGURE 8.9
Constituents of a typical data packet.
The exact constituents of a packet depend on the communication protocol. In general,

however, all packets have much the same construction. They begin with a symbol or character string that allows systems listening in to the communication channel to recognize
the bit-pattern that follows as a packet.
Each packet bears an address that tells where it is bound. Devices listening in on the communication channel check the address. If it does not match their own or does not indicate
that the packet is being broadcast to all devicesin other words, the packet wears an
address equivalent to occupantthe device ignores the rest of the packet.
Communications equipment is courteous enough not to listen in on messages meant for
someone else.
Most packets include some kind of identifying information that tells the recipient what to
do with the data. For example, a packet may bear a market to distinguish commands from
ordinary data.
The bulk of the packet is made from the data being transmitted. Packets vary in size, and
hence, the amount of data that they may contain. Although there are no hard-and-fast
limits, most packets range from 256 to 2048 bytes.
Error-Handling
Of course, a perfect world would also have fairies and other benevolent spirits to help
usher the data along and protect it from all the evil imps and energies lurking about trying
to debase and disgrace the singular purity of serial transfer.
The world is, alas, not perfect, and the world of computers is even less so. Many misfortunes can befall the vulnerable serial data bit as it crawls through its connection. One of
the bits of a full byte of data may go astray, leaving a piece of data with a smaller value on
arrival as it had at departurea problem akin to shipping alcohol by a courier service
operated by dipsomaniacs. With the vacancy in the data stream, all the other bits will slip
up a place and assume new values. Or the opposite casein the spirit of electronic camaraderie, an otherwise well-meaning signal might adopt a stray bit like a child takes on a
kitten, only to later discover the miracle of pregnancy and a progeny of errors that ripple
through the communications stream, pushing all the bits backward. In either case, the
prognosis is not good. With this elementary form of serial communications, one mistaken
bit either way, and every byte that follows will be in error.
Establishing reliable serial communications means overcoming these bit-error problems

and many others as well. Thanks to some digital ingenuity, however, serial communications work and work wellwell enough that you and your computer can depend on them.
Because no communication channel is error free, most packets include error-detection or
error-correction information. The principle behind error detection is simple: Include
duplicate or redundant information that you can compare to the original. Because communication errors are random, they are unlikely to affect both of two copies of the transmitted data. Compare two copies sent along and if they do not match, you can be sure one
of them changed during transmission and became corrupted.
Many communications systems dont rely on complex error-correction algorithms as are
used in storage and high-quality memory systems. Communications systems have a luxury
storage systems do not: They can get a second chance. If an error occurs in transmission,
the system can try againand againuntil an error-free copy gets through.
As a function of communication protocol, packets are part of the software standard used
by the communication system. Even so, they are essential to making the hardwarethe
entire communication systemwork properly and reliably.
Control
In a perfect world, a single circuitnothing more than two wires, a signal line, and a
groundwould be all that was necessary to move this serial signal from the one place to
another without further ado. But such a serial connection would be like a fire hose squirting out a big, fat stream of bits without rhyme, reason, or nozzle. Theres nothing to control the flow of data. If the receiving end of the serial channel cant keep up with the
sending end, data might flood out every which way and drown the poor recipient.
Laugh if you want, but controlling the flow of data is a major issue of the designers of serial communications systems. Engineers working on the problem have come up with not
one but several solutions. Some add an extra connection just for controlling the data flow,
a technology called hardware flow control. Some reserve special character values in the data
stream to produce software flow control. And some take a further step and redefine the flow
of data into digestible chunks and provide a software-based control system built in to
these chunks. This system, quickly becoming the most popular, is termed packetized communications.
One difference between these three technologies is range related. Hardware flow control,
with its need for extra physical connections, naturally limits its application to short ranges,
typically connections between computers and peripherals. Adding the extra wires for control is a small burden when they only need to go a few feet. But the cost and complexity of
those wires quickly gets out of hand when distances stretch to miles and signals must go
through switching or routing centers. Software flow control works with any range because
it requires only a single data pathway. However, because it relies on some of the characters
257
258
in the data stream and only a few characters can be reserved without compromising the
capacity of the channel, it works best only in one-on-one communications. That is, software flow control works best between two devices. Packetized communication, however,
has greater versatility. By putting control information into packets, a packet-based system
can handle signals meant for a nearly unlimited number of devices. Hence, packetized
communications work to link together entire networks.
Hardware and software flow controls are usually reserved for local, direct connections.
The most important application for these techniques of flow control is in serial ports, simple links between your computer and its peripherals. single In these applications, flow
control is often termed handshaking.
In its most basic form, hardware handshaking adds one signal from the receiving device
back to the sending device. This signal has two states: on and off. One state says, I am
ready and willing to accept data from you. The other state says, Whoa! I cant take
another byte. Often these two signals are called ready and busy. The receiving device is in
control. When it sends the ready signal, the sending device knows it can send data and
does, if it has any ready. When the recipient sends the busy signal back, the sending
device stops and waits for the next ready signal.
Software handshaking uses special characters in the data stream to indicate whether they
should transfer data. The device receiving data normally sends out one character to indicate it can accept data and a different character to indicate it is busy and cannot accommodate more. In the most common connection system to use software handshakingthe
legacy serial porteither of the two characters pairs, XON/XOFF or ETX/ACK, is often
used.
In the XON/XOFF scheme, the XOFF character sent from a receiving device to the
sending device says to hold off sending data. This character is also sometimes called DC1
and has an ASCII value of 19 or 013(hex). It is sometimes called Control-S. When the
receiving device is ready to receive data again, it sends out XON, also known as DC3, to
the sender. This character has an ASCII value of 17 or 011(hex). It is sometimes called
Control-Q.
ETX/ACK works similarly. ETX, which is an abbreviation for end text, says to hold off on
sending more text. This character has an ASCII value of 3 (decimal or hexadecimal) and is
sometimes called Control-C. ACK, short for acknowledge, says to resume sending data. It
has an ASCII value of 6 (decimal or hexadecimal) and is sometimes called Control-F.
Theres no issue as to whether hardware or software flow control is better. Both work, and
thats all thats necessary. The important issue is what kind of flow control is used in
asingle given system. Both ends must use the same handshaking, even the same characters.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
Expansion Buses
Given an old house, closets packed tighter than the economy section of a
Boeing 737, and a charge card, most people will broach the idea of
adding to their homes. After all, when you need more room, nothing
seems more natural than building an addition. If you have more ambition than second thoughts, the most humble bungalow can become a
manse of overwhelming grandeur with a mortgage to match.
Making the same magic on your computer might not seem as natural,
but it is. Adding to your computer is easier than adding to your house
and less likely to lead to bankruptcy and divorcebecause, unlike a
house, computers are designed to be added to. They sprout connectors
and ports like a Chia pet does hair.
Computers are different, however, in that the best way to add to them is
not on the outside. Odd as it may seem, you can effectively increase the
power and capabilities of most computers by expanding them on the inside.
The design feature key to this interior growth is the expansion bus. By
plugging a card into your computers expansion bus, you can make the
machine into almost anything you want it to bewithin reason, of
course. You cant expect to turn your computer into a flying boat, but a
video production system, international communications center, cryptographic analyzer, and medical instrument interface are all easily within the
power of your expansion bus. Even if your aim is more modest, youll
need to become familiar with this bus. All the most common options for
customizing your computermodems, network interface cards, and television tunersplug into it. Even if your computer has a modem, sound,
and video built in to its motherboard, forever fixed and irremovable, these
features invisibly connect to the computers expansion bus.
262
The expansion bus is your computers electrical umbilical cord, a direct connection with
the computers logical circulatory system that enables whatever expansion brainchild you
have to link to your system. The purpose of the expansion bus is straightforward: It
enables you to plug things into the machine and, hopefully, enhance the computers
operation.
Background
Expansion buses are much more than the simple electrical connections you make when
plugging in a lamp. Through the bus circuits, your computer transfers not only electricity
but also information. Like all the data your computer must deal with, that information is
defined by a special coding in the sequence and pattern of digital bits.
The bus connection must flawlessly transfer that data. To prevent mistakes, every bus
design also includes extra signals to control the flow of that information, adjust its rate to
accommodate the speed limits of your computer and its expansion accessories, and adjust
the digital pattern itself to match design variations. Each different bus design takes it own
approach to the signals required for control and translation. Some use extra hardware signals to control transfers across the bus. Others use extensive signaling systems that package information into blocks or packets and route them across the bus by adding addresses
and identification bytes to these blocks.
Expansion bus standards define how your computer and the cards you add to it negotiate
their data transfers. As a result, the standard that your computers bus follows is a primary
determinant of what enhancement products work with it. Nearly all of todays computers
follow a single standard, the oft-mentioned Peripheral Component Interconnect (PCI), which
ensures that the most popular computer cards will work inside them.
Older computers often had two expansion buses: a high-speed bus (inevitably PCI) that
was the preferred expansion connection and a compatibility bus to accommodate older
add-in devices that frugal-minded computer owners refused to discard. The compatibility
bus followed the Industry Standard Architecture design, often called the PC bus. In addition, most new computers have an accelerated graphics port (AGP) for their video cards.
Although AGP is a high-speed variation on PCI, it is not compatible with ordinary expansion cards. PCI replaced ISA because of speed. PCI was about eight times faster than ISA.
But time has taken its toll, and engineers are now eyeing higher-performance alternatives
to PCI. Some of these, such as PCI-X, are straightforward extrapolations on the basic PCI
design. Others, such as HyperTransport and InfiniBand, represent new ways of thinking
about the interconnections inside a computer.
Notebook computers have their own style of expansion buses that use PC Cards rather
than the slide-in circuit boards used by PCI. This design follows the standards set by the
Personal Computer Memory Card International Association (PCMCIA). The basic
CHAPTER 9 EXPANSION BUSES
Background
PC Card design allows for performance and operation patterned after the old ISA design,
but the CardBus design expands on PC Card with PCI-style bus connections that are
backward compatible. That means a CardBus slot will accept either a CardBus or PC
Card expansion card (with one exceptionfat Type 3 cards wont fit in ordinary skinny
slots), so your old cards will happily work even in the newest of computers.
Unlike those of desktop computers, the expansion buses of notebook machines are externally accessibleyou dont have to open up the computer to slide in a card. External
upgrades are, in fact, a necessity for notebook computers in that cracking open the case of
a portable computer is about as messy as opening an egg and, for most people, an equally
irreversible process. In addition, both the PC Card and CardBus standards allow for hotswapping. That is, you can plug in a card or unplug it while the electricity is on to your
computer and the machine is running. The expansion slots inside desktop computers generally do not allow for hot-swapping.
Expansion board technology is mature, at least for now. The necessary standards are in place
and stable. You can take it for granted that you can slide a PCI expansion board in a PCI
slot and have it automatically configure itself and operate in your computer. You can even
select from a wealth of PC Card and CardBus modules for your notebook computerall
without any concern about compatibility. Although those statements sound so obvious as to
be trivial, reaching this point has been a long journey marked by pragmatism and insight,
cooperation and argument, and even battles between industry giants. At least 11 major standards for computer expansion have been the result, as summarized in Table 9.1.
TA B L E 9 . 1
Comparison of Expansion Bus Standards
Name
Date
Bus Width
Clock Speed
Addressing
PC bus
1981
8 bits
4.77MHz
1MB
ISA
1984
16 bits
8MHz
16MB
Micro Channel
1987
16/32 bits
10MHz
16MB/4GB
EISA
1988
32 bits
8MHz
4GB
VL bus
1992
32/64 bits
50MHz
4GB
PCI
1992
32/64 bits
33MHz
4GB
PCIX
1998
32/64 bits
66 or 133MHz
4GB
PC Card
1990
16 bits
8MHz
64MB
CardBus
1994
32 bits
33MHz
4GB
InfiniBand
2000
12 bits
2500MHz
Packetized
HyperTransport
2001
32 bits
1600MHz
Packetized
263
264
The PC bus, so called because it was the expansion bus used by the first IBM personal
computer, was little more than an extension of the connections from the 8088 microprocessor, with a few control signals thrown in. The AT bus was IBMs extrapolation on
the PC bus, adding the signals needed to match the 80286 microprocessor used in the
IBM personal computer AT (for Advanced Technology). Compaq Corporation altered the
design by moving memory to a separate bus when the company introduced its Deskpro
386 in 1987. This basic design was made into an official standard by the Institute of
Electrical and Electronic Engineers as Industry Standard Architecture. Micro Channel
Architecture represented IBMs take on a new, revolutionary bus design, but did not survive
poor marketing. EISA, short for Enhanced ISA, was an industry response to Micro
Channel, which found little acceptance. Neither of the new buses (Micro Channel nor
EISA) made a visible improvement on performance. The VESA Local Bus (or VL Bus, created by the Video Electronics Standards Association) did offer a visible improvement
because it allowed for faster video displays. It was a promising new bus standard, but it
was out-promoted (and out-developed) by Intels Peripheral Component Interconnect (PCI),
which became the most popular expansion bus ever used by the computer industry.
Peripheral Component Interconnect

No one might have predicted that Peripheral Component Interconnect would become the
most popular expansion bus in the world when Intel Corporation introduced it in July
1992. PCI was long awaited as a high-speed expansion bus specification, but the initial
announcement proved to be more and less than the industry hoped for. The first PCI
standard defined mandatory design rules, including hardware guidelines to help ensure
proper circuit operation of motherboards at high speeds with a minimum of design complications. It showed how to link together computer circuitsincluding the expansion
busfor high-speed operation but failed to provide a standard for the actual signals and
connections that would make it a real expansion bus.
The design first took the form of a true expansion bus with Intels development of PCI
Release 2.0 in May 1993, which added the missing pieces to create a full 64-bit expansion
bus. Although it got off to a rocky start when the initial chipsets for empowering it proved
flawed, computer-makers have almost universally announced support. The standard has
been revised many times since then. By the time this book appears, Version 3.0 should be
the reigning standard.
Architectural Overview
The original explicit purpose of the PCI design was to make the lives of those who engineer chipsets and motherboards easier. It wasnt so much an expansion bus as an interconnection system, hence its pompous name (Peripheral Component is just a haughty way of
saying chip, and Interconnect means simply link). And that is what PCI is meant to bea
fast and easy chip link.
Even when PCI was without pretensions of being a bus standard, its streamlined linking
capabilities held promise for revolutionizing computer designs. Whereas each new Intel
microprocessor family required the makers of chipsets and motherboards to completely
redesign their products with every new generation of microprocessor, PCI promised a
common standard, one independent of the microprocessor generation or family. As originally envisioned, PCI would allow designers to link together entire universes of processors, coprocessors, and support chips without glue logicthe pesky profusion of chips
needed to match the signals between different integrated circuitsusing a connection
whose speed was unfettered by frequency (and clock) limits. All computer chips that follow the PCI standard can be connected together on a circuit board without the need for
glue logic. In itself, this could lower computer prices by making designs more economical
while increasing reliability by minimizing the number of circuit components.
A key tenant of the PCI design is processor independence; that is, its circuits and signals
are not tied to the requirements of a specific microprocessor or family. Even though the
standard was developed by Intel, the PCI design is not limited to Intel microprocessors.
In fact, Apples PowerMac computers use PCI.
Bus Speed
PCI can operate synchronously or asynchronously. In the former case, the speed of operation of the PCI bus is dependent on the host microprocessors clock and PCI components
are synchronized with the host microprocessor. Typically the PCI bus will operate at a
fraction of the external interface of the host microprocessor. With todays high microprocessor speeds, however, the bus speed often is synchronized to the system bus or frontside bus, which may operate at 66, 100, or 133MHz. (The 400 and 533MHz buses used
by the latest Pentium 4 chips actually run at 100 and 133MHz, respectively, and ship multiple bytes per clock cycle to achieve their high data rates.) The PCI bus can operate at
speeds up to 66MHz under the revised PCI 2.2 (and later) standards. PCI-bus derivations,
such as PCI-Express, use this higher speed.
PCI is designed to maintain data integrity at operating speeds down to 0 Hz, a dead stop.
Although it wont pass data at 0 Hz, the design allows notebook computers to freely shift
to standby mode or suspend mode.
Although all PCI peripherals should be able to operate at 33MHz, the PCI design allows
you to connect slower peripherals. To accommodate PCI devices that cannot operate at
the full speed of the PCI bus, the design incorporates three flow-control signals that indicate when a given peripheral or board is ready to send or receive data. One of these signals halts the current transaction. Consequently, PCI transactions can take place at a rate
far lower than the maximum 33MHz bus speed implies.
265
266
The PCI design provides for expansion connectors extending the bus off the motherboard, but it limits such expansion to a maximum of three connectors (none are required
by the standard). As with VL bus, this limit is imposed by the high operating frequency of
the PCI bus. More connectors would increase bus capacitance and make full-speed operation less reliable.
To attain reliable operation at high speeds without the need for terminations (as required
by the SCSI bus), Intel chose a reflected rather than direct signaling system for PCI. To
activate a bus signal, a device raises (or lowers) the signal on the bus only to half its
required activation level. As with any bus, the high-frequency signals meant for the slots
propagate down the bus lines and are reflected back by the unterminated ends of the conductors. The reflected signal combines with the original signal, doubling its value up to
the required activation voltage.
The basic PCI interface requires only 47 discrete connections for slave boards (or
devices), with two more on bus-mastering boards. To accommodate multiple power supply
and ground signals and blanked off spaces to key the connectors for proper insertion, the
physical 32-bit PCI bus connector actually includes 124 pins. Every active signal on the
PCI bus is adjacent to (either next to or on the opposite side of the board from) a power
supply or ground signal to minimize extraneous radiation.
Multiplexing
Although the number of connections used by the PCI system sounds high, Intel actually
had to resort to a powerful trick to keep the number of bus pins manageable. The address
and data signals on the PCI bus are time-multiplexed on the same 32 pins. That is, the
address and data signals share the same bus connections (AD00 through AD31). On the
one clock cycle, the combined address/data lines carry the address values and set up the
location to move information to or from. On the next cycle, the same lines switch to carrying the actual data.
This address/data cycling of the bus does not slow the bus. Even in nonmultiplexed
designs, the address lines are used on one bus cycle and then the data lines are used on the
next. Moreover, PCI has its own burst mode that eliminates the need for alteration
between address and data cycles. PCI also can operate in its own burst mode. During
burst mode transfers, a single address cycle can be followed by multiple data cycles that
access sequential memory locations.
PCI achieves its multiplexing using a special bus signal called Cycle Frame (FRAME#).
The appearance of the Cycle Frame signal identifies the beginning of a transfer cycle and
indicates the address/data bus holds a valid address. The Cycle Frame signal is then held
active for the duration of the data transfer.
Burst Mode
During burst mode transfers, a single address cycle can be followed by multiple data
cycles that access sequential memory locations, limited only by the needs of other devices
to use the bus and other system functions (such as memory refresh). The burst can continue as long as the Cycle Frame signal remains active. With each clock cycle that Cycle
Frame is high, new data is placed on the bus. If Cycle Frame is active only for one data
cycle, an ordinary transfer takes place. When it stays active across multiple data cycles, a
burst occurs.
This burst mode underlies the 132MBps throughput claimed for the 32-bit PCI design.
(With the 64-bit extension, PCI claims a peak transfer rate of 264MBps.) Of course, PCI
attains that rate only during the burst. The initial address cycle steals away a bit of time
and lowers the data rate (the penalty for which declines with the increasing length of the
burst). System overhead, however, holds down the ultimate throughput.
PCI need not use all 32 (or 64) bits of the buss data lines. Four Byte Enable signals
(C/BE0# through C/BE3#) are used to indicate which of the four-byte-wide blocks of
PCIs 32-bit signals contain valid data. In 64-bit systems, another four signals (C/BE4#
through C/BE7#) indicate the additional active byte lanes.
To accommodate devices that cannot operate at the full speed of the PCI bus, the design
incorporates three flow-control signals: Initiator Ready (IRDY#, at pin B35), Target
Ready (TRDY#, at pin A36), and Stop (STOP#, at pin A38). Target Ready is activated to
indicate that a bus device is ready to supply data during a read cycle or accept it during a
write cycle. When Initiator Ready is activated, it signals that a bus master is ready to complete an ongoing transaction. A Stop signal is sent from a target device to a master to stop
the current transaction.
Data Integrity Signals

To ensure the integrity of information traversing the bus, the PCI specification makes
mandatory the parity-checking of both the address and data cycles. One bit (signal PAR) is
used to confirm parity across 32 address/data lines and the four associated Byte Enable
signals. A second parity signal is used in 64-bit implementations. The parity signal lags
the data it verifies by one cycle, and its state is set so that the sum of it, the address/data
values, and the Byte Enable values are a logical high (1).
If a parity error is detected during a data transfer, the bus controller asserts the Parity
Error signal (PERR#). The action taken on error detection (for example, resending data)
depends on how the system is configured. Another signal, System Error (SERR#) handles
address parity and other errors.
267
268
Parity-checking of the data bus becomes particularly important as the bus width and speed
grow. Every increase in bus complexity also raises the chance of errors creeping in. Paritychecking prevents such problems from affecting the information transferred across the bus.
Serialized Interrupts
Because of the design of the PCI system, the lack of the old IRQ signals poses a problem.
Under standard PCI architecture, the compatibility expansion bus (ISA) links to the host
microprocessor through the PCI bus and its host bridge. The IRQ signals cannot be
passed directly through this channel because the PCI specification does not define them.
To accommodate the old IRQ system under PCI architecture, several chipset and computer makers, including Compaq, Cirrus Logic, National Semiconductor, OPTi, Standard
Microsystems, Texas Instruments, and VLSI Technology, developed a standard they called
Serialized IRQ Support for PCI Systems.
The serialized IRQ system relies on a special signal called IRQSER that encodes all available interrupts as pulses in a series. One long series of pulses, called an IRQSER cycle,
sends data about the state of all interrupts in the system across the PCI channel.
The IRQSER cycle begins with an extended pulse of the IRQSER signal, lasting from
four to eight cycles of the PCI clock (each of which is nominally 33MHz but may be
slower in systems with slower bus clocks). After a delay of two PCI clock cycles, the
IRQSER cycle is divided into frames, each of which is three PCI clock cycles long. Each
frame encodes the state of one interruptif the IRQSER signal pulses during the first
third of the frame, it indicates the interrupt assigned to that frame is active. Table 9.2 lists
which interrupts are assigned to each frame position.
TA B L E 9 . 2
PCI Serialized Interrupt Frame Encoding
Interrupt Encoded
Frame Position
Clocks Past Start
IRQ0
IRQ1
SMI#
IRQ3
11
IRQ4
14
IRQ5
17
IRQ6
20
IRQ7
23
IRQ8
26
Interrupt Encoded
Frame Position
Clocks Past Start
IRQ9
10
29
IRQ10
11
32
IRQ11
12
35
IRQ12
13
38
IRQ13
14
41
IRQ14
15
44
IRQ15
16
47
IOCHCK#
17
50
INTA#
18
53
INTB#
19
56
INTC#
20
59
INTD#
21
62
Unassigned
22 to 32
65 to 95
In addition to the 16 IRQ signals used by the old interrupt system, the PCI serialized
interrupt scheme also carries data about the state of the system management interrupt
(SMI#) and the I/O check (IOCHCK#) signals as well as the four native PCI interrupts
and 10 unassigned values that may be used by system designers. According to the serialized interrupt scheme, support for the last 14 frames is optional.
The IRQSER cycle ends with a Stop signal, a pulse of the IRQSER signal that lasts two
or three PCI clocks, depending on the operating mode of the serialized interrupt system.
The PCI serialized interrupt system is only a means of data transportation. It carries the
information across the PCI bus and delivers it to the microprocessor and its support circuitry. The information about the old IRQ signals gets delivered to a conventional 8259A
interrupt controller or its equivalent in the microprocessor support chipset. Once at the
controller, the interrupts are handled conventionally.
Although the PCI interrupt-sharing scheme helps eliminate setup problems, some systems
demonstrate their own difficulties. For example, some computers force the video and
audio systems to share interrupts. Any video routine that generates an interrupt, such as
scrolling a window, will briefly halt the playing of audio. The audio effects can be unlistenable. The cure is to reassign one of the interrupts, if your system allows it.
269
270
Bus-Mastering and Arbitration

The basic PCI design supports arbitrated bus-mastering like that of other advanced
expansion buses, but PCI has its own bus command language (a four-bit code) and supports secondary cache memory.
In operation, a bus master board sends a signal to its host to request control of the bus
and starts to transfer when it receives a confirmation. Each PCI board gets its own slotspecific signals to request bus control and receive confirmation that control has been
granted. This approach allows great flexibility in assigning the priorities, even the arbitration protocol, of the complete computer system. The designer of a PCI-based computer
can adapt the arbitration procedure to suit his needs rather than having to adapt to the
ideas of the obscure engineers who conceived the original bus specification.
Bus mastering across the PCI bus is achieved with two special signals: Request (REQ#)
and Grant (GNT#). A master asserts its Request signal when it wants to take control of
the bus. In return, the central resource (Intels name for the circuitry shared by all bus
devices on the motherboard, including the bus control logic) sends a Grant signal to the
master to give it permission to take control. Each PCI device gets its own dedicated
Request and Grant signal.
As a self-contained expansion bus, PCI naturally provides for hardware interrupts. PCI
includes four level-sensitive interrupts (INTA# through INTD#, at pins A6, B7, A7, and
B8, respectively) that enable interrupt sharing. The specification does not itself define what
the interrupts are or how they are to be shared. Even the relationship between the four signals is left to the designer (for example, each can indicate its own interrupt, or they can
define up to 16 separate interrupts as binary values). Typically, these details are implemented in a device driver for the PCI board. The interrupt lines are not synchronized to
the other bus signals and may therefore be activated at any time during a bus cycle.
Low-Voltage Evolution
As the world shifts to lower-power systems and lower-voltage operation, PCI has been
adapted to fit. Although the early incarnations of the standard provided for 3.3-volt operation in addition to the then-standard 5-volt level, the acceptance of the lower voltage standard became official only with PCI version 2.3. Version 3.0 (not yet released at the time of
this writing) takes the next step and eliminates the 5-volt connector from the standard.
Slot Limits
High frequencies, radiation, and other electrical effects also conspire to limit the number
of expansion slots that can be attached in a given bus system. These limits become
especially apparent with local bus systems that operate at high clock speeds. All current
local bus standards limit to three the number of high-speed devices that can be connected
to a single bus.
Note that the limit is measured in devices and not slots. Many local bus systems use a
local bus connection for their motherboard-based display systems. These circuits count as
one local bus device, so computers with local bus video on the motherboard can offer at
most two local bus expansion slots.
The three-device limit results from speed considerations. The larger the bus, the higher
the capacitance between its circuits (because they have a longer distance over which to
interact). Every connector adds more capacitance. As speed increases, circuit capacitance
increasingly degrades its signals. The only way to overcome the capacitive losses is to start
with more signals. To keep local bus signals at reasonable levels and yet maintain high
speeds, the standards enforce the three-device limit.
A single computer can accommodate multiple PCI expansion buses bridged together to
allow more than three slots. Each of these sub-buses then uses its own bus-control circuitry. From an expansion standpointor from the standpoint of an expansion board
splitting the system into multiple buses makes no difference. The signals get where they
are supposed to, and thats all that counts. The only worries are for the engineer who has
to design the system to begin withand even that is no big deal. The chipset takes care of
most expansion bus issues.
Hot-Plugging
Standard PCI cards do not allow for hot-plugging. That is, you cannot and should not
remove or insert a PCI expansion board into a connector in a running computer. Try it,
and you risk damage to both the board and the computer.
In some applications, however, hot-plugging is desirable. For example, in fault-tolerant
computers you can replace defective boards without shutting down the host system. You
can also add new boards to a system while it is operating.
To facilitate using PCI expansion boards in such circumstances, engineers developed a
variation on the PCI standard called PCI Hot Plug and published a specification that
defines the requirements for expansion cards, computers, and their software to make it all
work. The specification, now available as revision 1.1, is a supplement to the ordinary PCI
Specification.
Unless you have all threethe board, computer, and softwareyou should not attempt
hot-plugging with PCI boards. Products that meet the standard are identifiable by a
special PCI Hot Plug logo.
271
272
Setup
PCI builds upon the Plug-and-Play system to automatically configure itself and the
devices connected to it without the need to set jumpers or DIP switches. Under the PCI
specification, expansion boards include Plug-and-Play registers to store configuration
information that can be tapped into for automatic configuration. The PCI setup system
requires 256 registers. This configuration space is tightly defined by the PCI specification
to ensure compatibility. A special signal, Initialization Device Select (IDSEL), dedicated
to each slot activates the configuration read and write operations as required by the
Plug-and-Play system.
PCI-X
In the late 1990s, engineers at Compaq, Hewlett-Packard, and IBM realized that microprocessor speeds were quickly outrunning the throughput capabilities of the PCI bus, so
they began to develop a new, higher-speed alternative aimed particularly at servers.
Working jointly, they developed a new bus specification, which they submitted to the PCI
Special Interest group in September 1998. After evaluating the specification for a year, in
September 1999, the PCI SIG adopted the specification and published it as the official
PCI-X Version 1.0 standard.
The PCI-X design not only increases the potential bus speed of PCI but also adds a new
kind of transfer that makes the higher speeds practical. On July 23, 2002, PCI-X Version
2.0 was released to provide a further upgrade path built upon PCI-X technology. Table
9.3 compares the speed and bandwidth available under various existing and proposed PCI
and PCI-X specifications.
TA B L E 9 . 3
PCI and PCI-X Implementations Compared
Specification
Bus Width
Bus Speed
Maximum Peak
Bandwidth
PCI .2.2
32 bits
33MHz
133Mbps
PCI 2.2
64 bits
33MHz
266Mbps
PCI 2.2
64 bits
66MHz
533Mbps
PCI-X 1.0
64 bits
133MHz
1066Mbps
PCI-X 2.0
64 bits
266MHz
2133Mbps
PCI-X 2.0
64 bits
533MHz
4266Mbps
The PCI-X design follows the PCI specification in regard to signal assignments on the
bus. It supports both 32-bit and 64-bit bus designs. In fact, a PCI card will function
normally in a PCI-X expansion slot, and a PCI-X expansion board will work in a standard
PCI slot, although at less than its full potential speed.
The speed of the PCI-X expansion bus is not always the 133MHz of the specifications. It
depends on how many slots are connected in a single circuit with the bus controller and
whats in the slots. To accommodate ordinary PCI boards, for example, all interconnected
PCI-X slots automatically slow to the highest speed the PCI board will accepttypically
33MHz. In addition, motherboard layouts limit the high-frequency capabilities of any bus
design. Under PCI-X, only a single slot is possible at 133MHz. Two-slot designs require
retarding the bus to 100MHz. With four slots, the practical top speed falls to 66MHz.
The higher speeds possible under PCI-X 2.0 keep the clock at 133MHz but switch to
double-clocking or quad-clocking the data to achieve higher transfer rates.
To reach even the 133MHz rate (and more reliable operation at 66MHz), PCI-X adds a
new twist to the bus design, known as register-to-register transfers. On the PCI bus, cards
read data directly from the bus, but the timing of signals on the bus leaves only a short
window for valid data. At 33MHz, a PCI card has a window of about 7 milliseconds to
read from the bus; at 66MHz, only about 3 milliseconds is available. In contrast, PCI-X
uses a register to latch the signals on the bus. When a signal appears on the bus, the PCIX registers lock down that value until data appears during the next clock cycle. A PCI-X
card therefore has a full clock cycle to read the dataabout 15 milliseconds at 66MHz
and about 7.5 milliseconds at 133MHz.
Although register-to-register transfers make higher bus speeds feasible, at any given speed
they actually slow the throughput of the bus by adding one cycle of delay to each transfer.
Because transfers require multiple clock cycles, however, the penalty is not great, especially for bursts of data. But PCI-X incorporates several improvements in transfer efficiency that help make the real-world throughput of the PCI-X bus actually higher than
PCI at the same speed.
One such addition is attribute phase, an additional phase in data transfers that allows
devices to send each other a 36-bit attribute field that adds a detailed description to the
transfer. The field contains a transaction byte count that describes the total size of the transfer to permit more efficient use of bursts. Transfers can be designated to allow relaxed
ordering so that transactions do not have to be completed in the order they are requested.
For example, a transfer of time-critical data can zoom around previously requested transfers. Relaxed ordering can keep video streaming without interruption. Transactions that
are flagged as non-cache coherent in the attribute field tell the system that it need not waste
time snooping through its cache for changes if the transfer wont affect the cache. By
applying a sequence number in the attribute field, bus transfers in the same sequence can be
managed together, which can improve the efficiency in caching algorithms.
273
274
In addition to the attribute phase, PCI-X allows for split transactions. If a device starts a
bus transaction but delays in finishing it, the bus controller can use the intervening idle
time for other transactions. The PCI-X bus also eliminates wait states (except for the
inevitable lag at the beginning of a transaction) through split transactions by disconnecting idle devices from the bus to free up bandwidth. All PCI-X transactions are one standard size that matches the 128-bit cache line used by Intel microprocessors, permitting
more efficient operation of the cache. When all the new features are taken together, the
net result is that PCI-X throughput may be 10 percent or more higher than using standard PCI technology.
Because PCI-X is designed for servers, the integrity of data and the system are paramount
concerns. Although PCI allows for parity-checking transfers across the bus, its error-handling mechanism is both simple and inelegantan error would shut down the host system. PCI-X allows for more graceful recovery. For example, the controller can request
that an erroneous transmission be repeated, it can reinitialize the device that erred or disable it entirely, or it can notify the operator that the error occurred.
PCI-X cards can operate at either 5.0 or 3.3 volts. However, high-speed operation is
allowed only at the lower operating voltage. The PCI-X design allows for 128 bus segments in a given computer system. A segment runs from the PCI-X controller to a PCI-X
bridge or between the bridge and an actual expansion bus encompassing one to four slots,
so a single PCI-X system can handle any practical number of devices.
PCI Express
According to the PCI Special Interest Group, the designated successor to todays PCI-X
(and PCI) expansion buses is PCI Express. Under development for years as 3GIO (indicating it was to be the third generation input/output bus design), PCI Express represents a
radical change from previous expansion bus architectures. Instead of using relatively lowspeed parallel data lines, it opts for high-speed serial signaling. Instead of operating as a
bus, it uses a switched design for point-to-point communications between deviceseach
gets the full bandwidth of the system during transfers. Instead of special signals for service
functions such as interrupts, it uses a packet-based system to exchange both data and commands. The PCI-SIG announced the first PCI Express specification at the same time as
PCI-X 2.0, on July 23, 2002.
In its initial implementation, PCI Express uses a four-wire interconnection system, two
wires each (a balanced pair) for separate sending and receiving channels. Each channel
operates at a speed of 2.5GHz, which yields a peak throughput of 200MBps (ignoring
packet overhead). The system uses the 8b/10b encoding scheme, which embeds the clock
in the data stream so that no additional clock signal is required. The initial design contemplates future increases in speed, up to 10GHz, the theoretic maximum speed that can
be achieved in standard copper circuits.
To accommodate devices that require higher data rates, PCI Express allows for multiple
lanes within a single channel. In effect, each lane is a parallel signal path between two
devices with its own four-wire connection. The PCI Express hardware divides the data
between the multiple lanes for transmission and reconstructs it at the other end of the
connection. The PCI Express specifications allow for channels with 1, 2, 4, 8, 12, 16, or
32 lanes (effectively boosting the speed of the connection by the equivalent factor). A 32lane system at todays 2.5GHz bus speed would deliver throughput of 6400MBps.
The switched design is integral to the high-speed operation of PCI Express. It eliminates
most of the electrical problems inherent in a bus, such as changes in termination and the
unpredictable loading that occur as different cards are installed. Each PCI Express expansion connector links a single circuit designed for high-speed operation, so sliding a card
into one slot affects no other. The wiring from each PCI Express connector runs directly
back to a single centralized switch that selects which device has access to the system functions, much like a bus controller in older PCI designs. This switch can be either a standalone circuit or a part of the host computers chipset.
All data and commands for PCI Express devices are contained in packets, which incorporate error correction to ensure the integrity of transfers. Even interrupts are packetized
using the Messaged Signal Interrupt system introduced with PCI version 2.2, the implementation of which is optional for ordinary PCI but mandatory for PCI Express. The
packets used by PCI Express use both 32-bit and extended 64-bit addressing.
The primary concern in creating PCI Express was to accommodate the conflicting needs
of compatibility while keeping up with advancing technology. Consequently, the designers
chose a layered approach with the top software layers designed to match current PCI protocols, while the lowest layer, the physical, permits multiple variations.
In the PCI Express scheme, there are five layers, designated as follows:
Config/OSHandles configuration at the operating system level based on the current PCI Plug-and-Play specifications for initializing, enumerating, and configuring
expansion devices.
S/WThis is the main software layer, which uses the same drivers as the standard
PCI bus. This is the main layer that interacts with the host operating system.
TransactionThis is a packet-based protocol for passing data between the devices.
This layer handles the send and receive functions.
Data LinkThis layer ensures the integrity of the transfers with full error-checking
using a cyclic redundancy check (CRC) code.
PhysicalThis is the PCI Express hardware itself.
275
276
The top two layers, Config/OS and S/W, require no change from ordinary PCI. In other
words, the high-speed innovations of PCI Express are invisible to the host computers
operating system.
As to the actual hardware, PCI Express retains the standard PCI board design and dimensions. It envisions dual-standard boards that have both conventional and high-speed PCI
Express connections. Such dual-standard boards are restricted to one or two lanes on a
extra edge connector thats collinear with the standard PCI expansion connector but
mounted between it and the back of the computer. Devices requiring a greater number of
lanes need a new PCI Express connector system.
Standards and Coordination

The PCI, PCI-X, and PCI Express standards are managed and maintained by the PCI
Special Interest Group. The latest revision of each specification is available from the following address:
PCI Special Interest Group
2575 NE Kathryn St. #17
Hillsboro, OR 97124
Fax: 503-693-8344
Web site: http://www.pcisig.com
InfiniBand
At one time, engineers (who are as unlikely to gamble as the parish priest) put their
money on InfiniBand Architecture as the likely successor to PCI and PCI-X as the next
interconnection standard for servers and, eventually, individual computers. The standard
had wide industry backing and proven advantages in moving data around inside computers. Its success seemed all but assured by its mixed parentage that pulled together factions
once as friendly as the Montagues and Capulets. But as this is written, some industry
insiders believe that the introduction of a new PCI variation, PCI Express, may usurp the
role originally reserved for InfiniBand.
InfiniBand Architecture (also known as IBA) marks a major design change. It hardly
resembles whats come to be known as an expansion bus. Instead of being loaded with
clocks and control signals as well as data links, InfiniBand more resembles a network wire.
Its stripped down to nothing but the connections carrying the data signals and, if necessary, some power lines to run peripherals. As with a network connection, InfiniBand packages data into packets. The control signals usually carried over that forest of extra bus
connections are put into packets, too. Moreover, InfiniBand has no shared bus but rather
operates as a switched fabric. In fact, InfiniBand sounds more like a network than an
InfiniBand
expansion system. In truth, its both. And it also operates as an interconnection system
that links together motherboard components.
Designed to overcome many of the inherent limitations of PCI and all other bus designs,
IBA uses data-switching technology instead of a shared bus. Not only does this design
increase the potential speed of each connection because the bandwidth is never shared, it
also allows more effective management of individual connected devices and almost unlimited scalability. That is, it has no device limit (as does PCI) but allows you to link together
as many peripherals as you want, both as expansion boards inside a computer and as external devices.
As with PCI, InfiniBand starts as an interconnection system to link components together
on motherboards. But the specification doesnt stop there. It defines a complete system
that includes expansion boards and an entire network.
The initial implementation of InfiniBand operates with a base frequency of 2.5GHz.
Because it is a packetized communication system, it suffers from some signaling overhead.
As a result, the maximum throughput of an InfiniBand circuit is 250MBps or 2.0Gbps.
The speed is limited by necessity. Affordable wiring systems simply cannot handle higher
rates.
To achieve better performance, InfiniBand defines a multicircuit communication system.
In effect, it allows a sort of parallel channel capable of boosting performance by a factor
of 12.
InfiniBand is a radical change for system expansion, as different from PCI as PCI was
from ISA. Good as it is, however, dont expect to find an InfiniBand expansion bus in new
personal computers. As an expansion bus, it is expensive, designed foremost for servers.
Youll find the physical dimensions for InfiniBand expansion in Chapter 30, Expansion
Boards.
As an interconnection system for motherboard circuits, however, InfiniBand will find its
way into personal computers, along with the next generation of Intel microprocessors,
those that use 64-bit architecture.
History
The shortcomings of PCI architecture were apparent to engineers almost by the time the
computer industry adopted the standard. It was quicker than ISA but not all that swift.
Within a few years, high-bandwidth applications outgrew the PCI design, resulting in the
AGP design for video subsystems (and eventually 2x, 4x, and 8x versions of AGP).
To the engineers working on InfiniBand, PCI-X was only a band-aid to the bandwidth
problems suffered by high-powered systems. Most major manufacturers were already
aware that PCI-X shared the same shortcomings as PCI, burdened with its heritage of
277
278
interrupts and hardware-mediated flow control. Increasing the PCI speed exacerbated the
problems rather than curing them.
Two separate groups sought to create a radically different expansion system to supercede
PCI. One group sought a standard that would be able to link all the devices in a company
without limit and without regard to cost. The group working on this initiative called it
Future I/O. Another group sought a lower-cost alternative that would break through the
barriers of the PCI design inside individual computers. This group called its initiative
Next Generation I/O.
On August 31, 2000, the two groups announced they had joined together to work on a
single new standard, which became InfiniBand. To create and support the new standard,
the two groups formed the InfiniBand Trade Association (ITA) with seven founding members serving as the organizations steering committee. These included Compaq, Dell,
Hewlett-Packard, IBM, Intel, Microsoft, and Sun Microsystems. Other companies from
the FI/O and NGI/O initiatives joined as sponsoring members, including 3Com, Adaptec,
Cisco, Fujitsu-Siemens, Hitachi, Lucent, NEC, and Nortel Networks. On November 30,
2000, the ITA announced that three new members joinedAgilent Technologies Inc.,
Brocade Communications Systems Inc., and EMC Corporation.
The group released the initial version of the InfiniBand Architecture specification (version
1.0) on October 23, 2000. The current version, 1.0.a, was released on June 19, 2001. The
two-volume specification is distributed in electronic form without cost from the
InfiniBand Trade Association Web site, www.infinibandta.com.
Communications
In traditional terms, InfiniBand is a serial communication system. Unlike other expansion
designs, it is not a bus but rather a switched fabric. That means it weaves together devices
that must communicate together, providing each one with a full-bandwidth channel.
InfiniBand works more like the telephone system than a traditional bus. A controller
routes the high-speed InfiniBand signals to the appropriate device. In technical terms, it is
a point-to-point interconnection system.
InfiniBand moves information as packets across its channel. As with other packet-based
designs, each block of data contains addressing, control signals, and information.
InfiniBand uses a packet design based on the Internet Protocol so that engineers will have
an easier time designing links between InfiniBand and external networks (including, of
course, the Internet itself).
All InfiniBand connections use the same signaling rate, 2.5GHz. Because of the packet
structure and advanced data coding used by the system, this signaling rate amounts to an
actual throughput of about 500MBps. The InfiniBand system achieves even higher
throughputs by moving to a form of parallel technology, moving its signals through 4 or
InfiniBand
12 separate connections simultaneously. Table 9.4 summarizes the three current transfer
speeds of the InfiniBand Architecture.
TA B L E 9 . 4
InfiniBand Peak Throughput Versus Bus Width
Peak Throughput
Signaling Rate
Wire Pairs
500MBps
2.5GHz
2GBps
2.5GHz
6GBps
2.5GHz
12
These connections are full-duplex, so devices can transfer information at the same data
rate in either direction. The signals are differential (the system uses two wires for each of
its signals) to help minimize noise and interference at the high frequencies it uses.
The InfiniBand design allows the system to use copper traces on a printed circuit board
like a conventional expansion bus. In addition, the same signaling scheme works on copper wires like a conventional networking system or through optical fiber.
As with modern interconnection designs, InfiniBand uses intelligent controllers that handle most of the work of passing information around so that it requires a minimum of
intervention from the host computer and its operating system. The controller packages
the data into packets, adds all the necessary routing and control information to each one,
and sends them on their way. The host computer or other device need only send raw data
to the controller, so it loses a minimal share of its processing power in data transfers.
The InfiniBand design is inherently modular. Its design enables you to add devices up to
the limit of a switch or to add more switches to increase its capacity without limit.
Individual switches create subnetworks, which exchange data through routers, much like a
conventional network.
Structure
InfiniBand is a system area network (SAN), which simply means a network that lives
inside a computer system. It can also reach outside the computer to act as a real network.
In any of its applicationsinternal network, expansion system, or true local network
InfiniBand uses the same signaling scheme and same protocol. In fact, it uses Internet
Protocol, version 6 (IPv6), the next generation of the protocol currently used by the
Internet, for addressing and routing data.
In InfiniBand terminology, a complete IBA system is a network. The individual endpoints
that connect to hardware devices, such as microprocessors and output devices, are called
nodes. The wiring and other hardware that ties the network together make up the IBA
fabric.
279
280
The hardware link between a device and the InfiniBand network that makes up a node is a
channel adapter. The channel adapter translates the logic signals of the device connected to
it into the form that will be passed along the InfiniBand fabric. For example, a channel
adapter may convert 32-bit parallel data into a serial data stream spread across four differential channels. It also includes enough intelligence to manage communications, providing
functions similar to the handshaking of a conventional serial connection.
InfiniBand uses two types of channel adapters: the Host Channel Adapter (HCA) and the
Target Channel Adapter (TCA). As the name implies, the HCA resides in the host computer. More specifically, it is typically built in to the north bridge in the computers
chipset to take full advantage of the hosts performance. (Traditionally, Ethernet would
link to a system through its south bridge and suffer the speed restrictions of the hosts
expansion bus circuitry.) The distinction serves only to identify the two ends of a communications channel; the HCA is a microprocessor or similar device. The TCA is an
input/output device such as a connection to a storage system. Either the HCA or the
TCA can originate and control the flow of data.
The InfiniBand fabric separates the system from conventional expansion buses. Instead of
using a bus structure to connect multiple devices into the expansion system, the
InfiniBand fabric uses a switch, which sets up a direct channel from one device to another.
Switching signals rather than busing them together ensures higher speedevery device
has available to it the full bandwidth of the system. The switch design also improves reliability. Because multiple connections are not shared, a problem in one device or channel
does not affect others. Based on the address information in the header of each packet, the
switch directs the packet toward its destination, a journey that may pass through several
switches.
A switch directs packets only within an InfiniBand network (or within a device such as a
server). The IPv6 nature of InfiniBands packets allows easy interfacing of an individual
InfiniBand network with external networks (which may, in turn, be linked to other
InfiniBand systems). A router serves as the connection between an InfiniBand network and
another network.
The physical connections within the InfiniBand network (between channel adapters,
switches, and routers) are termed links in IBA parlance. The fabric of an InfiniBand network may also include one or more repeaters that clean up and boost the signals in a link
to allow greater range.
In a simple InfiniBand system, two nodes connect through links to a switch. The links and
the switch are the fabric of the network. A complex InfiniBand system may involve multiple switches, routers, and repeaters knitted into a far-reaching web.
HyperTransport

The InfiniBand specification is managed and maintained by the InfiniBand Trade
Association. The latest revision of the specification is available from the following address:
InfiniBand Trade Association
5440 SW Westgate Drive, Suite 217
Portland, OR 97221
Telephone: 503-291-2565
Fax: 503-297-1090
E-mail: [email protected]
Web site: www.infinibandta.org
HyperTransport
The fastest way to send signals inside a computer is via HyperTransport, but
HyperTransport is not an expansion bus. It isnt a bus at all. Strictly defined, it is a pointto-point communications system. The HyperTransport specification makes no provision
for removable expansion boards. However, HyperTransport is destined to play a major
role in the expansion of high-performance computers because it ties the circuitry of the
computer together. For example, it can link the microprocessor to the bus control circuitry, serving as a bridge between the north and south bridges in the motherboard
chipset. It may also link the video system to the north bridge. Anywhere signals need to
move their quickest, HyperTransport can move them.
Quick? HyperTransport tops out with a claimed peak data rate of 12.8GBps. Thats well
beyond the bandwidth of todays microprocessor memory buses, the fastest connections
currently in use. It easily eclipses the InfiniBand Architecture connection system, which
peaks out at 2GBps.
But thats it. Unlike InfiniBand, which does everything from host expansion boards to
connect to the World Wide Web, HyperTransport is nothing but fast. It is limited to the
confines of a single circuit board, albeit a board that may span several square feet of real
estate. As far as your operating system or applications are concerned, HyperTransport is
invisible. It works without any change in a computers software and requires no special
drivers or other attention at all.
On the other hand, fast is enough. HyperTransport breaks the interconnection bottlenecks in common system designs. A system with HyperTransport may be several times
faster than one without it on chores that reach beyond the microprocessor and its
memory.
281
282
History
HyperTransport was the brainchild of Advanced Micro Devices (AMD). The Texas chipmaker created the interconnection design to complement its highest performance microprocessors, initially codenamed Hammer.
Originally named Lightning Data Transport, the design gained the HyperTransport name
when AMD shared its design with other companies (primarily the designers of peripherals) who were facing similar bandwidth problems. On July 23, 2001, AMD and a handful
of other interested companies created the HyperTransport Technology Consortium to
further standardize, develop, and promote the interface. The charter members of the
HyperTransport Consortium included Advanced Micro Devices, API NetWorks, Apple
Computers, PMC-Sierra, Cisco Systems, NVidia, and Sun Microsystems.
The Consortium publishes an official HyperTransport I/O Link Specification, which
is currently available for download without charge from the groups Web site at
www.hypertransport.org. The current version of the specification is 1.03 and was released
on October 10, 2001.
Performance
In terms of data transport speed, HyperTransport is the fastest interface currently in use.
Its top speed is nearly 100 times quicker than the standard PCI expansion bus and three
times faster than the InfiniBand specification currently allows. Grand as those numbers
look, they are also misleading. Both InfiniBand and HyperTransport are duplex interfaces,
which means they can transfer data in two directions simultaneously. Optimistic engineers
calculate throughput by adding the two duplex channels together. Information actually
moves from one device to another through the InfiniBand and HyperTransport interfaces
at half the maximum rate shown, at most. Overhead required by the packet-based interfaces eats up more of the actual throughput performance.
Not all HyperTransport systems operate at these hyperspeeds. Most run more slowly. In
any electronic design, higher speeds make parts layout more critical and add increased
worries about interference. Consequently, HyperTransport operates at a variety of speeds.
The current specification allows for six discrete speeds, only the fastest of which reaches
the maximum value.
The speed of a HyperTransport connection is measured in the number of bits transferred
through a single data path in a second. The actual clock speed of the connection on the
HyperTransport bus is 800MHz. The link achieves its higher bit-rates by transferring two
bits per clock cycle, much as double data-rate memory does. One transfer is keyed to each
the rise and fall of the clock signal. The six speeds are 400, 600, 800, 1000, 1200, and
1600Mbps.
HyperTransport
In addition, HyperTransport uses several connections in parallel for each of its links. It
achieves its maximum data rate only through its widest connection with 32 parallel circuits. The design also supports bus widths of 16, 8, 4, and 2 bits.
Table 9.5 lists the link speeds as well as the maximum bus throughputs in megabytes per
second at each of the allowed bus widths.
TA B L E 9 . 5
HyperTransport Peak Throughput As a Function of Bus Width and Speed
HyperTransport Link Width (In Bits)

Link Speed in Mbps
16
32
400
100
200
400
800
1600
600
150
300
600
1200
2400
800
200
400
800
1600
3200
1000
250
500
1000
2000
4000
1200
300
600
1200
2400
4800
1600
400
800
1600
3200
6400
Certainly, a wider bus gives a performance edge, but the wider connections also rapidly
increase in complexity because of the structure of the interface.
Structure
HyperTransport is a point-to-point communications system. Functionally, it is equivalent
to a wire with two ends. The signals on the HyperTransport channel cannot be shared
with multiple devices. However, a device can have several HyperTransport channels leading to several different peripheral devices.
The basic HyperTransport design is a duplex communications system based on differential
signals. Duplex means that HyperTransport uses separate channels for sending and receiving data. Differential means that HyperTransport uses two wires for each signal, sending
the same digital code down both wires at the same time but with the polarity of the two
signals opposite. The system registers the difference between the two signals. Any noise
picked up along the way should, in theory, be the same on the two lines because they are
so close together, so it does not register on the differential receiver.
HyperTransport uses multiple sets of differential wire pairs for each communications
channel. In its most basic form, HyperTransport uses two pairs for sending and two for
receiving, effectively creating two parallel paths in the channel. That is, the most basic
form of HyperTransport moves data two bits at a time. The HyperTransport specification
283
284
allows for channel widths of 2, 4, 8, 16, or 32 parallel paths to move data the corresponding number of bits at a time. Of course, the more data paths in a channel, the faster the
channel can move data. The two ends of the connection negotiate the width of the channel when they power up, and the negotiated width remains set until the next power up or
reset. The channel width cannot change dynamically.
The HyperTransport specification allows for asymmetrical channels. That is, the specification allows for systems that might have a 32-bit data path in one direction but only 2
bits in return.
Although HyperTransport moves data in packets, it also uses several control lines to help
reduce software overhead and maintain the integrity of the network. A clock signal (CLK)
indicates when data on the channel is valid and may be read. Channels 2-, 4-, or 8-bits
wide have single clock signals; 16-bit channels have two synchronized clock signals; 32-bit
channels have four clock signals. A control signal (CTL) indicates whether the bits on the
data lines are actually data or control information. Activating this signal indicates the data
on the channel is control information. The line can be activated even during a data
packet, allowing control information to take immediate control. A separate signal
(POWEROK) ensures the integrity of the connection, indicating that the power to the
system and the clock are operating properly. A reset signal (RESET#) does exactly what
youd expectit resets the system so that it reloads its configuration.
For compatibility with power-managed computers based on Intel architecture, the
HyperTransport interface also includes two signals to allow a connection to be set up
again after the system goes into standby or hibernate mode. The LDTSTOP# signal,
when present, enables the data link. When it shifts off, the link is disabled. The host system or a device can request the link be reenabled (which also reconfigures the link) with
the LDTREQ# signal. Table 9.6 summarizes the signals in a HyperTransport link.
TA B L E 9 . 6
HyperTransport Bus Signals
Mnemonic
Function
Width
CAD
Command, address, and

data lines
2, 4, 8, 16, or 32
CTL
Control
CLK
Clock
1, 2, or 4
PWROK
Power and clock okay
RESET#
Reset chain
LDTSTOP#
Enable/disable link
LDTREQ#
Request link enable
PCMCIA
The design of the HyperTransport mechanical bus is meant to minimize the number of
connections used because more connections need more power, generate more heat, and
reduce the overall reliability of the system. Even so, a HyperTransport channel requires a
large number of connectionseach CAD bit requires four conductors, a two-wire differential pair for each of the two duplex signals. A 32-bit HyperTransport channel therefore
requires 128 conductors for its CAD signals. At a minimum, the six control signals, which
are single ended, require six more conductors, increasing to nine in a 32-bit system
because of the four differential clock signals.
Each differential pair uses a variation on the low-voltage differential signaling (LVDS) standard. The differential voltage shift between the pairs of a signal is only 600 millivolts.

The HyperTransport specification is maintained by the HyperTransport Technology
Consortium, which can be reached at the following addresses:
HyperTransport Technology Consortium
1030 E. El Camino Real #447
Sunnyvale, CA 94087
Phone: 800-538-8450 (Ext. 47739)
Web site: www.hypertransport.org
PCMCIA
While the desktop remains a battlefield for bus designers, notebook computermakers
have selected a single standard to rally around: PC Card, promulgated by the Personal
Computer Memory Card International Association (PCMCIA). Moreover, the PC Card
bus is flexible and cooperative. Because it is operating system and device independent, you
can plug the same PC Card peripheral into a computer, Mac, Newton, or whatever the
next generation holds in store. The PC Card expansion system can cohabitate in a computer with a desktop bus such as ISA or PCI. Whats more, it will work in devices that
arent even computersfrom calculators to hair curlers, from CAD workstations to autoeverything cameras. Someday you may even find a PC Card lurking in your toaster oven
or your music synthesizer.
The PC Card system is self-configuring, so you do not have to deal with DIP switches,
fiddle with jumpers, or search for a reference diskette. PC Card differs from other bus
standards in that it allows for external expansionyou dont have to open up your computer to add a PC Card.
285
286
The design is so robust that you can insert or remove a PC Card with the power on without worrying that you will damage it, your computer, or the data stored on the card. In
other words, PC Cards are designed for hot-swapping. The system is designed to notify
your operating system what youve done, so it can reallocate its resources as you switch
cards. The operating system sets up its own rules and may complain if you switch cards
without first warning it, although the resulting damage accrues mostly to your pride.
The engineers who created the PC Card standard originally envisioned only a system that
put memory in credit-card formathence the name, Personal Computer Memory Card
International Association. The U.S.-based association drew on the work of the Japan
Electronics and Information Technology Industries Association, which had previously developed
four memory-only card specifications.
They saw PC Card as an easy way to put programs into portable computers without the
need for disk drives. Since then the standard has grown to embrace nearly any expansion
optionmodems, network interface cards, SCSI host adapters, video subsystems, all the
way to hard disk drives. Along the way, the PCMCIA and JEITIA standards merged, so
one specification now governs PC Cards worldwide.
The first release of the PC Card standard allowed only for memory and came in September
1990. In September 1991, it was updated to Release 2.0 to include storage and input/output devices as well as thicker cards. After some minor changes, the standard jumped to
Release 5.0 in February 1995. The big jump in numbering (from Release 2.1 to Release
5.0) put the PCMCIA and JEITIA number schemes on equal footingthe specification
was number 5.0 for both. Since that event, the numbering of the standard as issued by both
organizations has tracked, culminating in the current standard, Release 8.0, published in
April 2001. Note, too, the specification title changed from PCMCIA Release 2.1 to PC
Card Release 5.0. Table 9.7 summarizes the history of the PCMCIA standard.
TA B L E 9 . 7
History of the PCMCIA/JEITIA/PC Card Standard
Version
Introduction Date
Features
PCMCIA 1.0
June 1990
First PCMCIA Standard, JEITIA Version 4.0; memory

only.
PCMCIA 2.0
September 1991
I/O interface added; socket services introduced.
PCMCIA 2.01
November 1992
PC Card ATA specification; card services introduced.
PCMCIA 2.1
July 1993
Card and socket services enhanced.
PC Card 5.0
February 1995
Compatibility enhancements; low-voltage added,

CardBus introduced; power management.
PC Card 6.0
March 1997
Zoomed video; hot dock; streamlined configuration.
PCMCIA
Version
Introduction Date
Features
PC Card 6.1
April 1998
Small PC Card form factor added; Win32 bindings.
PC Card 7.0
February 1999
Windows NT bindings; memory paging.
PC Card 7.1
March 2000
OpenCable custom interface; DMA references

removed.
PC Card 7.2
November 2000
New maximum current requirements; new guidelines.
PC Card 8.0
April 2001
New low-voltage support; 3-watt maximum per card

limit.
The PCMCIA/JEITIA specifications cover every aspect of the cards you slide into your computer (or other devicethe standard envisions you using the same cards in other kinds of
electronic gear). It details the size and shape of cards, how they are protected, how they are
connected (including the physical connector), the firmware the cards use to identify themselves, and the software your computer wields to recognize and use the cards. Moreover, the
specifications dont describe a hard-and-fast hardware interface. The same connector and
software system can embrace a wide variety of different purposes and interfaces. The PC
Card was only the first of the interconnection schemes documented by the PCMCIA/JEITIA
specifications. By redefining the signals on the PC Card connector, a high-speed 32-bit
expansion bus called CardBus, a dedicated high-speed video system called Zoomed Video, and a
cable televisionoriented video system called OpenCable all fit within the system.
All types of PC Cards use the same 68-pin connector, whose contacts are arranged in
two parallel rows of 34 pins. The lines are spaced at 1.27 mm (0.050-inch) intervals
between rows and between adjacent pins in the same row. Male pins on the card engage
a single molded socket on the host. Well look at the PCMCIA packages in more detail
in Chapter 30.
To ensure proper powering up of the card, the pins are arranged so that the power and
ground connections are longer (3.6 mm) than the signal leads (3.2 mm). Because of their
greater length, power leads engage first so that potentially damaging signals are not
applied to unpowered circuits. The two pins (36 and 67) that signal that the card has been
inserted all the way are shorter (2.6 mm) than the signal leads.
PC Card
The original and still most popular of the PCMCIA card specifications is PC Card. Based
on the ISA design, the standard expansion bus at the time the specifications were written,
PC Card is a 16-bit interface. Unlike modern expansion buses, it lacks such advanced features as bus-mastering. However, the PC Card standard is more than the old bus. It is a
system designed specifically for removable cards that includes both hardware and software
support.
287
288
The hardware side of the standard sets its limits so high that engineers are unlikely to
bump into them. For example, it allows for a single system to use from 1 to 255 PCMCIA
adapters (that is, circuits that match the signals of PC Cards to the host). Up to 16 separate PC Card sockets can be connected to each adapter, so under the PCMCIA specifications you could potentially plug up to 4080 PC Cards into one system. Most computers
shortchange you a bit on thattypically you get only one or two slots.
The memory and I/O registers of each PC Card are individually mapped into a window in
the address range of the host device. The entire memory on a PC Card can be mapped
into a single large window (for simple memory expansion, for example), or it can be paged
through one or more windows. The PC Card itself determines the access method through
configuration information it stores in its own memory.
Signals and Operation

The PCMCIA specifications allow for three PC Card variations: memory-only (which
essentially conforms to the Release 1.0 standard), I/O cards, and multimedia cards.
When a PCMCIA connector hosts PC Cards, all but 10 pins of the standard 68 share
common functions between the memory and I/O-style cards. Four memory card signals
are differently defined for I/O cards (pins 16, 33, 62, and 63); three memory card signals
are modified for I/O functions (pins 18, 52, and 61); and three pins reserved on memory
cards are used by I/O cards (pins 44, 45, and 60).
The PC Card specifications allow for card implementations that use either 8- or 16-bit
data buses. In memory operations, two Card Enable signals (pins 7 and 42) set the bus
width; pin 7 enables even-numbered address bytes; and pin 42 enables odd bytes. All bytes
can be read by an 8-bit system by activating pin 7 but not 42 and toggling the lowest
address line (A0, pin 29) to step to the next byte.
Memory Control
The original PC Card specifications allow the use of 26 address lines, which permitted
direct addressing of up to 64MB of data. The memory areas on each card are independent. That is, each PC Card can define its own 64MB address range as its common memory.
In addition to common memory, each card has a second 64MB address space devoted to
attribute memory, which holds the cards setup information. The typical PC Card devotes
only a few kilobytes of this range to actual CIS storage.
Activating the Register Select signal (pin 61) shifts the 26 address lines normally used
to address common memory to specifying locations in attribute memory instead. The
address space assigned to attribute memory need not correspond to a block of memory
separate from common memory. To avoid the need for two distinct memory systems,
a PC Card can be designed so that activating the Register Select signal simply points
PCMCIA
to a block of common memory devoted to storing setup information. All PC Cards

limit access to attribute memory to an eight-bit link using the eight least-significant
data lines.
To accommodate more storage, the specifications provide address extension registers, a bankswitching system that enables the computer host to page through much larger memory
areas on PC Cards. A number of different address extension register options are permitted,
all the way up to 16 bits, which extends the total card common memory to four terabytes
(4TB, which is 2 to the 42nd powerthat is, 26 addresses and 65,536 possible pages).
Memory cards that use EPROM memory often require higher-than-normal voltages to
reprogram their chips. Pins 18 and 52 on the PCMCIA interface provide these voltages
when needed.
Data Transfers
To open or close access to data read from a PC Card, the host computer activates a signal
on the cards Output Enable line (pin 9). A Ready/Busy line (pin 16) on memory cards
allows the card to signal when it is busy processing and cannot accept a data-transfer
operation. The same pin is used on I/O cards to make interrupt requests to the host system. During setup, however, an I/O card can redefine pin 16 back to its Ready/Busy function. Since Release 2.0, memory or I/O PC Cards also can delay the completion of an
operation in progressin effect, slowing the host to accommodate the time needs of the
cardby activating an Extend Bus Cycle signal on pin 59.
The Write Protect pin (pin 33) relays the status of the write-protect switch on memory
cards to the computer host. On I/O cards, this pin indicates that a given I/O port has a
16-bit width.
The same 26 lines used for addressing common and attribute memory serve as port-selection addresses on I/O cards. Two pins, I/O read (44) and I/O write (45), signal that the
address pins will be used for identifying ports and whether the operation is a read or a write.
Unlike memory addresses, however, the I/O facilities available to all PC Cards in a system
share only one 67,108,864-byte (64MB) range of port addresses. Although small compared to total memory addressing, this allotment of I/O ports is indeed generous. When
the specification was first contrived, the most popular computer expansion bus (the AT
bus) allowed only 64KB of I/O ports, of which some systems recognized a mere 16KB.
Whether ports are 8 or 16 bit is indicated by the signal on pin 33.
I/O PC Cards each have a single interrupt request signal. The signal is mapped to one of
the computer interrupt lines by the computer host. In other words, the PC Card generates a generic interrupt, and it is the responsibility of the host computers software and
operation system to route each interrupt to the appropriate channel.
289
290
Configuration
The PC Card specifications provide for automatic setup of each card. When the host
computer activates the Reset signal on pin 58 of the PCMCIA socket, the card returns to
its preinitialization settings, with I/O cards returning to their power-on memory card
emulation. The host can then read from the cards memory to identify it and run the
appropriate configuration software.
When a PC Card is plugged into a slot, the host computers PCMCIA adapter circuitry
initially assumes that it is a memory card. The card defines itself as an I/O card through
its onboard CIS data, which the host computer reads upon initializing the PC Card.
Multimedia cards similarly identify themselves through their software and automatically
reconfigure the PC Card to accept their special signals.
Audio
An audio output line also is available from I/O PC Cards. This connection is not intended
for high-quality sound, however, because it allows only binary digital (on/off) signals
(much like the basic speaker in early desktop computers). The audio lines of all PC Cards
in a system are linked together by an XOR (exclusive OR) logic gate fed to a single common loudspeaker, equivalent to the sound system of a primitive computer lacking a sound
board. Under the CardBus standard (discussed later in this chapter) the same socket supports high-quality digital audio.
Power
Pins 62 and 63 on memory cards output two battery status signals. Pin 63 indicates the
status of the battery: When activated, the battery is in good condition; when not activated,
it indicates that the battery needs to be replaced. Pin 62 refines this to indicate that the
battery level is sufficient to maintain card memory without errors; if this signal is not activated, it indicates that the integrity of on-card memory may already be compromised by
low battery power.
Although the PCMCIA specifications were originally written to accommodate only the 5volt TTL signaling that was the industry standard at the time of their conception, the
standard has evolved. It now supports cards using 3.3-volt signaling and makes provisions
for future lower operating voltages. Cards are protected so you cannot harm a card or
your computer by mismatching voltages.
Software Interface
The PC Card specifications require layers of software to ensure its compatibility across
device architectures. Most computers, for example, require two layers of software driversSocket Services and Card Servicesto match the card slots in addition to whatever
drivers an individual card requires (for example, a modem requires a modem driver).
PCMCIA
As with other hardware advances, Windows 95 helped out the PC Card. Support for the
expansion standard is built in to the operating system (as well as Windows 98). Although
Windows NT (through version 4.0) lacks integral PC Card support, Windows 2000
embraced the standard. Most systems accept cards after you slide them in, automatically
installing the required software. Unusual or obscure boards usually require only that you
install drivers for their manufacturer-specific features using the built-in Windows installation process.
Socket Services
To link the PC Card to an Intel-architecture computer host, PCMCIA has defined a software interface called Socket Services. By using a set of function calls under interrupt 1A
(which Socket Services shares with the CMOS real-time clock), software can access PC
Card features without specific knowledge of the underlying hardware. In other words,
Socket Services make access to the PC Card hardware independent, much like the BIOS
of a computer. In fact, Socket Services are designed so that they can be built in to the
computer BIOS. However, Socket Services also can be implemented in the form of a
device driver so that PCMCIA functionality can be added to existing computers.
Using Socket Services, the host establishes the windows used by the PC Card for access.
Memory or registers then can be directly addressed by the host. Alternatively, individual
or multiple bytes can be read or written through Socket Services function calls.
Card Services
With Release 2.01, PCMCIA approved a Card Services standard that defines a program
interface for accessing PC Cards. This standard establishes a set of program calls that link
to those Socket Services independent of the host operating system. Like the Socket
Services associated with interrupt 1A, Card Services can either be implemented as a driver
or be built in as part of an operating system. (Protected-mode operating systems such as
OS/2 and Windows NT require the latter implementation.)
Setup
For an advanced PC Card system to work effectively, each PC Card must be able to identify itself and its characteristics to its computer host. Specifically, it must be able to tell the
computer how much storage it contains; the device type (solid-state memory, disk, I/O
devices, or other peripherals); the format of the data; the speed capabilities of the card;
and any of a multitude of other variables about how the card operates.
Card Identification Structure
Asking you to enter all the required data every time you install a PC Card would be both
inconvenient and dangerous. Considerable typing would be required, and a single errant
keystroke could forever erase the data off the card. Therefore, PCMCIA developed a selfcontained system through which the basic card setup information can be passed to the
291
292
host regardless of either the data structure of the on-card storage or the operating system
of the host.
Called the Card Identification Structure (CIS) or metaformat of the card, the PCMCIA configuration system works through a succession of compatibility layers to establish the necessary link between the PC Card and its host. As with the hardware interface, each layer
of CIS is increasingly device specific.
Only the first layer, the Basic Compatibility Layer, is mandatory. This layer indicates how
the cards storage is organized. Only two kinds of information are relevant here: the data
structures used by the layer itself and such standard and physical device information as the
number of heads, cylinders, and sectors of a physical or emulated disk. Since Release 5.0,
each card must bear a Card Information Structure so it can identify itself to the host system.
Data Recording Format Layer
The next layer up is the Data Recording Format Layer, which specifies how the stored
data is organized at the block level. Four data formats are supported under Release 2.0:
unchecked blocks, blocks with checksum error correction, blocks with cyclic redundancy
error checking, and unblocked data that does not correspond to disk organization (for
example, random access to the data, such as is permitted for memory).
Data Organization Layer
The third CIS layer, the Data Organization Layer, specifies how information is logically
organized on the card; that is, it specifies the operating system format to which the data
conforms. PCMCIA recognizes four possibilities: DOS, Microsofts Flash File System for
Flash RAM, PCMCIAs own Execute-in-Place (XIP) ROM image, and application-specific
organization. Microsofts Flash File System is an operating system specifically designed for
the constraints of Flash memory. It minimizes rewriting specific memory areas to extend
the limited life of the medium and to allow for speedy updates of required block writes.
System-Specific Standards
The fourth CIS layer is assigned to system-specific standards that comply with particular
operating environments. For example, the Execute-in-Place (XIP) standard defines how
programs encoded on ROM cards are to be read and executed.
XIP is PCMCIAs own specification that allows program code in read-only memory to
execute without being first loaded into main system (read/write) memory. Applicationspecific organization allows card developers to create data organizations unique to their
products so as to implement special features.
Attribute Memory
The setup information for all these layers is stored in a reserved area on the PC Card
called attribute memory. This area is isolated from the cards ordinary storage, which under
PCMCIA
PCMCIA 2.1 is called common memory. The CIS information is structured as a linked
chain of data blocks called tuples, each of which can be up to 128 bytes long. To give all
systems a common starting point to search for CIS data, the first tuple of the metaformat
is located at the first address in attribute memory. This ensures that the data is within the
addressing range of even those primitive microprocessors that can address only 1MB of
RAM. Because the CIS system must work in any computer or other host, it assumes that
memory can be accessed only in byte widths.
The first two bytes of each tuple, as well as the format of many predefined tuples, are
strictly defined. The first byte encodes the function of the tuple and the parameters it
describes. The second byte links to the next tuple in the chain (if any); it specifies the
number of data bytes in the tuple, which, of course, indicates where the next tuple begins.
The PCMCIA 2.1 specifications define the options available for many common tuples. PC
Card manufacturers are free to add their own tuples to store data for setting up cards that
contain proprietary features.
As the storage and expansion needs of computers and other electronic devices continue to
evolve, the PCMCIA PC Card standard will likely follow in lockstep. Undoubtedly it is
the computer expansion system of the futureand the first truly universal data-interchange system.
CardBus
To bring the PC Card into the 32-bit world, PCMCIA adapted the highly regarded PCI
expansion bus to the credit-card format to create the CardBus in November 1994, formalized in Release 5.0. Just as PC Card is a miniaturized version of ISA, CardBus shrinks
down PCI while yielding the same high data-transfer rates, with one important exception:
Due to limitations in the connector design, CardBus extends only to a 32-bit bus width,
whereas the PCI standard allows for 64-bit buses. Although CardBus is not truly a PCI
systemit is designed to be a platform-independent 32-bit systemit is functionally
equivalent to PCI and uses the PCI protocol for bus operations.
Under CardBus, effectively all the active signals on the 68-pin PCMCIA connector are
redefined. Only the grounds, fixed voltages, and card-identification pins are unchanged.
To accommodate the full 32-bit addressing range and 32-bit data buses of the PCI bus,
the CardBus system multiplexes the two onto the same 32 connections of the PC Card,
just as they are in the full-size PCI implementation. In addition, the CardBus standard
boosts the signaling rate to match that of PCI, up to 33MHz, while allowing lower-speed
operation as well (all the way down to zero).
Besides a wider, faster bus and PCI-based protocol, CardBus adds several new features to
the PC Card repertory. CardBus supports arbitration and bus-mastering with multiple bus
293
294
masters, again patterned after PCI. A CardBus device can take command of its own transfers to the host system and free the host computers microprocessor from the chore.
CardBus also incorporates a new pulse width modulation (PWM) audio mode. This mode
allows a CardBus device to transfer audio in digital form across a single bus line. As with
all digital audio, the PWM signal is a series of pulses; however, the length of each pulse
rather than a coded pattern determines its audio equivalent. PWM mode allows the
CardBus device to transfer multiple high-quality audio signals in real time.
All CardBus cards operate at 3.3 volts, in keeping with the trend toward lower-voltage
operation in both desktop and notebook computers. The CardBus cards take advantage of
the PC Card voltage keying system so that you cannot insert a CardBus product into a PC
Card slot that supports only 5-volt operation (and thereby damaging the card, your computer, or both).
The low operating voltage helps ensure minimal power drain on the computer host. The
configuration system, too, was designed with power in mind. When a CardBus product
starts, only a limited part of its circuitry gets powered up, only enough to allow the host
computer to read the Card Identification Structure. The host can then determine what
kind of card it is, how much power it requires, and whether it has enough resources to
operate the card. Only when the host computer determines the compatibility of the
CardBus product and accepts it, does full power flow into the card. Besides allowing the
host system to reject products that might otherwise drain its energy resources, the
stepped startup procedure also prevents a surge in power demand when the host computer boots up.
One primary design intention with CardBus was continuity with the PC Card standard.
PCMCIA wanted CardBus and PC Cards to be as compatible with one another as
possible.
One incompatibility cannot be breeched. Because CardBus supports only 3.3-volt operation, its slots are off-limits to 5-volt-only PC Cards. The voltage key prevents you from
inserting 5-volt PC Cards into CardBus slots. Except for the voltage issue, the two types
of slots accept cards made to either standard. That is, all 3.3-volt PC Cards fit in and
work with CardBus slots.
The primary challenge in making CardBus backward compatible with PC Card was fitting
all the necessary 32 bits of addressing and 32 bits of data signaling onto a 68-pin connector while still having enough connections for support functions. The simple design expedient was the same as used by PCImultiplexing. CardBus connections do double-duty,
handling both address and data signals, cutting the number of pins required in half and
making a 32-bit bus on a 68-pin card viable.
PCMCIA
Zoomed Video
For higher-quality sound or video applications, the PC Card standard defines special multimedia cards. When the PC Card software in your computer recognizes a multimedia
card, it redefines many of the connections in the card connector to switch it into operating as a Zoomed Video port.
Specifically in multimedia mode, 21 address signals and three control signals get replaced
by four audio and 19 video signals as well as a new control signal. The video signals are
meant to write directly to the frame buffer of the host computer, allowing a PC Card to
generate quick displays. In effect, whatever device is on the PC Card with Zoomed Video
has direct access to your notebook computers screen. It can push pixels without the delays
inherent in bus transfers.
In addition, a Zoomed Video connection also gives a PC Card a direct connection to the
audio system in the host computer. The card provides pulse code modulation (PCM) digital audio directly to a codec in the host computer, which in turn links to the sound system.
The PC Card connection is only part of the Zoomed Video system. The host computer
must also be designed to give access to its frame buffer from the port connector.
Despite its appealing name, Zoomed Video does not make the ordinary displays on your
computer screen any faster. The only time it helps the display speed is when you have a
graphics accelerator or other high-performance video device that installs on a PC Card or
CardBus card.

One key to the PC Cards success is that it is not a proprietary standard foisted on the
industry by a single company or small coterie. The design is the product of a group called
the Personal Computer Memory Card International Association (PCMCIA), which has
more than 220 members involved in all aspects of the computer and electronics industry.
CardBus continues in that tradition and is similarly maintained. The various PCMCIA
standards are completely open, and all specifications are available to anyone requesting
them from the organization. (The charge for the latest specification release was not set at
the time of this writing, but it is estimated to be between $200 and $300.)
Rather than operate completely independently, PCMCIA cooperates with other standardsetting organizations. For example, by working jointly with the Japan Electronic Industry
Technology Association (JEITA) in December 1989, PCMCIA was able to ensure that the
standards it developed would be truly international in scope. Today, each organization
sends a delegation to the meetings of the other.
Besides the efforts at extending PCMCIA to future technologies, such as 32-bit data paths
and bus-mastering, PCMCIA is also developing standards for incorporating specific
295
296
device types into the system. Already the group has fully described the needs for XIP,
which allows programs to execute from their storage locations on PC Cards instead of
needing to be loaded as if from disk into normal system RAM. PCMCIA has also developed standards for linking AT Attachmentstyle IDE hard disks into PC Card sockets.
To obtain more information about PCMCIA products and standards, contact the following address:
Personal Computer Memory Card International Association
1030 East Duane Avenue, Suite G
Sunnyvale, CA 94086
Phone: 408-720-0107
Fax: 408-720-9416
Web site: www.pc-card.com
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
10
Interfaces
If you could pry one of the expansion bus connectors off the motherboard of your computer and solder a length of cable to it, youd probably
ruin your computer. But youd also make the equivalent of a mass storage interface. This may be educational, but dont try it at home. Not
only will it be expensive and wasteful (of your computer), but unnecessary. Some clever engineers have already done it for you.
In fact, the most popular interface for hard disk, CD, and DVD drives, is
nothing more than a pried-off expansion board connector. The AT
Attachment interface earned its name by putting the AT expansion bus
connector on a short cable that plugged into hard disk drives. The interface has matured somewhat since thenits about 16 times faster, has
developed its own language, and trimmed off a lot of fat. But deep down,
underneath all the gloss and the layers of specifications, is the heart of
the AT expansion bus.
At that, AT Attachment is almost the perfect interface. It brings devices
both electrically and logically closer to your computers circuitry than
any other connection system. And thats exactly what the interface is supposed to doconnect individual devices to your computer.
AT Attachment isnt the only popular interface, however. It has a number of rivals and predecessors. If you have a high-performance computer
and disk system, youre likely to covet its closet rival, SCSI, for its speed
and versatilityor even ATAs designated successor, Serial ATA, which
promises to be the fastest yet lowest cost interface developed so far.
At the other end of the spectrum, your computer likely also has the
slowest interface in common use still stuffed within its circuitry
300
a floppy disk interface. You probably dont need an engineering degree to figure out what
it does. It runs the floppy disk drive that is the one remaining (or at least the most prominent) vestige of the technology of the first personal computer in modern machines. Its the
equivalent of you wearing your grandmothers locket around your neck.
AT Attachment
The dominant hard disk interface for the computer has become the AT Attachment
designa natural outcome considering that the interface is based on the ISA expansion
bus, at one time the most widely used computer bus. Despite its wide application, you still
might never have heard of the AT Attachment name. It goes under more aliases than a
candidate for the Ten Most Wanted list. The most familiar of its alternate monikers is
IDE, which stands for Integrated Drive Electronics and is a term that correctly describes a
technology rather than a specific interface. Nevertheless, the name IDE gets used indiscriminately by people in the computer industry for the AT Attachment design. Sometimes
engineers call ATA the AT interface, which strictly speaking it is not.
Although engineers created ATA in its original form for low-cost small-capacity drives, it
has grown along with the needs of modern computers, keeping in step with the need for
both higher capacities and faster access to information. In fact, todays ATA interface can
handle more capacity than any disk drive foreseeable in your lifetime. And its speed far
exceeds the capabilities of any current disk drive mechanism.
But AT Attachment isnt a single interface. During its long evolution since its introduction
in 1991 (several preliminary versions were used by products before then and after 1985,
when engineers first began work on what eventually became the standard), it has undergone several stages of refinement. Every year or two, a new version of the standard gets
approved. The current version is ATA-6, although ATA-7 is already in the works. Table
10.1 briefly summarizes the history of the AT Attachment standard.
TA B L E 1 0 . 1
The ATA Standards
Standard
Year Introduced
Maximum Capacity
Maximum Speed
ATA
1994
504MB (CHS only)
4.17Mbps
ATA-2
1996
137.5GB
11.3Mbps
ATA-3
1997
137.5GB
16.7Mbps
ATA-4
1998
137.5GB
33.3Mbps
ATA-5
2000
137.5GB
66.6Mbps
ATA-6
2002
16PB
133Mbps
ATA-7
Proposed
16PB
133Mbps
CHAPTER 10 INTERFACES
AT Attachment
Performance
Much of the development of ATA has been focused on improving its performance to keep
it in step with disk drive and computer technology. The designers of the interface have
done an admirable job, not only keeping it ahead of current technology but also pushing
far beyond it. Todays hard disk drive mechanisms are limited by physical and design factors to far lower speeds than the ATA interface can deliver.
As engineers refined the ATA standard, they increased performance by adding new transfer
modes to the operating features of the drive. Each new drive usually supports the fastest
mode defined at the time it is made as well as all the older, slower modes. The host computer, its operating system, and the drive negotiate the fastest transfer mode that all
understand to use for their transfers. That way, you always get the fastest possible performance from your disk drive. Moreover, even the newest drives can work with old computers (or old drives can work with new computers).
The current ATA standard recognizes 11 transfer modes, each with its own performance
limit. Table 10.2 summarizes the principal transfer modes of the AT Attachment standard.
TA B L E 1 0 . 2
AT Attachment Transfer Modes and Speeds
Transfer Mode
Cycle Time (Nanoseconds)
Speed (Mbps)
Standard
PIO mode 0
600
1.67
ATA
PIO mode 1
383
2.61
ATA
PIO mode 2
240
4.17
ATA
PIO mode 3
180
11.1
ATA-2
PIO mode 4
120
16.7
ATA-3
DMA, single-word, mode 0
960
1.04
ATA
480
2.08
ATA
240
4.17
ATA
DMA, multi-word, mode 0
480
4.17
ATA
150
13.3
ATA-2
120
16.7
ATA-3
UltraDMA, mode 0
235
16
ATA-4
UltraDMA, mode 1
160
24
ATA-4
UltraDMA, mode 2
120
33.3
ATA-4
UltraDMA, mode 3
90
45
ATA-5
continues
301
302
TA B L E 1 0 . 2
Continued
Transfer Mode
Speed (Mbps)
Standard
UltraDMA, mode 4
60
66.6
ATA-5
UltraDMA, mode 5
40
100
ATA-6
UltraDMA, mode 6
30
133
ATA-7
AT Attachment supports two broad classes of transfers: Programmed Input/Output (PIO)

and Direct Memory Access (DMA). Under the more recent iterations of the ATA standard, DMA is further divided into Single-Word DMA, Multi-Word DMA, and
UltraDMA (once called synchronous DMA) modes. With todays computers, UDMA modes
are preferred for their higher performance.
The difference between PIO and DMA modes is how they use the resources in your computer. The DMA modes provide an opportunity for improved overall system performance
through bus-mastering, but the full potential of this benefit requires matching hardware
and software in your computer. The fastest modes match only with the most recent computers.
Programmed Input/Output puts your microprocessor to work. The microprocessor in
your system directly controls every byte that moves through the interface. The microprocessor directly writes values from its registers or memory to a special I/O port, which
transfers the data to the control circuitry of the interface. PIO transfers can be of one of
two types: blind and flow-control.
Blind transfers dont give the AT Attachment drive full control of the transfers. It has no
information about the ability of the microprocessor host to accept data. In effect, it is
blind to the capabilities of the host. By design, any error the system makes will be on the
conservative side to maximize the reliability of the system. Because sometimes only a part
of the full bandwidth of your computer is available for making disk transfers, blind transfers throttle back to the worst case. Consequently, AT Attachment operates blind transfers
at a slow rate regardless of the host computers ability to capture the data. It moves bytes
at the lower, throttled rate even when 100 percent of the bandwidth is available.
Flow-control transfers use a form of handshaking to ensure that disk transfers take advantage of all the available bandwidth. Specifically, controlled transfers use the I/O Channel
Ready (IORDY) line in the AT Attachment interface to signal to the computer host when
the drive needs to make a high-speed transfer. Using the IORDY line, the drive can call
for maximum drive bandwidth support and increase its transfer rate.
The current ATA specifications allow for five modes of PIO transfers, as listed in Table
10.3. Modes 0, 1, and 2 use blind transfers. Modes 3 and 4 use flow control.
AT Attachment
TA B L E 1 0 . 3
AT Attachment Programmed I/O Modes
Transfer Mode
Speed (Mbps)
PIO mode 0
600
1.67
PIO mode 1
383
2.61
PIO mode 2
240
4.17
PIO mode 3
180
11.1
PIO mode 4
120
16.7
The AT Attachment standard allows for Direct Memory Access transfers, which in turn
allow for bypassing the host computers microprocessor and moving data directly to memory. But their real speed comes from the reduced cycle times, bursting blocks of data, and
bus-mastering. Moreover, when the host operating system does not serialize input/output
functions, DMA transfers allow a degree of parallel processing. The host microprocessor
can engage in other activities while the DMA transfer progresses. Versions of Windows
based on DOS (including Windows 95, 98, and Me) serialize their I/O functions, so they
do not gain this benefit. Unix, Windows NT, Windows 2000, and Windows XP can
achieve gains from this transfer strategy.
AT Attachment allows both single- and multi-word DMA transfers. Single-word DMA
transfers move one word at a time. The host sets up the transfer, selects the data to be
transferred, and then makes the transfer. The next word repeats the process. The ATA
specifications acknowledge three single-word DMA modes, as listed in Table 10.4. But the
one-word-at-a-time transfers incur substantial unnecessary overhead, so these modes are
classified as obsolete.
TA B L E 1 0 . 4
AT Attachment Single-Word DMA Transfer Modes and Speeds
Transfer Mode
Speed (Mbps)
Standard
960
1.04
ATA
DMA, single-word mode 1,
480
2.08
ATA
240
4.17
ATA
Multi-word DMA transfers replace single-word modes under more recent standards, delivering improved performance by operating as a burst mode. After the host sets up the
transfer, it selects the starting and ending words for the transfer; then the interface moves
the entire block of data from start to end with no further intervention. Table 10.5 lists
these ATA multi-word DMA modes.
303
304
TA B L E 1 0 . 5
AT Attachment Multi-Word DMA Transfer Modes and Speeds
Transfer Mode
Speed (Mbps)
Standard
480
4.17
ATA
150
13.3
ATA-2
120
16.7
ATA-3
The UltraDMA modes combine two performance-enhancing technologies. The most

straightforward boost comes from quickening the clock or reducing cycle time and taking
advantage of bus-mastering technology. By shifting control of the bus to the drive system,
the UDMA modes can quicken the pace of transfers.
In addition, UltraDMA doubles-up transfers. Normally a DMA transfer is made on one
edge of the clock cycletypically on the leading or rising edge when the cycle beginsso
one transfer occurs during each clock tick. An UltraDMA system can transfer data on
both the rising and the falling edge of each clock cyclethats two transfers for every
clock tick, an effective doubling of throughput. To make this system work reliably, the
ATA designers shifted from asynchronous transfers to synchronous transfers. The transmitting device (the computer when writing data; the disk drive when reading) generates
the clock and synchronizes the data pulses to the clock. Because one device controls both
the clock and the data, it can control them better.
In order for the disk drive to take control of the timing, it must act as a bus master, so this
technology is sometimes termed bus master DMA. In addition, error protection is built in
to the transfer protocol using a cyclical redundancy check (CRC) algorithm. The CRC
applies only to the cable transfer, not the data stored on disk.
Plain UltraDMA takes advantage of the 16.6MHz clocking speed available in ATA PIO
mode 4 but enhances it with double-clocking. Other UltraDMA modes use faster clocks
to achieve higher peak data rates. These UltraDMA modes are often referred to by their
peak transfer rates, so UltraDMA mode 4 is commonly termed ATA/66 or UDMA/66;
mode 5, ATA/100 or UDMA/100; and mode 6, ATA/133 or UDMA/133.
Ordinarily you dont have to worry about the transfer mode used by an ATA drive system.
Your computer should automatically use the fastest mode available without your
interventionor knowledge.
Addressing and Capacity

The AT Attachment interface can limit the storage you connect to your computer because
of limitations in addressing data in various ATA versions. Although a drive may physically
be able to store hundreds of gigabytes, that capacity might not be useful because the
AT Attachment
interface used by the drive does not allow all the bytes to be addressed. In effect, the bytes
on the drive beyond the addressing limit imposed by the interface have unlisted numbers,
and no amount of begging or cajoling the operator will get you through to them.
Addressing limitations have constrained the usable capacity of AT Attachment drives
almost since their inception. As the need for greater addressing reach has appeared,
designers have added new addressing modes to the ATA standard. The latest move finally
pushes the upper limit beyond any currently feasible drive capacity.
CHS Addressing
When engineers first conceived ATA, they based its addressing system on physical attributes of the hard disk the interface was to control. They assigned blocks of storage based
on the heads, tracks, and sectors of the disk drive. This addressing system is sometimes
called CHS addressing for Cylinder (the techie term for track), Head, Sector addressing.
Somewhat arbitrarily, the engineers allowed for addressing disks with 16 separate heads or
disk surfaces, each of which may have up to 65,536 tracks spread across it. Each track
could contain up to 255 sectors of 512 bytes eachall generous figures for the state of
technology at the time. Multiply these numbers, and youll see the that CHS scheme
allowed for a capacity limit of 136,902,082,560 bytes or 127.5GB (in the days when a large
hard disk measured 40MB).
Most people round that capacity number to 128GB, even though its a bit less, because of
the limit of 255 rather than 256 sectors per track. (For historical reasons, drive-makers
start numbering sectors with one rather than zero.) The scheme should have worked until
the year 2002, when the drive-makers introduced the first 130GB drive, the first to exceed
the CHS imposed limit.
Unfortunately, in practical application, the BIOSs of older computers severely constrained
ATA addressing. All software that relies on interrupt 13(hex) code routines in your computers BIOS must abide by the BIOSs own disk addressing system. The BIOS routines
originally developed in 1982 for the IBM personal computer XT allowed for disks with up
to 255 heads or disk surfaces (IBM started number heads with one so the limit is 255
rather than 256), each with up to 1024 tracks on them. Each track can hold up to 63 sectors. This somewhat different geometry allowed for a small total capacity8,422,686,720
bytes or 7.8GBwhen using standard 512-byte sectors. Most sources round this figure up
to 8GB.
A problem arises when the two limits of these two geometries get combined. The largest
value that the system can handle for any of the three drive parameters is the smaller of the
limits set by the two standards. Table 10.6 summarizes the interaction between the BIOS
and ATA limits.
305
306
TA B L E 1 0 . 6
ATA and Interrupt 13(hex) Addressing Limits
Feature
ATA Limit
BIOS Limit
Combined Limit
Heads
16
255
16
Tracks
65,536
1024
1024
Sectors
255
63
63
Total sectors
267,386,880
16,450,560
1,032,192
Capacity
127.5GB
7.8GB
0.5GB
The result is that ATA disks that are addressed through the interrupt 13 facilities of the
BIOS of a computer cannot address more than 63 sectors per track (the BIOS limit), 16
heads (the ATA limit), and 1024 tracks (the BIOS limit again). With 512-byte sectors, the
top capacity is 528,482,304 bytes (about 504MB). Bytes beyond this limit simply cannot
be addressed through the standard interrupt 13(hex) routines of the standard computer
BIOS.
CHS Translation
Because the AT Attachment design as a system-level interface allows the inner workings of
the ATA device to be hidden from your computer, manufacturers are free to alter the
geometry of the drive as long as they make it look like something recognizable to your
computer. The number of physical heads, tracks, and sectors inside the drive are usually
totally different from the number of heads, tracks, and sectors your computer uses to
address the sectors on the disk. The drive translates the incoming addresses to its own,
internal physical addresses when it seeks out a sector.
Engineers broke through the 504MB limit by adding CHS translation to the BIOSs of
computers. The BIOS translates commands from your software and operating system to
its own CHS addressing system with the 8GB capacity limit. CHS translation is more a
band-aid than solution and, thanks to revisions in the ATA specification, is not needed by
modern drives (which are larger than the 8GB limit, anyway).
Logical Block Addressing

The ATA-2 specification introduced an alternate way of organizing disks called logical block
addressing (LBA). It sidesteps the BIOS and substitutes 28-bit logical block addresses for
disk sectors. Each sector on the drive is assigned a unique logical address that would be
used by the BIOS to reach its contents. The full 28-bit addressing scheme allows for
268,435,456 sectorsabout 137.5GB. The choice of a 28-bit scheme was essentially arbitraryit simply matched the CHS limit well, allowing slightly more capacity.
Of course, this scheme was foreign to BIOSs and operating systems. BIOSs written after
the development of LBA technology accommodate it by translating LBA addresses into
AT Attachment
CHS addresses (sound familiar?). Operating systems require drivers to understand the
technology (which are included with all modern versions of Windows). Problems with
some BIOSs limited the capacity of hard disk that could be addressed (sometimes to as
small as 2 or 3GB, but more often to 8GB) because the BIOS translated LBA addresses
into CHS addresses with all their limitations.
48-bit Addressing
As the ATA standard bearers, known formally as the ANSI T13 Technical Committee, were
preparing the fifth version of their specification, they decided to put to rest capacity limits
once and foreverat least for the foreseeable future. With the ATA-6 specification, they
added a new addressing system that accommodated logical block addresses 48 bits long,
equivalent to about a million times more on-disk capacity (a total of
144,115,188,075,855,360 bytes). In more convenient language, thats 144 quadrillion
bytes, or 144 petabytes (PB).
The new addressing scheme uses the same registers as the older 28-bit addressing system.
It doubles the width of the registers used for LBA addresses by layering them.
(Technically speaking, the registers act as a two-byte First In/First Out buffer.) For example, the operating system first writes the most significant bits of the LBA of a sector and
then writes again to load the least significant bits of the LBA.
Just adding more bits is not sufficient to make a new addressing system work, of course.
The addressing revision also includes new commands to access and use the longer
addresses. At the same time, the committee elected to increase the largest possible transfer
per command from 256 sectors to 65,536 sectors, allowing new drives to swallow and disgorge up to 32MB with a single command. The ATA system accommodates such large
transfers by doubling the width of the sector count register by layering it in the same way
as the LBA registers.
An operating system thats aware of the ATA-6 specifications can determine whether a
drive uses 48-bit addressing using the ATA Identify Device command. For compatibility,
all drives with 48-bit addressing also support 28-bit addressing, although only part of their
capacity may be available using it.
Unfortunately, 48-bit addressing doesnt entirely eliminate addressing limits on disk
capacity. It just shifts them back to the operating system. Most operating systems are
designed for handling only 32-bit addresses, so most (including most versions of Linux
and the entire Windows family) cannot handle capacities larger than 2.19 terabytes (trillion bytes, abbreviated TB).
Other Features
Although speed and capacity issues highlight most discussions of disk interfaces, the AT
Attachment standard covers other issues as well. The ATA Packet Interface allows you to
307
308
connect other devices such as CD and DVD drives to an interface that was originally
designed just to handle hard disks. Power management under the ATA rubric makes your
computer more energy efficientand lets you keep computing on your notebook computer even after a mixture of fog and pilots lost in the lounge leaves you waiting at the airport gate long after any of your hopes of making your connection have departed. Built-in
security features protect you from prying eyes and your data from errant erasures. And the
automatic drive-identification capabilities of ATA makes setting up a computer a simpler
chore, one that makes lengthy lists of drive parameters as happily behind you as a visit
from in-laws who have finally finished an overly prolonged visit.
ATA Packet Interface

As it was originally conceived, the AT interface lacked facilities to control everything that
a CD or DVD drive needed to do, such as play audio and video or simply change disks. It
was designed for hard disks that never change and merely absorb and disgorge data.
When it comes to ordinary data, however, CD-ROMs are simpler stillthey only discharge data.
To accommodate other kinds of storage media, the T13 committee added the AT
Attachment Packet Interface (ATAPI) to the ATA specifications. The committee based the
command set on those used by the SCSI system but otherwise kept the scheme completely compatible with existing ATA hardware and drivers. They did not entirely duplicate SCSI, however. Although ATAPI uses many of the same block and command
definitions described by SCSI, it does not use many of the other features of SCSI protocol, such as messaging, bus sharing with multiple computers, disconnect/reconnect, and
linking and queuing of commands.
ATAPI changes nothing on the computer side of the AT connection, and it does not affect
the design or operation of ATA hard disk drives. It just gives the makers of other storage
devices and programmers guidance as to how to link their products to computers in a
standard way.
Normally, an ATA hard disk gets its commands through eight registers called the Task File,
which passes along all the commands and parameters needed to operate the disk.
Unfortunately, these eight registers are not sufficient for the needed CD or DVD control.
ATAPI adds one new command, the Packet Command, that initiates a mode in which
multiple writes to the Task File will send packets of commands to the CD or DVD drive.
Most ATAPI command packets contain 12 bytes, although the standard also defines 16byte packets for compatibility with future devices.
The first of the 12 bytes in the ATAPI Command Packet (byte 0) is an operation code that
defines the command itself. The initial ATAPI specification defined 29 operation codes.
The third through sixth byte (bytes 25) of each packet hold the logical block address of
the data to be used if the command involves the use of data. The logical addresses start
AT Attachment
with zero as the first block and increase sequentially up to the last block. The eight and
ninth bytes (bytes 7 and 8) of the packet define the length of the transfer, parameter list,
or allocation involved in the command. Special extended commands add an extra byte for
indicating this length. The remaining bytes in the packet were not originally defined but
are left reserved for future implementations.
Power Management
In computers where electrical consumption is a critical issue (for example, battery powered notebook machines), the hard disk is one of the major drains of power. Most disk
drives meant for portable machines help conserve power by allowing their electrical use to
be trimmed when the drives fastest response is unnecessary. The T13 committee added
power-management commands to the specification starting with ATA-3. This design adds
four power modes, which allow frugal computers to economize as the occasions arise:
Active mode. This mode means normal operation for a disk drive. All functions of
the drive are available, the drive electronics are fully operational, and the disk platters are spinning. The drive can react immediately to any command, including a
seek to move its head to read or write data. The only delay is that imposed in normal disk access. The price for this fast access is that this mode consumes the most
power, particularly during seeks.
Idle mode. This mode is provided more for safety than power savings. The head
may move away from the active data area of the disk, but otherwise the drive is fully
operational. A seek command sends the head instantly scurrying to the appointed
sector with little additional delay. Because the drive does not transfer data in idle
mode, the part of its electronics associated with decoding, deserializing, and transferring data from (or to) the disk can be shut down. The part of the electronics that
processes commands stays alive, ready to receive instructions from your computer
and carry them out.
Standby mode. This mode saves power by stopping the spin of the platters and
shutting down the spindle motor to eliminate that power drain. The electronics of
the drive are left in essentially the same state as idle mode. Only as much of the circuitry is kept awake as is needed to receive commands and activate the rest of the
drive when requested. Because the drive must spin up to its normal operating speed
before it can read or write data, this mode imposes a delay on all data-handling
operations. Typically this delay will be on the order of ten seconds, although the
ATA standard allows for a delay of up to 30 seconds.
Sleep mode. This mode totally shuts down the hard disk and its electronics. The
drive does not listen for commands and cannot act on commands sent to it. The
only way of leaving sleep mode is by sending a hardware or software reset command.
The reset causes the drive to enter its normal power-up mode as determined by its
manufacturer. Sleep mode reduces the power consumption of a drive nearly to zero
309
310
but incurs the longest delay in accessing data. Not only must the drive spin up its
platter, but the rest of the system must also take its normal course through its reset
action.
Table 10.7 summarizes and distinguishes the four power-management modes of the ATA
specification.
TA B L E 1 0 . 7
AT Attachment Power Management Modes
Mode
Electronics
Motor (Spin)
Response
Power Savings
Active
On
On
Instant
None
Idle
Partially off
On
Instant
Small
Standby
Partially off
Off
Delay
Substantial
Sleep
Off
Off
Delay
Complete
AT Attachment drives with the power-management option also include a built-in standby
timer that can shift the drive into its lower-power standby mode after the drive has been
inactive for a preset period. This changeover is a function of the drive and occurs automatically without the intervention of the host computer. The host computer can, however,
switch off the standby timer so that the drive does not shift to standby automatically. The
host computer can then take full command of the power mode of the drive.
To switch the drive from active mode to a lower-power mode, the host computer sends
the drive a power-management command. Any drive-access command automatically forces
the drive back into active mode. In addition, the AT Attachment standard includes a
Check Power Mode command that allows the host computer at any time to determine in
which mode the drive is currently operating.
Security
Starting with ATA-3, the T13 committee added device-level password security to limit
access to disk storage. Although you set the ATA drive access password through your computer, the drive stores the password on its own media and it uses the same password, even
if you unplug the drive from one computer and attach it to another. Consequently, if you
activate its password security system and someone later steals your drive, this person cannot access the data stored on the drive.
Depending on the secrecy of your data, you can set security to one of two levels: normal
or maximum. The chief ramification of your choice is what happens should you forget
your password. Choose normal security, and the drive-maker can recover your data for
you. When you set your drive for maximum security, even the drive-maker cannot retrieve
your data.
AT Attachment
The AT Attachment standard allows for two kinds of passwords:

User passwords. These passwords you set yourself to limit access to your drive.
The command structure used by the AT Attachment standard allows for passwords
up to 32 bytes long. The standard provides a common means of setting these passwords and gaining access to the drive that operates at the interface level. Normally,
youll deal with a password through your BIOS, operating system, or application
software, which links to the drive security system through the ATA interface.
Master passwords. These passwords are set at the factory. They are not normally
used in the everyday operation of the drive but rather allow for recovery of data by
the drive-maker should you forget your password. If youve set your drive for normal security, the drive-maker can use its master password to access your data or
inactivate the old password so you can change your password to something that you
can remember. When you set security to maximum, the master password only allows
the drive-maker to erase your drive and restore it to its original conditionblank
with no password set. Your files will be gone forever.
Under the ATA security system, your drive operates in one of three modes:
Unlocked mode. This is the normal operating mode of the disk drive. The drive
carries out all commands sent to it. It can read and write to any of the data areas on
the disk. You can also add or change your password in unlocked mode.
A new drive will normally arrive in unlocked mode. It will operate as unlocked until
you set a user password. Thereafter, the drive will always start operating in the
locked mode and remain in locked mode until it receives a valid user or master password. Normally, the drive will accept five tries at sending it a password. It will then
reject subsequent password attempts until it has been reset.
Locked mode. This mode prevents you from accessing or changing any data thats
stored on the disk. In locked mode, the drive automatically aborts all read and write
commands without executing them. The drive carries out normally all commands
that do not access or alter the data stored on the disk (for example, to enter a lowpower mode).
Frozen mode. This mode prevents the security features of the drive from being
altered. The drive carries out all normal read and write operations but will not
change its security level or password. If you value your data, frozen mode will be
your standard operating mode so that no one can change your drive password should
you leave your computer unattended.
Device Identification
Each device includes 512 bytes of storage for identifying information. Device-makers are
free to locate this storage anywhere they want. It can be stored on a nonremovable storage
medium, such as a hard disk, or in EEPROM or Flash memory.
311
312
Regardless of its location, the AT Attachment standard includes a standard command for
reading this identifying block. Using this command, your computer can interrogate the
device to find out what it is and automatically configure itself for optimum operation,
including maximizing its data-transfer rate. Included are the device parameters, the features supported (including transfer modes and thus speeds), the model number, and the
serial number of the drive.
Logical Interface
The AT Attachment interface logically links to the rest of your computer through a series
of registers. The standard does not constrain the port addresses used by these registers.
Their location is set by the BIOS writer and chipset designer of your computer. The registers are generally inside the motherboard chipset, and the BIOS must be written to
match. Add-in host adapters include their own registers and matching add-on BIOS.
Applications and operating systems that use the BIOS never need know the location of
these ports. Software drivers for operating systems that bypass the BIOS must be written
with the proper register addresses.
Most chipset-makers use the same range of addresses for the I/O ports used in communicating with their AT Attachment interfaces. Consequently, a single set of register
addresses has become a de facto standard in nearly all computers. Table 10.8 lists these register addresses.
TA B L E 1 0 . 8
AT Attachment Primary Port Host System Registers
Register
Read Function
Write Function
01F0h
Read data
Write data (16 bits)
01F1h
Error register
Set features data
01F2h
Status of sector count
Write sector count
01F3h
Starting sector
Write sector
01F4h
Cylinder low location
Write cylinder low location
01F5h
Cylinder high location
Write cylinder high location
01F6h
Head/device selection
Write device/head selection
01F7h
Device status
Device command
03F6h
Alternate status
Device control
03F7h
Drive address
(Read only)
A single ATA connection is meant to handle the control of two drives. Most modern host
adapters include a secondary portanother connector to plug in a cable for two more ATA
AT Attachment
devices. The logical location of the ports used by both the primary and secondary AT
Attachment host adapters is not fixed and can be set at any of several ranges of values that
the hardware manufacturer chooses. The chosen I/O port addresses only need to be
reflected in the BIOS routines for interrupt 13(hex) service and the software drivers used
to control the board. Nevertheless, most manufacturers use the same ports and interrupts
for their primary and secondary AT Attachment ports.
Signals and Operation

The hardware side of the AT Attachment interface is based on the Industry Standard
Architecture (ISA) bus, but it is not the complete bus. For example, hard disk drives are
not normally memory-mapped devices, so theres no need to give a hard disk control of
the addressing signals (24 of them) of the ISA bus. Instead, the registers of the interface
connect to the input/output address space of the ISA bus, and only those signals used in
addressing the ATA registers need to be used.
In addition, AT Attachment uses a 16-bit data bus (although it allows for drives that only
use eight) as well as a variety of control signals. These include signals to request reading
or writing data, to make DMA transfers, to check the results of running diagnostics, and
to indicate which of the two drives that can be connected to an AT interface port is to
perform a given function. ATA also allows for a spindle sync signal so that two drives
can spin synchronously, as is required for some implementations of drive arrays.
Connectors
An AT Attachment interface uses any of three connectors for its interface. The most common has 40 pins to accommodate its addressing, data, and control signals. As shown in
Figure 10.1, one pin is sometime removed and the connector shell is notched. The connector has a tab corresponding to the notch. Together the tab/notch and missing pin key
the connector against improper insertion.
FIGURE 10.1
Key notch
The 40-pin ATA connector.
Pin
one
Key pin
The 44-pin connector used by 2.5-inch drives adds the three essential power connections
as well as a coding pin. A 50-pin connector additionally allows four vendor-unique pin
assignments, which are typically used for indicating master or slave status for a given
drive. (Two pin spaces are skipped.) Figure 10.2 shows how the power and vendor-unique
pins are arrayed on a 50-pin connector.
313
314
FIGURE 10.2
The 50-pin ATA connector (of which the
44-pin connector is a
subset).
Power
connections
Key
pin
Drive
designation
Regardless of the connector used by a drive, the function assigned each pin remains the
same. Table 10.9 lists the signal assignments of the interface.
TA B L E 1 0 . 9
AT Attachment Signal Assignments
Pin
Function
Pin
Function
Vendor Unique
Vendor unique
Vendor Unique
Vendor unique
(Key pin)
(Key pin)
RESET-
Ground
Data line 7
Data line 8
Data line 6
Data line 9
Data line 5
Data line 10
Data line 4
10
Data line 11
11
Data line 3
12
Data line 12
13
Data line 2
14
Data line 13
15
Data line 1
16
Data line 14
17
Data line 0
18
Data line 15
19
Ground
20
(Key pin)
21
DMARQ
22
Ground
23
DIOW-
24
Ground
25
DIOR-
26
Ground
27
IORDY
28
PSYNC:CSEL
29
DMACK-
30
Ground
31
INTRQ
32
IOCS16-
33
DA1
34
PDIAG-
35
DAO
36
DA2
37
CS1FX-
38
CS3FX-
AT Attachment
Pin
Function
Pin
Function
39
DASP-
40
Ground
41*
+5V (Logic)
42*
+5V (Motor)
43*
Ground (Return)
44*
Type- (0=ATA)
* Note: Pins 41 through 44 are used in 44-pin connectors only.
ATA drives that are built in to PC Cards or CardBus cards use the 68-pin connection of
the associated PCMCIA standards. These are essentially bus connections and are covered
in Chapter 9, Expansion Buses.
Signals
Seven of the connections of the AT Attachment interface (numbers 2, 19, 22, 24, 26, 30,
and 40) are grounds, scattered among the signals to provide some degree of isolation of
one signal from another. Sixteen pins (3 through 18) are devoted to ATAs 16-bit data bus.
Beside the 16-bit connection system used by all of todays devices, the AT Attachment
specification allows for eight-bit connections through the interface. Such narrow-bus systems use only the odd numbered eight pins in the standard sequence.
The remaining 16 pins of the standard 40-pin connector are assigned various signalcontrol functions, such as those to manage reading or writing data, make DMA transfers,
and coordinate the operation of two drives.
The AT Attachment signals primarily concerned with controlling the transfer of data
across the interface are given their own dedicated connections. Commands to the drives
and the responses from the drives (including error indications) are passed through 17
eight-bit registers.
The two drives permitted under the AT Attachment standard share the connection and
receive all signals across the interface indiscriminately. To signal which drive should act on
a given command, the AT Attachment uses a special control register. The same register
also determines the head, track, and sector that is to be used at any given time.
Seven signals are used to select among the registers. The registers are divided into two
groups: Control Block Registers and Command Registers, indicated by two interface signals. Activating the Drive Chip Select 0 signal (pin 37) selects the Control Block registers.
Activating the Drive Chip Select 1 signal (pin 38) selects the Command Block Registers.
When the Drive I/O Write signal (pin 23) is active, the registers that accept commands
from the host are accessible through the interface data lines. When the Drive I/O Read
signal (pin 25) is active, the registers indicate drive status through the data lines.
315
316
Drive Address Bus 0 through 2 are located on pins 35, 33, and 36 and control which register is currently selected and accessible through the data lines. The AT Attachment standard defines two read and one write control-block registers as well as seven read and
seven write command-block registers. Of these, one write command-block register selects
the active drive and head (up to 16 heads are allowed). Two registers select the drive track
(allowing up to 65,536 tracks on a single drive), another register selects the sector to start
reading from or writing to, and another register selects the number of the track to be read
or written. The read registers indicate which drive and head are active and which track
and head are being scanned. Other registers provide status information and define errors
that occur during drive operation.
The number of register bits available sets the logical limits on device size: up to 16 heads,
65,536 tracks, and 256 sectors per track. Because many computers are incapable of handling more than 1024 tracks, the maximum practical capacity of an AT Attachment drive
is 2,147,483,648 bytes (2GB).
During active data transfers, separate signals are used as strobes to indicate that the data
going to or coming from the drive or values in the control registers are valid and can be
used. The falling edge of the Drive I/O Read signal (pin 25) indicates to the host that
valid data read from the disk is on the bus. The falling edge of the Drive I/O Write signal
(pin 23) indicates that data on the bus to be written on-disk is valid.
The Drive 16-bit I/O signal (pin 32) indicates whether the read or write transfer comprises 8 or 16 bits. The signal is active to indicate 16-bit transfers.
Normally, AT Attachment transfers are accomplished through programmed I/O, the standard mode of operation using the standard AT hard disk BIOS. However, the AT
Attachment standard optionally supports Direct Memory Access (DMA) transfers. Two
signals control handshaking during DMA data moves. The drive signals that it is ready to
read data and transfer it in DMA mode by asserting the DMA Request signal (pin 21).
The computer host acknowledges that it is ready to accept that data with the DMA
Acknowledge signal (pin 29). If the host cannot accept all the data at once, it removes the
DMA Acknowledge signal until it is ready to receive more.
In DMA write operations, the host computer uses DMA Acknowledge to indicate that it
has data available, and the active drive uses DMA Request for handshaking to control the
flow of data. The Drive I/O Read and Write signals indicate in which direction the data
should flow (as a disk read or write).
An AT Attachment disk drive can interrupt the host computer to gain immediate attention
by activating the Drive Interrupt signal (pin 31). On programmed I/O transfers, the drive
generates an interrupt at the beginning of each block of data (typically a sector) to be
transferred. On DMA transfers, the interrupt is used only to indicate that the command
has been completed. (The interrupt used is determined by the hosts circuitry.)
AT Attachment
A drive can signal to the host computer that it is not ready to process a read or write
request using the I/O Channel Ready signal (pin 27). Normally, this signal is activated;
the drive switches it off when it cannot immediately respond to a request to transfer data.
The Drive Reset signal (pin 1) causes the drive to return to its normal power-on state,
ignoring transfers in progress and losing the contents of its registers (returning them to
their default values). Normally, the drive is activated briefly (for at least 25 microseconds)
when the host computer is turned on so that it will initialize itself. Activating this signal
thereafter will cancel the command in progress and reinitialize the drive.
The Passed Diagnostics signal (pin 34) is used by the slave drive to indicate to its host that
it is running its diagnostics. The Passed part of the name does not mean that the diagnostics are completed successfully but that the results are ready to be passed along to the host
system. Actual results (and the command to actually execute diagnostics) are given
through the AT Attachment registers.
The Spindle Sync/Cable Select signal (pin 28) can be used at the drive manufacturers
option to make the drives spin synchronously (as is required by some drive-array technologies) or to set drive identification as master or slave by the cable rather than using a
jumper or switch on the drive. When used as a spindle-synchronizing signal, the master
drive generates a periodic pulse (typically once each revolution of the disk, although the
actual timing is left to the drive manufacturer), and the slave uses this signal to lock its
spin to the master. When this connection is used as a cable-select signal, supplying a
ground on pin 28 causes a drive to function as the master (drive 0); leaving the connection
open causes the connected drive to act as the slave (drive 1).
A single signal, termed Drive Active/Drive 1 Present, located on pin 39, indicates that one
of the drives is active (for example, to illuminate the drive activity indicator on the system
front panel). The same pin is used by the host signal to determine whether one or two AT
Attachment drives are installed when the power is switched on. The drive assigned as the
slave is given a 400-millisecond period during system startup to put a signal on this pin to
indicate its availability; after waiting 450 milliseconds to give the slave drive time to signal,
the master drive puts its signal on the pin to indicate its presence to the host computer. It
switches its signal off and converts the function of the signal to drive activity when the
drive accepts its first command from the host computer or after waiting 31 seconds,
whichever comes first.
Compatibility
As a general rule, all ATA drives support the standard under which they were made as well
as all earlier versions of the ATA standard. That is, you can plug the most recent ATA
drive into the ATA connector of a vintage computer and have reasonable expectations that
it will work. Similarly, you can blow the cobwebs out of drives that conform to earlier
ATA standards, plug one into the newest computer, and have it work. (You may have to
317
318
manually configure your system to accept an old drive by entering parameters into its
setup program.)
Plugging a fast drive into a slow system yields a slow drive, however. The interface is not
magic and is constrained by the oldest technology in the system. Plug the latest ATA/133
drive into a old computer and you may get a drive the works only in PIO modes. Worse,
if your system doesnt recognize logical block addressing, only a fraction of the total
capacity of the drive may be accessible.
In general, you can mix drives with different speed ratings on a single cable. Because ATA
handles each transaction individually, the drives should operate at the highest speed at
which they are individually capable in the given system.
Wiring
Two different cables are used by the AT Attachment interface. The basic cable is a flat ribbon with 40 conductors and three connectors, all the same. The typical cable puts two
connectors close together at one end and a single connector at the other. Sometimes a
computer-maker cheaps out and only puts two connectors on the cable, so you can connect only one drive (the other end plugs into your motherboard or a host adapter).
The chief problem with the cable design is that it works only for speeds up to 33MHz
(that is, ATA/33 or UDMA/33). Higher-speed transfer modes require a more sophisticated 80-conductor cable. The signals in the cable are essentially the same, but every
other conductor in the cable is held at ground potential. That way, there is always a
ground between two active signals on the cable. The ground effectively shields the two
active signals from one another. The ground wires are all linked together in the cable connector. ATA/66 and above interfaces can sense this cable. If they do not find it, they will
restrict the transfer speed to 33MHz or less, regardless of the capabilities of the drives you
install. Figure 10.3 illustrates the difference between the two cable types.
FIGURE 10.3
ATA connections for
speeds greater than
33MHz require 80conductor cables
(left) instead of the
40-conductor cables
(right) that suffice for
33MHz and slower
speeds.
All connectors on an ATA cable are supposed to be keyed (with a missing pin 20) and on
the shell of the connectors. In addition, the ATA cable is marked with a red or blue stripe
on the edge corresponding to pin 1 on the connector.
AT Attachment
Single Drive
All three connectors on an ATA cable are the same, so in theory you should be able to
plug any connector into any drive or the host adapter. Thats not the best practice, however. Convenience dictates that you plug the connector spaced farthest from the rest at
one end of the cable into your host adapter. Use the two more closely spaced connectors
for disk drives.
In a one-drive system, plug the connector at the other end of the cable into the drive and
leave the middle connector unconnected. This properly terminates the cable. Otherwise, a
fraction of the signal travels down the loose stub and reflects off the unterminated end.
The reflected signal travels back and mixes with the signal at the middle drive connector.
Because of the time it takes the signal to travel back and forth through the stub, the
reflected signal is not a mirror image of the signal at the connector. The signal at the connector will have changed during the travel time. As a result, the combination of the new
and reflected signals will be something unpredictable. Figure 10.4 shows how to terminate
a single drive properly.
FIGURE 10.4
Recommended wiring
of a single AT
Attachment drive.
Two Drives
Connecting two drives to an AT Attachment cable eliminates the electrical concerns. One
drive necessarily will be at the end of the cable and terminate it properly. It doesnt matter
which drive connects to which connectormaster or slave, big or small, old or new. The
connectors are electrically identical.
Three or Four Drives

Most AT interface host adapters enable you to connect more than two drives by providing
two distinct interfaces: primary and secondary. Some systems insist you use the primary
connection for the drive that boots your computer. Others are more flexible.
In any case, treat both the primary and secondary as an individual ATA interface with its
own cable. Always use the last connectors on the cable first, then the connector in the
middle of the cable for the second drive. Most dual-interface systems will allow you to
connect a single drive to each interfacetwo drives, two cables, two interfaces.
319
320
Master/Slave Designation
The AT Attachment system distinguishes the two drives attached to a single cable by making one the master and the other the slave. The chief distinction is, as a default, the master
gets assigned a drive letter first (it gets the letter closer to the front of the alphabet), and
the master is usually preferred as the boot drive (although many BIOSs will let you select
which drive to boot your system).
All AT Attachment drives have the potential for being masters or slaves. The function of
each drive is determined by jumper settings on the drive. These are usually called drive
select jumpers or simply indicated with DS near the contact on the drives electronic circuit
board. Different manufacturers use various designation schemes. A jumper on the pins
marked with the DS, for example, indicates a master. Other manufacturers might use
position-sensitive system. A jumper in one position indicates the drive will act as the master, and the jumper in another position indicates it will be a slave. Most drives have a cable
select setting, although few computer-makers implement this feature (by which the cable
does, in fact, indicate whether the drive is master or slave).
Only one master and one slave are permitted in a single AT Attachment connection, so
each of the two drives in a single AT Attachment chain must have its master/slave jumpers
properly set for the system to work. Moreover, most computers will not boot with only a
slave drive. Most drives are shipped jumpered for operation as the master, so you only
have to adjust the second AT Attachment drive you add to your computerbut be sure to
check.
Sanctioning Organization
The ATA standards are published and maintained by the American National Standards
Institute, Inc., 11 West 42nd Street, New York, New York, 10036. The more recent versions of the ATA standards have been prepared by the T13 technical committee of
Accredited Standards Committee NCITS. The committee maintains a Web site that publishes the working documents and drafts of prospective standards at www.t13.org. The
final approved standards are available in final form only from ANSI.
Serial ATA
Over the last decade and a half, engineers have used several tricks to extend the life of the
ATA interface, such as altering timing and developing new wiring systems. And theyve
succeeded beyond anything that could have been expected at the introduction of the interface. Todays ATA is potentially 25 times faster than the original. Hard disk drives have yet
to exploit the full potential of its latest improvements. But disk developers know they
wont be able to stretch ATA out another 25 times. Nor do they want to. The design has
several weaknesses (not all of which are technical).
Serial ATA
Looking to the future, on the 15th of February, 2000, Intel, along with several drivemakers including APT, Dell, IBM, Maxtor, and Seagate Technology, announced a radically new conception of ATA that they called Serial ATA. Through an industry association
called the Serial ATA Working Group, they published a final specification on August 29,
2001. Although the group expects getting the official imprimatur of a standards organization is still a year off, several companies are already manufacturing to the standard.
Seagate demonstrated the first Serial ATA drive in June 2002.
The name indicates the basic concept. Todays ATA is a parallel interface that uses 16 separate data connections. Serial ATA will use a single data channel operating at a much
higher speedat least 20 times the clock rate of current parallel ATA systems.
The initial specification for Serial ATA calls for an initial peak transfer rate of 150MBps, a
little more than a 10-percent improvement over the current top speed for conventional
parallel ATA, 133MBps. But parallel ATA has no future, according to the Serial ATA
Working Group. Its unlikely that the speed of the parallel interface could be doubled.
The Serial ATA specification, on the other hand, provides a growth path to double and
quadruple that initial rate to peak rates of 300MBps and 600MBps. According to the
Serial ATA Working Group, that speed potential should be enough to satisfy speed needs
for the next decade.
To serialize data on the Serial ATA channel, the system uses conventional 8b/10b encoding (that is, it encodes bytes in ten-bit data groups). To reduce interference levels, it also
uses a combination of spread spectrum clock and data scrambling to minimize repetitive
bit-patterns in the data stream. The group-coding overhead (ten bits to the byte) results
in a 1.5GHz signaling rate on the Serial ATA channel. The higher-speed versions of Serial
ATA will push the signaling rate to 3.0 and then 6.0GHz.
The wiring uses four pairs of wires for two balanced signaling pairsone to the device
and one from the devicefor full-duplex communications. The total voltage swing
between the wires in each pair is only 250 millivolts.
Serial ATA uses a point-to-point connections scheme. That is, only one device connects to
each Serial ATA port. Drives will not daisy-chain as they do with todays parallel ATA systems. To accommodate multiple devices, a host computer needs multiple ports. The typical installation will follow current ATA practice and deliver four ports per computer
systemtwo primary and two secondary ports.
For computer-makers, one of the chief allures of the Serial ATA system is the smaller cable
that it used to connect the drive to the computer. Instead of a wide 40- or 80-conductor
ribbon cable, Serial ATA will likely use a four-conductor cable with compact connectors,
similar to those of USB and FireWire. Such a design will eliminate some of the clutter
inside computers. Beside aesthetics, the compact cable will also improve the cooling of the
host computer and will likely reduce interference.
321
322
Serial ATA retains the ATA designation despite the radical wiring change because it
retains full software compatibility with the ATA system. Drives will use the same
commands as parallel ATA. Consequently, Serial ATA will work with todays operating
systems and software.
On the other hand, Serial ATA will be incompatible with todays parallel ATA drives and
controllers at the hardware level. You simply wont be able to plug one directly into the
other, although manufacturers may produce adapters to mate one to the other.
The Serial ATA Working Group maintains a Web site at www.serialata.org.
SCSI Parallel Interface

Pronounced scuzzy by much of the computer industry (and much less often sexy by its
most fervent advocates), SCSI is a system-level interface that provides what is essentially a
complete expansion bus into which to plug peripherals. SCSI isnt simply a connection
that links a device or two to your computer. Rather, it functions like a sub-bus. SCSI
devices can exchange data between themselves without the intervention of the host computers microprocessor. In fact, they can act across the SCSI bus even while other transfers
are shifting across the host computers normal expansion bus.
SCSI is even older than ATA. In fact, its effectively the oldest interface still in use in personal computers. Originally simply a means to add hard disk drives to minicomputers,
SCSI has evolved into a complete interconnection system that can link hard disk drives,
CD-ROM players, scanners, and even arrays of hard disks to any kind of computer.
The basic SCSI hardware interface is a parallel connection on a multiconductor cable
called the SCSI Parallel Interface (SPI). In original form, the SCSI interface comprised an
eight-bit bus with a ninth parity bit for error detection and a maximum clock speed of
5MHz. Since then, four versions of the SPI system have been developed, the most recent
of which (SPI-4) pushes the width of the bus to 16 bits (not counting parity) and the peak
data rate to 320MBps.
More a place to plug in hard disks, SPI has become an expansion bus that can link up to
15 devices of various types to your computer through a single-port connection. All the
devices function independently, under the control of the host system through the host
adapter that provides the SCSI port.
Standards
The original SCSI standard evolved from another interface called SASI, the Shugart
Associates Standard Interface, developed in 1981 by hard-disk pioneer Shugart Associates
working with NCR Corporation. Together they developed SCSI as an eight-bit parallel
connection between host computers and disk drives. Later that year, the X3T9 committee
of the American National Standards Institute used the SASI specification as the foundation for its work on a parallel interface standard. That standard, now known as SCSI-1,
was formally approved in 1986.
Even before the ANSI acceptance of SCSI, manufacturers were wary of the shortcomings
in the original specification. In 1985 device manufacturers approached the group working
on the standard with proposals to increase the mandatory requirements of the SCSI standard. Because the standard was already near completion and the industry was impatient,
ANSI adopted SCSI while at the same time forming a new group to develop a set of commands for SCSI with broader application, eventually published (though not as an official
standard) as the Common Command Set.
Because the Common Command Set was not a mandatory part of SCSI, each device manufacturer made its own interpretation of it. As a result, individual SCSI devices often
required their own driver software, and many combinations of SCSI devices simply
wouldnt work together at all.
SCSI-2
In 1991, a revision of SCSI was introduced to help fix some of the problems in mating
SCSI devices, as well as to increase the speed of SCSI transfers. Referred to as SCSI-2, the
new standard integrated the newly developed Common Command Set with several
optional hardware enhancements.
Through the Common Command Set, SCSI-2 embraces communications as well as other
storage devices. It incorporates specific command sets for read-only and erasable optical
devices, in addition to CD-ROM. Other commands control devices that can change the
medium on which they store data (for example, cartridges), such as optical jukeboxes.
SCSI-2 also includes commands for printers (carried over from the original SCSI), scanners, and communications equipment.
Hardware improvements included the broadening of the eight-bit SCSI data bus to Wide
SCSI, which doubled the bus width to 16 bits. In addition, the top speed of SCSI transfers
was doubled with the addition of Fast SCSI. With an eight-bit bus width, Fast SCSI
pushed transfer rates up to 10MBps; Wide and Fast SCSI could peak at 20MBps. Wide
SCSI also expanded the number of possible devices connected to a single host adapter
channel from 7 to 15.
SCSI-3
Almost immediately after the SCSI-2 standard was approved, the industry began work on
its successor, SCSI-3, to refine the standard further. Although not formally approved at
the time this was written, SCSI-3 has been out in draft for long enough that most manufacturers have incorporated its requirements into their designs. For the purposes of this
discussion, SCSI-3 is considered the current standard.
323
324
Rather than a single standard, SCSI-3 has become an umbrella that covers a multitude of
individual hardware and protocol standards. Notably, it divorces software from hardware
so that the Common Command Set and the various cabling systems go their separate
directions. At that, SCSI itself becomes a command protocol, and the various interconnection standards control how you wire your peripherals together.
At the hardware level, SCSI-3 brought with it the first generation of the SCSI Parallel
Interface, along with three other interconnection systems. Serial Storage Architecture currently delivers two channels of 20MBps transfer rate for full-duplex operation (a device
can transmit and receive simultaneously at that data rate). P1394 gives a 100MBps serial
system that plugs together like a set of stereo components. Fibre Channel Arbitrated Loop
provides a 100MBps system that uses simple plug-in connections for the disk drive and a
bridge to optical fiber with even higher speeds. These hardware standards give SCSI-3
enough headroom to take peripheral connections into the next generationand the next
incarnation of SCSI.
After the introduction of the SCSI-3, the parallel interface was independently revised as
SPI 2 through 5 (currently). Table 10.10 summarizes some of the various SCSI standards,
their official designations, and the popular names of the speed modes included under
them.
TA B L E 1 0 . 1 0
SCSI Standard Documents
Common Name
Official Standard
Technology
SCSI-2
ANSI X3:131:1994
CCS and hardware
SCSI-3
ANSI X3:270:1996
Architecture
SCSI-3
ANSI X3:301:1997
Primary commands
SPI-2
ANSI X3:302:1998
Ultra2 SCSI
SPI-3
ANSI NCITS:336:2000
Ultra3 SCSI, Ultra160 SCSI-m
SPI-4
Draft T10 Project 1365D
Ultra320 SCSI
SPI-5
Draft T10 Project 1528D
Ultra640 SCSI
Clocking
SCSI Parallel Interface allows for both asynchronous and synchronous transfers.
Asynchronous transfers allow data to be sent at irregular intervals using start and stop bits to
mark the beginning and end of data bytes. Synchronous transfers use system timing to
define data bytes. Because synchronous transfers require less overheadfewer bits need to
be transferred and less processing of the data is requiredthey can be faster. The first
implementation of synchronous SCSI is sometimes called Fast SCSI, usually with a designation of the bus clock speed.
Asynchronous SCSI
Asynchronous SCSI operates as a handshaking system. First, a device issues a request (signaling REQ) and waits for an acknowledgment (ACK) to be sure its request was received
before doing anything else. In every step and every byte of the transfer, this two-step
process must take place. Actual transfers require four steps: First, one device sends the
REQ and it is answered by ACK. The first device turns off its REQ and waits for the
other device to switch off its ACK before the first device can go on to the next byte.
Each of these handshakes must travel down the SCSI cable, so the signal must go back
and forth through the cable twice for each transfer. Although electrical signals are fast,
they travel at a finite and measurable speed, about one-third the speed of light in a vacuum. In typical SCSI cables, it takes a signal about 5.25 nanoseconds to move one meter.
This travel time is termed the propagation delay, and it is the primary constraint on the
speed of asynchronous SCSI transfers in long cables. The longer the cable, the longer the
delay.
Although the original SCSI standard puts a speed limit of 1.25MBps on asynchronous
transfers, that rate applies only to the longest permitted cables, 25 meters. Shorter cables
can achieve higher rates using asynchronous transfers.
The overhead involved in transferring each byte across the parallel SCSI bus using asynchronous signaling is about 160 nanoseconds. Adding in the propagation delay sets the
minimum cycle time and maximum transfer rate of the connection.
Synchronous SCSI
Synchronous SCSI requires an acknowledgment for each request sent, but it allows the
acknowledgment to be delayed. Consequently, a device can send packets one after another
without enduring the propagation delays required in interlocked handshaking.
Synchronous SCSI operates at a speed determined by its cycle time without regard to
propagation delay. In other words, the speed of synchronous SCSI is independent of cable
length.
The first clock speed specified for synchronous transfers was 5MHz. Through the years,
the standard has raised this speed, with each improvement doubling the speed of its predecessor. Each improvement also earned a distinctive name. Fast SCSI boosts the clock
speed to 10MHz. Ultra SCSI kicks it up another increment to 20MHz. Ultra2 SCSI doubles the speed again to 40MHz. Ultra160 SCSI doubles it again, to 80MHz. Todays
fastest, Ultra320 SCSI, leaves the bus speed alone but double-clocks the data so that transfers are made on both the rising and falling edge of each clock cycle. Therefore, data
moves at twice the clock speed.
325
326
The most recent SPI standards also permit double-clocking, the transfer of two data bits on
every clock pulseone on the rising edge, one on the falling edge. In fact, double-clocking is necessary to achieve the highest peak rates for all the most recent SPI standards.
Double-clocking was introduced with the SPI-3 specification.
Bus Width
The current SCSI parallel interface standards allow the use of either an 8- or 16-bit data
bus, with the latter moving data twice as fast as the former. The 16-bit version usually is
called Wide SCSI. Under early SCSI standards, a wider, 32-bit bus was also allowed,
although it was never commercially implemented. The SPI-3 standard declared the 32-bit
bus obsolete.
Under the SCSI-3 standard, Wide SCSI is an optional implementation. However, the
wide bus is mandatory for Ultra160 and Ultra320 SCSI connections.
Table 10.11 lists the nomenclature and speed limits (more correctly, peak transfer rates) of
todays common parallel SCSI connection systems.
TA B L E 1 0 . 1 1
Comparison of SCSI Parallel Interface Performance Specifications
Interface Name
Bus Width
Bus Speed
Peak Transfer Rate
Plain SCSI (SCSI-2)
8 bits
5MHz
5MBps
Fast SCSI
8 bits
10MHz
10MBps
Wide SCSI
16 bits
5MHz
10MBps
Fast Wide SCSI
16 bits
10MHz
20MBps
Ultra SCSI
8 bits
20MHz
20MBps
Ultra Wide SCSI
16 bits
20MHz
40MBps
Ultra2 SCSI
8 bits
40MHz
40MBps
Ultra2 Wide SCSI
16 bits
40MHz
80MBps
Ultra160/m SCSI
16 bits
40MHz
160MBps
Ultra320
16 bits
80MHz
320MBps
Capacity and Addressing

SCSI devices store data in the form of blocks. Each block is assigned an address, and a
host indicates which block to read or write by its address. The size of each block and the
number of available addresses constrain the capacity of any given device.
The original SCSI command structure provided 21 bits for identifying block addresses,
sufficient for 2,097,152 blocks. With hard disk drives that used blocks of standard 512byte sectors, this addressing scheme allowed for drives storing up to 1GB.
The SCSI-2 refinement allows for a full three bytes (24 bits) for encoding block addresses
by taking advantage of three bits reserved under the earlier implementation. SCSI host
adapters designed to that standard can access up to 16,777,216 blocks, which with sectorsize blocks limits capacity to 8GB, or 8,589,934,592 bytes. Larger drives could be accommodated by defining blocks as multiple sectors.
Increasing block size is a band-aid solution, one that only covers up an underlying problem. Consequently, the current SCSI Block Commands now include a second form of
read, write, and related commands that allocate four bytes (32 bits) for block addresses.
Using the new commands, the system can address up to 4,294,967,296 blocks, enough to
handle devices storing 2TB (2,199,023,255,552 bytes) of data.
Note that some early SCSI host adapters suffered addressing limits imposed by design
choices made by their manufacturers. In particular, many old host adapters made before
about 1995 may limit addressing to 1GB or less. In other words, large-capacity SCSI
devices may not work properly with older host adapters.
Signaling
SCSI systems can use any of three kinds of signaling: single-ended and two forms of differential signaling. Early generations of SCSI devices used single-ended signals, as do
many modern low-cost and/or low-speed devices. Differential signaling allows for higher
speed and is mandatory under the fastest SPI standards.
Single-ended SCSI uses an unbalanced (or single-ended) electrical signala single wire for
each signal, with all signals in the bus using a single common ground return. Differential
SCSI uses balanced (or differential) signals. Each signal on the SCSI bus has its own return
line that is isolated from the reference ground. Differential SCSI signals use twisted-pair
wiring. Most SCSI implementations have been single-ended because they require half the
pins, cheaper wire, and simpler electronics than do differential SCSI implementations.
As with all single-ended electrical systems, single-ended SCSI is more prone to picking up
noise and interference than differential SCSI. As a result, the specifications for SCSI systems that use single-ended signals limit cable lengths to no more than 6 meters (just under
20 feet). Differential SCSI allows for bus lengths up to 25 meters (about 82 feet). You
must have at least one-third meter (about 12 inches) of cable between SCSI devices, so the
shortest possible SCSI cable is that length. External SCSI cables should be shielded.
The voltages on the single-ended SCSI vary a bit with speed. The levels used are summarized in Table 10.12.
327
328
TA B L E 1 0 . 1 2
Single-Ended SCSI Signaling Levels
Signal Type
Logic Zero
Logic One
Fast-5
2.0 to 5.25 VDC
0 to 0.8 VDC
Fast-10
2.0 to 5.25 VDC
0 to 0.8 VDC
Fast-20
1.9 VDC maximum
1.0 VDC maximum
Instead of centering the logic signals on zero volts, differential SCSI signals are designed
so that the currents flowing during logic transitions exactly cancel out each other in the
two signal conductors. When one of the two differential signal lines transitions from low
to high, the other goes from high to low. Even though both high and low are positive values, the current flows cancel. The current flow generates the magnetic field that limits
performance as well as generates interference.
The first differential signaling system used by the SCSI system was quite tolerant of high
common-mode voltages (for instance, noise). Called High-Voltage Differential (HVD)
SCSI, the difference between low and high signals was required to be one volt or more.
The maximum voltage for a low signal is 1.7 volts; the minimum for a high signal is 2.7
volts. The common mode voltage, which is the same on both the positive and negative
signals lines, can range from 7 to +12 volts.
The SPI-3 standard declares HVD signals obsolete. They should no longer be used in
new SCSI devices.
The SPI-2 specification introduced another differential signaling scheme that ensured the
appearance of only lower voltages on the SCSI lines, making the system compatible with
single-ended signaling. Called Low-Voltage Differential (LVD) SCSI signaling, this scheme
is optional for speeds of 20MHz and under and mandatory for clock speeds of 40MHz
and higher, including Ultra2 SCSI, Ultra160 SCSI, and Ultra320 SCSI.
Wiring
Because SCSI is a bus, its wiring is essentially simple. The pins of the connectors on one
device are directly linked to the pins of the same position (and function). You can run a
straight cable (no twists or crossovers) from the host adapter to one SCSI device, then on
to the next. Unfortunately, reality interferes with this idealized connection system. All the
variations of the parallel interface need to be taken into account. As a result, the mechanics of the actual cabling vary with bus speed and width, whether your SCSI connections
are strictly inside your computer, external to it, or both.
SCSI also saddles you with two other worries: device identification and termination. The
proper operation of a SCSI system requires that you get both of these setup chores right.
SCSI Identification
As a bus, all devices are simultaneously connected and receive all transmissions (commands and data). Commands are routed to individual SCSI devices by identifying them by
their SCSI address. Standard SCSI systems use eight dedicated lines in the SCSI cable for
identifying devices. Each device uses one line, allowing eight unambiguous addresses that
translate into SCSI identification (SCSI ID) numbers. In eight-bit SCSI systems, these are
usually designated in standard Arabic numerals as SCSI ID numbers 0 through 7. Wide
SCSI systems expand the identification possibilities to 0 through 15. In either case, one
number (usually the highest) is reserved for the SCSI host adapter. The other addresses, 0
through 6 or 0 through 14, can be assigned to any device connected anywhere in the SCSI
chain. Each address is assigned a priority to the device using it, with the host adapter having the highest address (7 or 15) and therefore the top priority. The device with a SCSI
ID of 0 always has the lowest priority.
In the latest SCSI systems that operate under the Plug-and-Play aegis, you dont have to
worry about these numbers. The SCAM system (which stands for SCSI Configured
Automagically) does exactly what its name impliesit queries each SCSI device and assigns
nonconflicting ID numbers to each one it finds. For SCAM to work, of course, your host
adapter, operating system, and SCSI devices must comply with the standard. Note, by the
way, the magically part of the name was applied in the initial standard, although some less
inspired engineers are apt to interpret the name to be SCSI Configured Automatically.
If you have an older host adapter and a mess of old SCSI devices, youll be left to configure them without prestidigitation. And thats where the complication of SCSI begins.
Except for the arbitrary address of 7 assigned to the host adapter (or 15 in Wide SCSI
systems), you assign the addresses that each SCSI device uses. The most basic rule is simple: Each SCSI ID number must be unique, so never assign the same ID number to two
or more devices in the same SCSI chain.
In the formal SCSI scheme of things, you can assign any number to any device, as long as
you configure your software properly. The SCSI specification does not reserve any particular number for any device. As a practical matter, you cannot be so cavalier about the ID
numbers that you use for different devices. One concern is matching your software. A
SCSI driver may require that you indicate the SCSI address that you assign to your CDROM player. The software for the CD-ROM player may demand that you assign a particular address to the player in order for it to work.
If there is any one guide to follow, it is the model set by Adaptec, the largest maker of
SCSI host adapters. To provide some consistency for its host adapters that emulate
Western Digital WD1002 controllers so they can boot your computer, Adaptec chose
rather pragmatically that you must assign your boot hard disk drive (if it is a SCSI drive)
ID number 0. Newer SCSI systems allow you to choose the ID number for the boot
device.
329
330
The SCSI ID controls the arbitration priority of the device to which it is assigned. In the
SCSI scheme, device ID 0, oddly enough, receives the lowest priority in arbitration.
Device 7, usually the host adapter, receives the highest priority. In Wide SCSI systems
with a 16-bit bus, the priority system is a bit more complex, but device 7 maintains the
greatest priority. Table 10.13 summarizes the relationship between device ID and arbitration priority on the Wide SCSI bus.
TA B L E 1 0 . 1 3
Wide SCSI Arbitration Priorities by Device ID
SCSI ID
Priority
16
15
10
14
11
13
12
12
13
11
14
10
15
Early host adapters had built-in BIOSs that enabled the boot drive and one additional
drive to link to your computer without an additional software driver. In order for your
second drive to be recognized when using this kind of controller, the second hard disk
must have ID number 1. Many of the latest SCSI host adapters include an advanced BIOS
capable of recognizing up to seven hard disk drives, giving you the freedom to assign
them any ID, providing you give the boot device ID 0. The BIOS recognizes disk drives (and
dispenses drive letters) according to the order of the ID you assign. The lower numbers
get the first drive letters.
Some host adapters put one more restriction on your freedom in assigning device IDs. If
you cannot get a device to be properly recognized in a SCSI system based on an older
host adapter, you may need to check the host adapter manual to see if it enforces any
restrictions on SCSI ID assignment.
Internal SCSI devices such as hard disks typically have several jumpers or switches on
each device that set its SCSI ID. The location of these selectors varies, but theyre usually
easy to find. Look for a set of jumpers with some form of the legend ID silk-screened
on the circuit board. Older drives often offer a row of seven jumper pairs numbered 0
through 6, with one jumper spanning a single pair. With these, setting the ID is easy.
Move the jumper to the ID number you want. Other drive manufacturers use a binary
code to set the ID, so youll face a set of three header pin pairs that require from zero to
three jumpers, depending on the ID you want.
Newer drives typically use a binary code for assigning the ID number. Four pairs of pins
allow the choice of any ID from 0 to 15. Figure 10.5 gives an example of this binary coding scheme.
FIGURE 10.5
SCSI ID binary code
shown as jumper positions.
Drives vary in the location and labeling of these identification pins, so you should check
with the instructions supplied with a given drive or the drive-makers Web site for guidance in setting the ID number you want.
External SCSI devices commonly use two forms of selector switches for choosing the
SCSI ID number assigned to the device: pushbutton and rotary switches. Figure 10.6
illustrates both of these selectors.
FIGURE 10.6
The pushbutton (left)
and rotary selector
switches for setting
the SCSI ID.
Pushbutton selectors allow you to ratchet up and down the series of identification numbers by pressing two push-tabs on the selector. Some switches cycle around and around
331
332
without stopping; others limit their movement from 0 to 7. You only need to push the
appropriate button until you see the SCSI ID you want in the small adjacent window.
Rotary switches select the SCSI ID numbers like an old-fashioned television channel
selector, the miniature equivalent of the rotating knob. Most of these rotary switches
require a small screwdriver to adjust, preventing them from getting inadvertently changed
by bumping the back of the SCSI device.
Note that nothing in the design of the SCSI Parallel Interface precludes two computers
sharing one chain of SCSI devices except for SCSI identification numbers. Because each
device in a single chain must have a unique ID, two host adapters assigned SCSI ID 7
cannot coexist. If you can alter the ID number assigned to the SCSI host adapter in either
computer, however, you can connect both in a single SCSI chain. Either computer will be
able to access the various devices. In theory, the two computers could transfer files
between one another as well, but most SCSI software makes no provision for such installations.
Cable Connections
Once youve manually configured your SCSI devices with the proper ID numbers (or left
matters to the Fates under SCAM), you have to connect them all together. The SCSI system uses a straightforward daisy-chain wiring systemeach device connects to the next
one down the line.
The exact mechanics of the wiring system depends on whether you have internal or external SCSI devices or a mixture of both.
Internal Devices
An internal SCSI system is like a flexible expansion bus. Instead of connectors to a printed
circuit board, the bus takes the form of a ribbon cable, and the expansion connectors are
plugs on the cable. One end of the cable plugs into your SCSI host adapter. Each drive
attaches to one of the plugs in the cable.
All plugs on the cable have identical signals, so you can use any convenient connector for
any SCSI device. The devices and host adapter use SCSI ID numbers to sort out which
commands and data go where.
For best operation, always plug a device into the last connector at the end of the cable.
When this device is terminated (explained later in this chapter), it prevents signal reflections from the end of the wire that might interfere with the operation of the SCSI system.
Modern SCSI devices may use 50-, 68-, or 80-pin connectors. The last two (and one form
of the first) are molded in a vague D shape that acts as a key, so you cannot plug in a
drive improperly. Older 50-pin connectors sometimes are not keyed and can pose
orientation problems.
As with nearly all ribbon cable systems, a red (or sometimes blue) stripe on one edge of
the cable used by SCSI devices marks the polarity of the cable. The stripe corresponds to
pin 1 of the SCSI connector. Most internal 50-pin SCSI connectors and receptacles are
also keyed with a tab and corresponding notch in the middle of one long edge of the connector and receptacle. When you look into the holes of the cable-mounted female SCSI
connector with the keying tab on top, pin 1 (and the red stripe) appears on the left, as
shown in Figure 10.7.
FIGURE 10.7
Key tab
The keying of an
internal SCSI cable
connector.
The SCSI cable in systems with 50-pin and 68-pin connectors links both control and data
signals. However, it does not provide power to the individual internal SCSI devices. Each
SCSI device will also require you to plug in its own power connector. The 80-pin SCA
system includes power in its connections.
Drives that use the 80-pin system are designed for hot-swapping. You can plug them in at
any time, even when your computer is running. The other connection systems require
that you switch off your computer before you plug in a drive.
External Devices
External SCSI cabling is somewhat different from that of internal devices. Using the
external SCSI cabling system, you run a cable from the host adapter to the first device in
the external SCSI chain. For the next device, you plug another cable into the first device
and then plug the other end of the cable into a second device. Just continue in the same
manner, adding another cable for each additional device.
Most external SCSI devices have two SCSI connectors to facilitate their daisy-chain connection. It doesnt matter which of the two connectors on the SCSI device you use to
attach each cable. Functionally, both connectors are the same; each is equally adept at
handling incoming and outgoing signals.
The empty jack on the last device gets a terminator (unless the device has other provisions for termination). The complete wiring system is a daisy-chain as shown in
Figure 10.8.
Wider bus SCSI systems may require two cables to link each device in the chain. The
cables run in parallel following the same daisy-chain path as a single cable SCSI system.
333
334
FIGURE 10.8
PC host adapter
Terminator
External wiring of the

SCSI Parallel Interface.
As long as the SCSI system is properly terminated, cable length is not supposed to matter,
providing that each of the cables is between 18 inches (one-half meter) and six feet (two
meters) long. High-frequency electrical signals such as those used by SCSI sometimes act
differently from what the rules predict, so the length and placement of the cables may
affect SCSI operation. Rearranging the cables in a recalcitrant system can sometimes
bring the entire SCSI chain to life.
After you finish connecting your SCSI cables, be sure to snap in place the retaining clips
or wires on each SCSI connector to be sure that each connector is held securely in place.
This mechanical locking is particularly important with SCSI connections because the
wiring works like old-fashioned Christmas lightsif one goes out, they all go out. Not
only will whatever SCSI device that has a loose connector be out of touch, all other
devices after the loose connector in the chain also will lose communication. Moreover,
because the chain will no longer be terminated properly, even the devices earlier in the
chain may not work reliably. Locking the wire on each SCSI connector will help ensure
that none of the connectors accidentally gets loosened.
Mixed Devices
When your SCSI system has a mixture of internal and external devices, the wiring is
exactly as you would expecta mixture of both of the preceding schemes. Most SCSI host
adapters have connectors that allow you to link both internal and external devices at the
same time. The individual rules for internal and external wiring apply to the respective
parts of the system. The only change is in terminating the system, as discussed in the next
section.
Terminations
Eventually all SCSI daisy-chains must come to an end. You will have one last device to
which you have no more peripherals to connect. To prevent spurious signals bouncing
back and forth across the SCSI cable chain, the SCSI standard requires that you properly
terminate the entire SCSI system. The SCAM system can automatically activate the
proper terminations in a SCSI system. For manually settings, the SCSI-2 standard allows
for two methods of terminating SCSI buses.
Alternative 1 is the old-fashioned (original SCSI) method of passive terminations using
only resistors. Alternative 2 uses active terminations, which means the terminator uses a
voltage regulator to ensure the voltage on the SCSI line remains constant.
In classic SCSI implementations, the three most popular physical means of providing a
SCSI termination are internally with resistor packs, externally with dummy termination
plugs, and using switches. SCSI-2 systems with active terminations usually use switches.
Resistor packs are components attached directly to circuit boards. Unlike the other interfaces, SCSI devices typically use three resistor packs (instead of one) for their terminations. Most computer-based SCSI host adapters and hard disks come with termination
resistors already installed on them.
You can easily identify terminating resistors as three identical components about an inch
long, one-quarter to three-eighths of an inch high, and hardly an eighth inch thick. Most
commonly, these resistor packs are red, brownish yellow, or black and shiny, and they are
located adjacent to the SCSI connector on the SCSI device or host adapter. When necessary, you remove these terminations simply by pulling them out of their sockets on the
circuit board. Figure 10.9 shows two of the most common styles of terminating resistor
packsthose in DIP and SIP cases.
FIGURE 10.9
Passive terminators
(SIP shown on left,
DIP on right).
External SCSI terminators are plugs that look like short extensions to the SCSI jacks on
the back of SCSI devices. One end of the terminator plugs into one of the jacks on your
SCSI device, and the other end of the dummy plug yields another jack that can be
attached to another SCSI cable. Some external terminators, however, lack the second jack
on the back. Generally, the absence of a second connector is no problem because the
dummy plug should be attached only to the last device in the SCSI chain.
Switches, the third variety of termination, may be found on both external and internal
drives. Sometimes a single switch handles the entire termination, but occasionally a SCSI
drive will have three banks of DIP switches that all must be flipped to the same position
to select whether the termination is active. These switches are sometimes found on the
SCSI device or on the case of an external unit.
A few external SCSI devices rely on the terminators on the drive inside their cases for
their terminations. For these, you must take apart the device to adjust the terminators.
According to the SCSI specification, the first and last device in a SCSI chain must be terminated. The first device is almost always the SCSI host adapter in your computer. If you
install a single internal hard disk to the host adapter, it is the other end of the chain and
requires termination. Similarly, a single external hard disk also requires termination.
335
336
With multiple devices connected to a single host adapter, the termination issue becomes
complex. Generally, the host adapter will be one end of the SCSI chain, except when you
have both internal and external devices connected to it. Then, and only then, should you
remove the terminations from your host adapter. In that case, the device nearest the end
of the internal SCSI cable should be terminated, as should the external device at the end
of the daisy-chain of cables (the only external device that likely has a connector without a
cable plugged into it). Remove or switch off the terminators on all other devices.
Maximum Cable Length

SCSI host adapter maker Adaptec provides these rules of thumb as recommendations for
the maximum cable length under various SCSI standards, speeds, and number of connected devices. Table 10.14 lists these recommendations.
TA B L E 1 0 . 1 4
Maximum Recommended SCSI Cable Lengths
SCSI Standard
Bus Clock
Devices
Maximum Cable Length

Meters
Feet
SCSI
5MHz
<7
16
Fast SCSI
10MHz
<7
Fast/Wide SCSI
10MHz
<15
Ultra SCSI
20MHz
>4
1.5
4.5
Ultra SCSI
20MHz
<4
Ultra2 SCSI
40MHz
>2
12
36
The general rule underlying these recommendations is to divide 30 by the bus clock to
derive the maximum recommended cable length in meters. The rule does not apply to
Ultra2 SCSI because of the special low-voltage differential signals used under that standard.
Cables
In the current SCSI scheme of things, only two different cable types are used. These are
designated the A cable and the P cable.
The A cable has 50 conductors and serves for all eight-bit SPI systems, both single-ended
and differential. Similarly, the 68-conductor P cable serves wide signaling systems.
The purpose assigned each conductor in the cable varies with the type of signaling used
and the connector system. The allowed variety is wide. In fact, the cable is a minor concern compared to connector considerations.
Connectors
More than any other standard computer wiring system, SCSI suffers from a plague of
connectors. Not only do internal and external devices use different connector styles, but
both also come in a wide variety. Beside separate internal and external connector styles,
the system also suffers from generational differences.
The SPI-3 specifications acknowledge eight connector systemsfour nonshielded alternatives for internal devices and four shielded systems for external devices. In the past, computer-makers have used other designs to suit their particular requirements. Table 10.15
summarizes the SPI connector options.
TA B L E 1 0 . 1 5
SCSI Versions and Cabling Requirements
Implementation
Bus
Width
Cable
Configuration
Pin
Count
Maximum
Transfer
Devices
Rate
Supported
Notes
SCSI-1
50
Asynchronous
SCSI-2
50
10
Fast
SCSI-2
16
A+B
50+68
20
Fast + Wide
SCSI-2
32
A+B
50+68
40
Fast + Wide
SCSI-3
50
10
Fast
SCSI-3
16
68
20
16
Fast + Wide
SCSI-3
32
P+Q
68+68
40
32
Fast + Wide
The oldest SCSI devices use 50-pin connectors. More recent devices use 68-pin connectors that support wide as well as narrow SCSI implementations. Many hard disk drives use
80-pin connectors (also called SCA for Single Connector Attachment) that integrate power as
well as the SCSI signal onto a single connector, allowing for the easier hot-swapping of
devicesin particular, the hard disks used in massive disk arrays (see Chapter 17,
Magnetic Storage).
For the most part, a SCSI connector is like a shoe. If it fits, use it. All too often, however,
your SCSI device comes equipped with a connector that looks weirdly different from the
one on the cable youre trying to snap onto it. You can prevent problems by checking the
connector type used by the SCSI cables in your computer and matching any device you
buy to your cabling. SCSI deviceshard disk drives in particularoften have several connector options. When buying a new drive, choose the one that fits. Adapters are also
available for matching unlike connectors. Within the limitations noted in the next section,
an adapter is often sufficient to get any SCSI device to link up with any host adapter.
337
338
50-pin Internal Connectors

The initial incarnation of the SCSI interface was the basic 50-contact pin connector used
on the first SCSI drives. Officially, this system is termed the nonshielded connector alternative 2.
Basically a glorified header, this connector featured two parallel rows of pins in a rectangular array, the centers of the pins separated at increments of 0.1 inch, both horizontally
and vertically. Figure 10.7 shows the male connector of this type as you would find it on
the rear of a SCSI device.
At one time, nearly all SCSI devices that used the eight-bit implementation of the standard used this connector. However, both SPI-2 and SPI-3 offer another 50-pin nonshielded design thats more compact and more consistent with current circuit design
practices.
Termed the nonshielded connector alternative 1, this design also uses two parallel rows of
pins. As with the older connector, the rows are spaced on 0.1-inch centers, but the individual pins in each row are spaced at only half that distance, 0.05 inch. The female receptacle is molded with a vague D shape. The male connector has a matching extended
D-shaped perimeter that protects the pins. Drives use the female connector as a jack.
Figure 10.10 shows this connector.
FIGURE 10.10
The 50-pin nonshielded connector
alternative 1.
Although both nonshielded connector alternatives 1 and 2 have 50 pins, the functions
assigned each pin differ between the two alternatives. In fact, there are four variations:
single-ended alternative 1, single-ended alternative 2, differential alternative 1, and differential alternative 2. You cannot tell the four options apart by merely looking at the connector. You must check the specifications of the drive to determine the signal type it uses.
50-pin External Connectors

The most common form of external SCSI connectors has 50 pins arranged in two rows of
25 and looks like an enlarged Centronics printer connector. This connector is termed the
shielded connector alternative 2 under SPI-2 and SPI-3. It is still the most common 50-pin
connector used by external SCSI devices and is shown in Figure 10.11.
The shielded connector alternative 1 also offers 50-pins to external devices. It uses the
same layout and signal system as the alternative 1 nonshielded connector, except the mating portions of the connector are metal shields. Figure 10.12 shows this connector.
FIGURE 10.11
The 50-place ribbon
connector, shielded
connector
alternative 2.
FIGURE 10.12
The 50-place external
shielded connector
alternative 1.
As with the nonshielded 50-pin connectors, shielded connector alternatives 1 and 2 differ
in the functions assigned each pin. The pin-outs of the connector alternatives are the
same as for the nonshielded versions. Because of the physical similarities and identical pinouts, you can plug a raw drive into an external shielded alternative 1 connector (providing
you dont mix the single-ended and differential signaling systems). The nonshielded and
shielded versions of alternative 2 also share the same signal assignments with their pins,
but the connectors are physically incompatible.
Some older external SCSI devices and host adapters used classic 50-pin D-shell connectors with the same signal assignments as the A connector. They never attracted a following because they were simply ungainly, about four inches long. These resemble the
alternative 1 connectors but the pins are round rather than square, more widely spaced,
and staggered. Figure 10.13 shows this large connector.
FIGURE 10.13
The nonstandardized
50-pin D-shell
connector.
More popular was a design that trimmed the length of the D-shell connector in half by
similarly trimming the pin count. These shorter, 25-pin D-shell connectors saved space
and were popularized by Apple Computer for its Macintosh equipment. The connectors
are identical to those used by most computer-makers for their parallel port connections.
Despite the similarity, the signals on the SCSI and parallel ports are entirely different and
should never be connected together.
To reduce the number of pins required, this style of connector eliminates many of the
ground return signals used in single-ended SCSI. These connectors do not have enough
connections for differential signals. Figure 10.14 shows this connector.
339
340
FIGURE 10.14
The female 25-pin
D-shell connector
used by some SCSI
products.
68-pin Connectors
Wide SCSI requires more connections than are possible with a 50-pin connector. To
accommodate all the required signals, the SPI standards specify a set of 68-pin connectors.
This connector system offers both nonshielded and shielded alternatives for internal and
external devices.
Both the nonshielded and shielded versions use essentially the same design and the same
signal assignments. The chief difference is the roughly D-shaped metal shield surrounding
the contacts of both male and female connectors, with the shield also protecting the bare
pins of the male connector, much as in a classic D-shell connector.
Both connectors are termed alternative 3. That is, the internal set make up nonshielded connector alternative 3, and the external set nonshielded connector alternative 3. Many sources
refer to either the shielded or nonshielded connector simply as a SCSI-3 connector. Do not
be misled by this label, which is best seen as an abbreviation for SCSI alternative 3 connector. All SCSI-3 systems do not use this connector system. In fact, the SCSI-3 standard
allows any of the eight connector alternativesas well as the connectors for the various
other physical interconnection systems besides SPI.
The pins of the alternative 3 connectors are arranged in two rows of 34 male contacts.
The pins within each row are located on 0.05-inch centers, and the two rows are spaced
on 0.1-inch centers. Figure 10.15 shows the shielded version of this connector.
FIGURE 10.15
The 68-place shielded
connector
alternative 3.
The SPI specifications allow for a second style of shielded 68-pin connector that looks
like a miniaturized version of the classic 50-place external SCSI ribbon connector.
Termed the shielded connector alternative 4, it uses the same pin-out as alternative 3. Instead
of pins, it has two rows of ribbon contacts spaced at 0.0315-inch intervals. Figure 10.16
shows this connector.
80-pin SCA
Hot-swapping drives requires that both control signals and drive power be connected
quickly and easily. To accommodate the needed power signals, drive-makers developed
the 80-pin SCSI connector system. Under the SPI rubric, this is the nonshielded connector
alternative 4. Figure 10.17 shows this connector.
FIGURE 10.16
Shielded connector
alternative 4.
FIGURE 10.17
The 80-pin SCA connector.
The SCA connector is designed chiefly for internal usefor sliding drives into racks.
There is no external equivalent to the SCA connector.
All the pins on the connector on the SCA device are the same length. On the host end,
however, several of the pins are extended so that they make contact first when the drive
gets plugged in. These longer pins (numbers 1, 3643, 45, 46, and 7880) ground the
drive, supply power to it, and set its SCSI ID number. The added length ensures that the
drive powers up properly and is properly identified before it presents signals to the bus.
Mixing SCSI Standards

SCSI comes in so many flavors that youre likely to have devices that follow two or more
standards that youll want to hook together in a single chain. With modern host adapters
that allow you to configure the transfer characteristics assigned to each SCSI ID, you can
freely mix standards with a reasonable chance that you can get everything to work
together. Better still, you can tune all devices to operate at their maximum transfer rate,
with only a few exceptions.
341
342
In most cases, however, you cant just string together a mixed collection of SCSI peripherals with complete impunity. You still have to consider issues of bus width, signal type (single-ended or differential), and clock speed to make optimal connections. Some things just
wont fitand others you may not want to fit.
Bus Width
In theory, you can mix wide (16-bit) and narrow (8-bit) SCSI devices in a single chain.
However, you must follow some rules.
With an internal chain of mixed devices, the wide ribbon cable carries the full compliment
of signals to all the devices in the chain. The last device in the chain (the one that provides the termination) should be wide. You may need an adapter for each narrow SCSI
device you want to plug into the wide chain.
To mix wide and narrow SCSI devices together in an external SCSI daisy-chain, you need
to follow one rule: Wide goes first. The most important rule is that you must chain all of
the wide devices together, plugged into the host adapter, then chain the narrow devices
after the wide chain. A narrow device wont pass through the signals of a wide device, so
any wide device after a narrow one in the chain will lose its wide-bus option.
Signaling System
The original SCSI specifications allowed for two types of SCSI buses: single-ended and
differential. These quite different signaling systems are inherently incompatible. Although
adapters help mate them together, dangers remain. For example, connecting a singleended device such as a CD-ROM drive to a differential host adapter or a differential drive
to a single-ended host adapter may physically damage both the host adapter and the drive.
Because the SCSI connector system does not distinguish between signal types, you must
be sure of the signal types used by all the devices in your SCSI system before making the
first connection.
The single-ended and high-voltage differential signaling systems are incompatible and
cannot be used on the same bus. In addition, high-voltage and low-voltage differential signaling types are also incompatible. Connecting a single HVD device to an LVD system
could damage all the LVD devices in the system. In theory, single-ended and LVD devices
can be chained together, but you need to be sure the full array of differential signals get to
the LVD deviceswhich means not only differential signals but also the wide bus signals
for Ultra160 and Ultra320 SCSI, which demand a wide bus for proper operation.
Clock Speed
Although the different top speeds of SCSI devices might appear to be a problem and limit
all the devices, no matter how fast, to the speed constraints of the slowest, the issue never
arises. The SCSI host adapter only communicates with a single device at a time, and those
communications monopolize the SCSI bus while they are taking place. In todays SCSI
systems, transfers do not go directly between devices but instead must travel through the
host adapter. For example, to back up a hard disk, the disk first sends a block of data to
the host adapter, which temporarily stores the data in system memory. When the disk finishes sending its block, the host adapter moves the data from memory to the tape drive.
Consequently, the two SCSI deviceshard disk and tape drivecommunicate with the
host adapter independently of one another and may operate at different speeds.
A modern host adapter can adjust its speed to suit whatever devices with which it communicates. In general, it will set itself to transfer data to each device at the top speed at which
that device can operate.
The one big exception to the any speed rule is Ultra2 SCSI. Put simply, you cannot
attach an ordinary SCSI device to an Ultra2 SCSI port and expect Ultra2 SCSI peripherals to operate at top speed. The issue is not really speed, however. All Ultra2 SCSI devices
use LVD signaling. Most ordinary SCSI devices use single-ended (SE) signaling. When
you plug an SE device into the Ultra2 bus, the bus reverts to SE operation; otherwise,
damage would result. The SCSI host adapter would send the negative-going differential
voltages into a direct short-circuit because the associated signal pins are grounded in the
single-ended device. Although reverting to single-ended operation minimizes the chance
for damage, it also limits the speed potential to ordinary Ultra SCSI because the highest
speeds of Ultra2 SCSI require the added integrity of differential signaling.
Operation and Arbitration

All devices connected to a single SCSI bus function independently, under the control of
the host system through the SCSI adapter. Rather than just using signals on dedicated
conductors on the bus that can be understood by devices as dumb as a light bulb, SCSI
presupposes a high degree of intelligence in the devices it connects and provides its own
command setessentially, its own computer languagefor controlling these devices.
Boot Up
Most software drivers search for their target devices when they are booted into your system. Consequently, all your external SCSI devices should be running when you switch on
your computer. Turn on your SCSI devices before you switch on your computer, or use a
power director (outlet box) that ensures that your entire computer systemcomputer and
SCSI peripheralsswitch on simultaneously.
Arbitration
Not only is SCSI more like an expansion bus of a computer than a traditional hard disk
interface, but it also resembles todays more advanced Micro Channel and NuBus designs.
Like the latest computer buses, SCSI provides an arbitration scheme. Arbitration enables
the devices connected to the bus to determine which of them can send data across the bus
343
344
at a given time. Instead of being controlled by the host computer and suffering delays
while its microprocessor does other things, the arbitration of the SCSI bus is distributed
among all the devices on the bus.
Arbitration on the SCSI bus is handled by hardware. Each of the up to seven SCSI devices
is assigned a unique identifying number, usually by setting jumpers or DIP switches on
the drive in a manner similar to the Drive Select jumpers on an ST506 device.
When a device, called the initiator, wants to access the SCSI bus, it waits until the bus is
free and then identifies itself by sending a signal down one of the SCSI data lines. At the
same time, it transmits a signal down another SCSI data line corresponding to the other
SCSI device, called the target, that it wants to interact with. The eight data lines in the
SCSI connection allow for the unique identification of seven SCSI devices and one host.
Note that SCSI devices can initiate arbitration on their own, independent of the host.
Two SCSI devices also can transfer information between one another without host intervention. A SCSI hard disk, for example, may back itself up to a SCSI tape drive without
requiring the attention of (and thus robbing performance from) its host computer. Better
than background operation, this form of backup represents true parallel processing in the
computer system.
In addition, SCSI provides for reselecting. That is, a device that temporarily does not
need bus access can release the bus, carry out another operation, and then resume control.
You can command a disk drive to format, and it can carry out that operation without tying
up the bus, for example. The net result is, again, true parallel processing.
Because SCSI is a high-level interface, it also isolates the computer from the inner workings of the peripherals connected to it. The SCSI standard allows hard disks to monitor
their own bad tracks independently from the computer host. The disk drive reassigns bad
tracks and reports back to its computer host as if it were a perfect disk. In addition, hard
disk drives can be designed to automatically detect sectors that are going bad and reassign
the data they contain elsewhere, all without the host computer or the user ever being
aware of any problems.
The American National Standards Institute (ANSI) maintains the official SCSI standards.
SCSI standards in their final, published form are available only from ANSI.
The SCSI T10 Committee actively develops the various SCSI standards. It provides recommendations to ANSI, which, on its approval, sanctions the specifications as industry
standards. The committee acts as a forum for introducing and discussing proposals for the
standards. The committee maintains a Web address at www.symbios.com/t10.
Fibre Channel
The SCSI Trade Association promotes the use of SCSI and devices that adhere to that
standard. The association maintains a Web site at www.scsita.org.
Fibre Channel
At one time, some engineers regarded Fibre Channel as the successor to the SCSI Parallel
Interface for high-performance disk drives. Although it has been used successfully in many
drive products, it has yet to earn the following of the other interfaces. The steady advance
in the speeds of other interfaces has severely eroded its edge in performance. Both AT
Attachment and the latest versions of parallel SCSI outrace Fibre Channels 100MBps
capabilities. Among personal computers, it remains a viable, but not popular, interface.
The current implementation of Fibre Channel used for mass storage systems is actually a
specialized derivative subset of the technology called Fibre Channel Arbitrated Loop (FCAL). Although even this slice of the standard allows a variety of cabling options, including
twinaxial, coaxial, and optical connections, its chief use in mass storage currently is as a
backplane interface. That is, FC-AL drives link directly to printed circuit boards in their
computer hosts without any cabling between.
At that level, you dont have to understand FC-AL to use it. If you are physically capable
of putting a drive in a bay, youve mastered all you need to know. In modern computers,
the drives and interface are self-configuring and should require no intellectual intervention on your part.
Development of FC-AL began in 1992 with the creation of the Fibre Channel Systems
Initiative, a group of three leading workstation manufacturers (Hewlett-Packard, IBM,
and Sun Microsystems) with the specific goal of overcoming the inherent problems with
the SCSI Parallel Interface.
As with all modern serial systems, Fibre Channel is packet based. The same circuit and
protocol that carry the data also handle all control information. Commands to retrieve a
particular block from a disk drive, for example, are carried through the system as data in
the form of standard SCSI-3 commands. The drive collects the packets sent for it, interprets them, and carries out the commands. It sends data back, when requested to, in the
same data packet form.
Although with hard disk drives youll probably only have to deal with the backplane connector, the Fibre Channel Arbitrated Loop specifications allow for a great variety of
wiring systems. All FC-AL systems take the form of a loop, even the backplane connection. Each device in the loop has separate input and output connections. The output of
one device connects to the next device in the loop, until the last device, which connects
back to the first and completes the loop. Disconnecting a device from the loop may interrupt communications. Consequently, systems such as drive arrays that permit
345
346
hot-swapping of devices have built-in port bypass circuits that route the loop signaling
around the removed device.
The principal body governing the promotion and development of Fibre Channel standards is the Fibre Channel Association. You can get in-depth information about Fibre
Channel, products using it, and meeting schedules, from the association at the following
address:
Fibre Channel Association
12407 MoPac Expressway North 100-357
P.O. Box 9700
Austin, TX 78766-9700
Phone: 800-272-4618 or 512-328-8422
Fax: 512-328-8423
The Fibre Channel Loop Community, a working group thats part of the association, is
focused on the FC-AL standard. The address of the Fibre Channel Loop Community is
as follows:
Fibre Channel Loop Community
P.O. Box 2161
Saratoga, CA 95070
Phone: 408-867-6630
Web site: www.symbios.com/fclc/fclcmain.htm
Floppy Disk Interface

Some things are immutable, changing not at all as the years passthe constellations in the
sky, the crags and rocks of tall mountains, your credit card balance. The floppy disk interface fits in there pretty well, too. Its basically the same connection used on the first personal computer more than two decades ago. It sits there, waiting for you to slide in a boot
disk so it can show you all of its stuff.
Part of the reason for this immutability is that the function of the floppy disk hasnt
changed over the years. The floppy disk remains a boot device, a data-exchange system,
and a backup systemalthough an increasingly feeble one. At one time, however, it was
your only choice for each of these functions. The world has, of course, changed in the
intervening years, and the floppy disk is at best the fallback choice for each of these functions. Your needs and the rest of your computer hardware have long passed by floppy
technology. In the next few years, the floppy disk and its interface may be left behind.
But the floppy disk interface is not as unchangeable as you might think. During its short
history, its speed has notched up twice, its connectors have changed, and the range of
devices linking to it has broadened. And its role has changed. CDs have replaced the
floppy disk for just about every purpose except emergencies. You could live without a
floppy disk drive. Many sub-notebook computers get along perfectly well without them.
Every once in a while, though, the need for a floppy arises, and the interface proves
useful.
Controller
The floppy disk controller brings the interface to life. Its basic purpose is to convert the
requests from the BIOS or operating system, couched in terms of track and sector numbers, into the pulses that move the head to the proper location on the disk. In this translation function, the controller must make sense from the stream of unformatted pulses
delivered from the drive. It first must find the beginning of each track from the Index
pulse and then mark out each sector from the information embedded in the data stream.
When it identifies a requested sector, it must then read the information it contains and
convert that information from serial to parallel form so that it can be sent through the
computer bus. In writing, the controller must first identify the proper sector to write to
which is a read operationand then switch on the write current to put data into that sector before the next sector on the disk begins.
Most of the hard work of the controller is handled by a single integrated circuit, the 765
controller chip, or rather the equivalent of its circuitry inside your computers chipset.
The 765 circuitry works much like a microprocessor, carrying out certain operations in
response to commands it receives through registers connected to your computers I/O
ports. This programmability makes floppy disk controllers extremely versatile, able to
adapt to changes in media and storage format as the computer industry has evolved. None
of the essential floppy disk drive parameters are cast in stone or the silicon on the controller. The number of heads, tracks, and sectors on a disk are set by loading numbers into
the registers of the 765 circuitry. The values that the controller will use are normally
loaded into the controller when you boot up your computer.
Special software can reprogram your controller to make it read, write, and format floppy
disks that differ from the computer standard. The most common alternate format is the
Distribution Media Format used by Microsoft to stuff extra information on floppy disks
used for distributing software in the years before programs came on CD.
Operation
In its heart and in operation, the floppy disk interface is a glorified serial port to which
Dr. Frankensteins son, the mad engineer, grafted a cable and drive. The control electronics help the drive keep spinning at the right speed and know where to move its head.
347
348
Its so simple that if two drives get connected to a floppy disk controller, they get all the
same signals all the time, except for one. Two Drive Select signals are used to individually
select either the first or second drive, usually drives A or B. If the signal assigned to a particular drive is not present, all the other input and output circuits of the drive are deactivated, except for those that control the drive motor. In this way, two drives can share the
bulk of the wires in the controller cable without interference. However, this control
scheme also means that only one drive in a pair can be active at a time.
Two signals in the floppy disk interface control the head position of each of the attached
drives. One, Step Pulse, merely tells the stepper motor on the drive with its Drive Select
active to move one step (thats exactly one track) toward or away from the center of the
disk. The Direction signal controls which way the pulses move the head. If this signal is
active, the head moves toward the center.
To determine which of the two sides of a double-sided disk to read, one signal, called
Write Select, is used. When this signal is active, it tells the disk drive to use the upper
head. When no signal is present, the disk drive automatically uses the default (lower)
head.
Writing to disk requires two signals on the interface. Write Data contains the information
thats actually to be written magnetically onto the disk. It consists of nothing but a series
of pulses corresponding exactly to the flux transitions that are to be made on the disk. The
read/write head merely echoes these signals magnetically. As a fail-safe that precludes the
possibility of accidentally writing over valuable data, a second signal called Write Enable
is used. No write current is sent to the read/write head unless this signal is active.
Four signals are passed back from the floppy disk drive to the controller through the
interface. Two of these help the controller determine where the head is located. Track 0
indicates to the controller when the head is above the outermost track on the disk so that
the controller knows from where to start counting head-moving pulses. Index helps the
drive determine the location of each bit on a disk track. One pulse is generated on the
Index line for each revolution of the disk. The controller can time the distance between
ensuing data pulses based on the reference provided by the Index signal.
In addition, the Write Protect signal is derived from the sensor that detects the existence
or absence of a write-protect tab on a floppy disk. If a tab is present, this signal is active.
The Read Data signal comprises a series of electrical pulses that exactly matches the train
of flux transition on the floppy disk.
Cabling
The original computer floppy system is the standard for nearly all floppy disk systems. It
was designed so that you can install floppy disk drives with a minimum of thoughtwhich
gives you an idea of what manufacturers think of their assembly line workers. Once you
understand a few simple rules, you can do as good a job as they canor even better.
Drive Select Jumpers

New floppy disk drives simply plug into a floppy drive cable with no changes or settings.
Old drives, however, had drive-select jumpers or switches that allowed you to adjust a
given drive as unit zero or one. These settings allowed you to adjust a drive to suit systems that did not follow the current fashion (its not really a standard) in personal computers. When you install one of these older drives, you must adjust the switches or jumpers
to identify the drive as unit one rather than zero. Typically you install a jumper on the
pins marked DS1 (for Drive Select One).
Terminations
Similarly, older systems required the last drive (the one at the end of the cable) to be terminated with a small chip full of resistors. The second drive (if installed) was left unterminated. For practical reasons, termination is no longer a concern. The problems of
improper drive termination have proven minimal; manufacturers simply include a termination on all drives and design floppy disk controllers to work adequately whether they
have one or two terminations on a given cable.
Drive Cabling
The cable used by your floppy disk system is key to its proper operation. A twist in the
cable actually sets how the two drives get identified. The twist occurs in five conductors in
the ribbon cable between the two drive connectors. This twist reverses the drive-select
and motor-control signals in the cable as well as rearranges some of the ground wires in
the cable.
Because all drives are set up as the second drive, this reversal makes the drive attached to
the cable after the twist the first drive, drive A. In other words, drive A is attached to the
connector at the end of the cable, the one where the wire twist takes place. Drive B is
attached to the connector in the middle of the length of the cable. The third connector, at
the end of the cable with no twist, goes to the floppy disk controller or host adapter.
Figure 10.18 illustrates the floppy disk cable and the proper connections.
FIGURE 10.18
Classic floppy drive
cable showing proper
connections.
349
350
Connectors
Physical aspects of the floppy disk interface have changed subtly over the years. The original computer floppy disk controller used an edge connector to attach to the floppy disk
cable. Since about 1984, controllers generally have used pin connectors, although edge-connector products appeared into the 1990s. Similarly, the connectors on floppy disks
migrated from edge connectors to pin connectors. With older computers, this difference
can be critical when you need to get the correct cable. With modern computers, youll
want a cable that uses only pin connectors. Figure 10.19 contrasts the two connector
styles.
FIGURE 10.19
Floppy disk cable connectorsedge connector (left) and pin
connector.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
11
Ports
No computer has everything you need built in to it. The computer gains
its power from what you connect to itits peripherals. Somehow your
computer must be able to send data to its peripherals. It needs to connect, and it makes its connections through its ports.
Todays computers include two or three of these modern interfaces.
Those that youre likely to encounter include Universal Serial Bus (USB),
todays general-purpose choice; FireWire, most popular as a digital video
interface; IrDA, a wireless connection most often used for beaming data
between handheld computers; Bluetooth, a radio-based networking system
most suited for voice equipment; legacy serial ports, used by a variety of
slower devices such as external modems, drawing tablets, and even PCto-PC connections; and legacy parallel ports, most often used by printers.
Table 11.1 compares these port alternatives.
354
TA B L E 1 1 . 1
A Comparison of Serial Interfaces
Standard
Data Rate
(Current)
Medium
Devices
per Port
ACCESS.bus
100Kbps
Four-wire shielded cable
125
Bluetooth
723Kbps
Radio
FireWire (IEEE-1394)
400Mbps
Special six-wire cable
63
IrDA
4Mbps
Optical
126
Parallel (IEEE-1284)
2Mbps
25-conductor cable
Serial (RS-232C)
115,200bps
Twisted-pair
USB
480Mbps
Special four-wire cable
127
If you were only to buy new peripherals to plug into your computer, you could get along
with only USB ports. Designed to be hassle-free for installing new gear, USB is fast
enough (at least in its current version 2.0) that you need not consider any other connection. You can consider all the other ports special-purpose designsthe legacy pair (parallel and serial) for accommodating stuff that would otherwise be gathering dust in your
attic; FireWire for plugging in your digital camcorder; IrDA for talking to your notebook
computer; and Bluetooth for, well, youll think of something.
Universal Serial Bus

In 1995, Compaq, Digital, IBM, Intel, Microsoft, NEC, and Northern Telecom, determined to design a better interface, pooled their efforts and laid the groundwork for the
Universal Serial Bus, better known as USB. Later that year, they started the Universal
Serial Bus Implementers Forum and in 1996 unveiled the new interface to the world. The
world yawned.
Aimed at replacing both legacy serial and parallel port designs, the USB design corrected
all three of their shortcomings. To improve performance, they designed USB with a
12Mbps data rate (with an alternative low-speed signaling rate of 1.5Mbps). To eliminate
wiring hassles and worries about connector gender, crossover cables, and device types,
they developed a strict wiring system with exactly one type of cable to serve all interconnection needs. And to allow one jack on the back of a computer to handle as many peripherals as necessary, they designed the system to accommodate up to 127 devices per port.
In addition, they built in Plug-and-Play support so that every connection could be selfconfiguring. You could even hot-plug new devices and use them immediately without
reloading your operating system.
CHAPTER 11 PORTS
On April 27, 2000, a new group, led by Compaq, Hewlett-Packard, Intel, Lucent,
Microsoft, NEC, and Philips, published a revised USB standard, version 2.0. The key
change was an increase in performance, upping the speed from 12Mbps to 480Mbps. The
new system incorporates all the protocols of the old and is fully backward compatible.
Devices will negotiate the highest common speed and use it for their transfers.
Connectors and cabling remained unchanged.
Background
Designed for those who would rather compute than worry about hardware, the premise
underlying USB is the substitution of software intelligence for cabling confusion. USB
handles all the issues involved in linking multiple devices with different capabilities and
data rates with a layer-cake of software. Along the way, it introduces its own new technology and terminology.
USB divides serial hardware into two classes: hubs and functions. A USB hub provides
jacks into which you can plug functions. A USB function is a device that actually does
something. USBs designers imagined that a function may be anything you can connect
to your computer, including keyboards, mice, modems, printers, plotters, scanners, and
more.
Rather than a simple point-to-point port, the USB acts as an actual bus that allows you to
connect multiple peripherals to one jack on your computer with all the linked functions
(devices) sharing exactly the same signals. Information passes across the bus in the form of
packets, and all functions receive all packets. Your computer accesses individual functions
by adding a specific address to the packets, and only the function with the correct address
acts on the packets addressed to it.
The physical manifestation of USB is a porta jack on the back of your computer or in a
hub. Although your computers USB port can handle up to 127 devices, each physical
USB port connects to a single device. To connect multiple devices, you need multiple
jacks. Typically, a new computer comes equipped with two USB ports. When you need
more, you add a hub, which offers multiple jacks to let you plug in several devices. You can
plug one hub into another to provide several additional jacks and ports to connect more
devices.
The USB design envisions a hierarchical system with hubs connected to hubs connected
to hubs. In that each hub allows for multiple connections, the reach of the USB system
branches out like a treeor a trees roots. Figure 11.1 gives a conceptual view of the USB
wiring system.
355
356
FIGURE 11.1
The USB hierarchical
interconnection
scheme.
Host
Function
Hub
Function
Function
Function
Hub
Function
Hub
Function
Function
Your computer acts as the base hub for a USB system and is termed the host. The circuitry
in your computer that controls this integral hub and the rest of the USB system is called
the bus controller. Each USB system has one and only one bus controller.
Under USB 2.0, a device can operate at any of three speeds: Low speed is 1.5Mbps. Full
speed is 12Mbps, and high speed is 480Mbps.
USB 2.0 is backward compatible with USB 1.1all USB 1.1 devices will work with USB
2.0 devices, and vice versa, but USB 1.1 will impose its speed limit on USB 2.0 devices.
The mixing of speeds makes matters complicated. If you plug both a USB 2.0 and a USB
1.1 device into a USB 2.0 hub, both devices will operate at their respective top speeds. But
if a USB 1.1 hub appears in the chain between USB 2.0 devices, the slower hub will limit
the speed of the overall system. Plug any USB 2.0 device into a USB 1.1 hubeven a
USB 2.0 huband it will degrade to USB 1.1 operation (and if its a hub, so will all the
devices connected to that hub).
Other than this speed issue, the USB system doesnt care which device you plug into
which hub or how many levels down the hub hierarchy you put a particular device. All the
system requires is that you properly plug everything together following its simple rule:
Each device must plug into a hub. The USB software then sorts everything out. This software, making up the USB protocol, is the most complex part of the design. In comparison,
the actual hardware is simplebut the hardware wont work without the protocol.
The wiring hardware imposes no limit on the number of devices/functions you can connect in a USB system. You can plug hubs into hubs into hubs, fanning out into as many
ports as you like. You do face limits, however. The protocol constrains the number of
functions on one bus to 127 because of addressing limits. Seven bits are allowed for
encoding function addresses, and one of the potential addresses (128) is reserved.
CHAPTER 11 PORTS
In addition, the wiring limits the distance at which you can place functions from hubs.
The maximum length of a USB cable is five meters. Because hubs can regenerate signals,
however, your USB system can stretch out for greater distances by making multiple hops
through hubs.
As part of the Plug-and-Play process, the USB controller goes on a device hunt when you
start up your computer. It interrogates each device to find out what it is. It then builds a
map that locates each device by hub and port number. These become part of the packet
address. When the USB driver sends data out the port, it routes that data to the proper
device by this hub-and-port address.
USB requires specific software support. Any device with a USB connector will have the
necessary firmware to handle USB built in. But your computer will also require software
to make the USB system work. Your computers operating system must know how to send
the appropriate signals to its USB ports. All Windows versions starting with Windows 98
have USB support. Windows 95 and Windows NT do not. In addition, each function
must have a matching software driver. The function driver creates the commands or packages the data for its associated device. An overall USB driver acts as the delivery service,
providing the channelin USB terminology, a pipefor routing the data to the various
functions. Consequently, each USB you add to your computer requires software installation along with plugging in the hardware.
Connectors
The USB system involves four different styles of connectorstwo chassis-mounted jacks
and two plugs at the ends of cables. Each jack and plug comes in two varieties: A and B.
Hubs have A jacks. These are the primary outward manifestation of the USB portthe
wide, thin USB slots youll find on the back of your computer. The matching A plug
attaches to the cable that leads to the USB device. In the purest form of USB, this cable is
permanently affixed to the device, and you need not worry about any other plugs or jacks.
This configuration may someday become popular when manufacturers discover they can
save the cost of a connector by integrating the cable. Unfortunately, too many manufacturers have discovered that by putting a jack on their USB devices they save the cost of
the cable by not including it with the device.
To accommodate devices with removable cables (and manufacturers that dont want the
add the expense of a few feet of wire to their USB devices), the USB standard allows for a
second, different style of plug and jack meant only to be used for inputs to USB devices. If
a USB device (other than a hub) requires a connector so that, as a convenience, you can
remove the cable, it uses a USB B jack, which is a small, nearly square hole into which you
slide the mating B plug.
357
358
The motivation behind this multiplicity of connectors is to prevent rather than cause confusion. All USB cables will have an A plug at one end and a B plug at the other. One end
must attach to a hub and the other to a device. You cannot inadvertently plug things
together incorrectly.
Because all A jacks are outputs and all B jacks are inputs, only one form of detachable
USB cable existsone with an A plug at one end and a B plug at the other. No crossover
cables or adapters are needed for any USB wiring scheme.
Cable
The USB system uses two kinds of cablethat meant for low-speed connections and that
meant for full- and high-speed links. But you only have to worry about the higher-speed
variety. Low-speed cables, those capable of supporting only 1.5Mbps signaling rates, must
be permanently attached to the equipment using them. Higher-speed cables can be either
permanently attached or removable.
Both speeds of physical USB wiring use a special four-wire cable. Two conductors in the
cable transfer the data as a differential digital signal. That is, the voltage on the two conductors is of equal magnitude and opposite polarity so that when subtracted from one
another (finding the difference) the result cancels out any noise that ordinarily would add
equally to the signal on each line. In addition, the USB cable includes a power signal,
nominally five-volts DC, and a ground return. The power signal allows you to supply
power for external serial devices through the USB cable. The two data wires are twisted
together as a pair. The power cables may or may not be.
The difference between low- and higher-speed cables is that the capacitance of the lowspeed cable is adjusted to support its signaling rate. In addition, the low speed does not
need twisted-pair wires, and the standard doesnt require them.
All removable cables must be able to handle both full-speed and high-speed connections.
To achieve its high data rate, the USB specification requires that certain physical characteristics of the cable be carefully controlled. Even so, the maximum length permitted for
any USB cable is five meters.
One limit on cable length is the inevitable voltage drop suffered by the power signal. All
wires offer some resistance to electrical flow, and the resistance is proportional to the wire
gauge. Hence, lower wire gauges (thicker wires) have lower resistance. Longer cables
require lower wire gauges. At maximum length, the USB specification requires 20-gauge
wire, which is one step (two gauge numbers) thinner than ordinary lamp cord.
The individual wires in the USB cable are color-coded. The data signals form a greenwhite pair, with the +Data signal on green. The positive five-volt signal rides on the red
wire. The ground wire is black. Table 11.2 sums up this color code.
CHAPTER 11 PORTS
TA B L E 1 1 . 2
USB Cable Color Code
Signal
Color
+Data
Green
-Data
White
VCC
Red
Ground
Black
Normally, you cannot connect one computer to another using USB. The standard calls
for only one USB controller in the entire interconnection system. Physically the cabling
system prevents you from making such a connection. The exception is that some cables
have a bridge built in that allows two USB hosts to talk to each other. The bridge is active
circuitry that converts the signals.
Protocol
As with all more recent interface introductions, the USB design uses a packet-based protocol using No Return-to-Zero Inverted (NRZI) data coding. All message exchanges
require the swapping of three packets. The exchange begins with the host sending out a
token packet. The token packet bears the address of the device meant to participate in the
exchange as well as control information that describes the nature of the exchange. A data
packet holds the actual information that is to be exchanged. Depending on the type of
transfer, either the host or the device will send out the data packet. Despite the name, the
data packet may contain no information. Finally, the exchange ends with a handshake
packet, which acknowledges the receipt of the data or other successful completion of the
exchange. A fourth type of packet, called Special, handles additional functions.
Each packet starts with two componentsa Sync Field and a Packet Identifiereach one
byte long. The Sync Field is a series of bits that serves as a consistent burst of clock pulses
so that the devices connected to the USB bus can reset their timing and synchronize
themselves to the host. The Sync Field appears as three on/off pulses followed by a
marker two pulses wide. The Packet Identifier byte includes four bits to define the nature
of the packet itself and another four bits as check-bits that confirm the accuracy of the
first four. The four bits provide a code that allows for the definition of 16 different kinds
of packets.
USB uses the 16 values in a two-step hierarchy. The two more significant bits specify one
of the four types of packets. The two lesser significant bits subdivide the packet category.
Table 11.3 lists the PIDs of the four basic USB packet types.
359
360
TA B L E 1 1 . 3
USB Packet Identifications
Bit-Pattern
Packet Type
XX00XX11
Special packet
XX01XX10
Token packet
XX10XX01
Handshake packet
XX11XX00
Data packet
Token Packets
Only the USB host sends out token packets. Each token packet takes up four bytes, which
are divided up into five functional parts. Figure 11.2 graphically shows the layout of a
token packet.
FIGURE 11.2
Functional parts of a
USB Token Packet.
Sync Field
Packet
Identifier
Address
Field
Endpoint
Error
Check
8 bits
8 bits
7 bits
4 bits
5 bits
The two bytes take the standard form of all USB packets. The first byte is a Sync Field
that marks the beginning of the tokens bit-stream. The second byte is the Packet Identifier.
The PID byte defines four types of token packets. These include an Out packet that carries data from the host to a device; an In packet that carries data from the device to the
host; a Setup packet that targets a specific endpoint; and a Start of Frame packet that
helps synchronize the system.
Data Packets
The actual information transferred through the USB system takes the form of data packets.
As with all USB packets, a data packet begins with a one-byte Sync Field followed by the
Packet Identifier. The actual data follows as a sequence of 0 to 1,023 bytes. A two-byte cyclic
redundancy check verifies the accuracy of only the Data Field, as shown in Figure 11.3. The
PID field relies on its own redundancy check mechanism.
FIGURE 11.3
Constituents of a USB
data packet.
Sync Field
Packet
Identifier
Data
Field
Error Check
8 bits
8 bits
0 to 1032 bits
16 bits
CHAPTER 11 PORTS
FireWire
Handshake Packets
Handshake packets handle flow-control in the USB system. All are two bytes long, comprised of nothing more than the Sync Field and a Packet Identifier byte that acknowledges
proper receipt of a packet, as shown in Figure 11.4.
FIGURE 11.4
Constituents of a USB
handshake packet.
Sync Field
Packet
Identifier
8 bits
8 bits
Standard
The USB standard is maintained by USB Implementers Forum. You can download a complete copy of the current version of the specifications from the USB Website.
USB Implementers Forum
5440 SW Westgate Dr., Suite 217
Portland, OR 97221
Phone: 503-296-9892
Fax: 503-297-1090
Web site: www.usb.org
FireWire
Also known as IEEE-1394, I.link, and DV, FireWire is a serial interface thats aimed at
high-throughput devices such as a hard disk and tape drives, as well as consumer-level
multimedia devices such as digital camcorders, digital VCRs, and digital televisions.
Originally it was conceived as a general-purpose interface suitable for replacing legacy serial ports, but with blazing speed. However, it has been most used in digital videoat least
so far. Promised new performance and a choice of media may rekindle interest in
FireWire as a general-purpose, high-performance interconnection system.
For the most part, FireWire is a hardware interface. It specifies speeds, timing, and a connection system. The software side is based on SCSI. In fact, FireWire is one of the several
hardware interfaces included in the SCSI-3 standards.
As with other current port standards, FireWire continues to evolve. Development of the
standard began when the Institute of Electrical and Electronic Engineers (IEEE) assigned
a study group the task of clearing the murk of thickening morass of serial standards in
September 1986. Hardly four months later (in January 1987), the group had already outlined basic concepts underlying FireWire, some of which still survive in todays standard
including low cost, a simplified wiring scheme, and arbitrated signals supporting multiple
361
362
devices. The IEEE approved the first FireWire standard (as IEEE 1394-1995) in 1995,
based on a design with one connector style and two speeds (100 and 200Mbps).
In the year 2000, the institute approved the standard IEEE 1394a-2000, which boasts a
new, miniaturized connector, a higher speed (400Mbps), and streamlined signaling that
makes connections quicker (because of reduced overhead) and more reliable.
As this is written, the engineers at the institute are developing a successor standard, IEEE
1394b, that will quadruple the speed of connections, add both fiber-optic and twisted-pair
wiring schemes, and add a new, more reliable transport protocol.
For now, FireWire is best known as a 400Mbps connection system for plugging digital
camcorders into computers, letting you capture video images (live or tape), edit them, and
publish them on CD or DVD.
Overview
FireWire differs from todays other leading port standard, USB, in that it is a point-topoint connection system. That is, you plug a FireWire device directly into the port on
your computer. To accommodate more than one FireWire device (the standard allows for
a maximum of 63 interconnected devices), the computer host may have multiple jacks or
the FireWire device may have its own input jack so that you can daisy-chain multiple
devices to a single computer port.
Although FireWire does not use hubs in the network or USB sense, in its own nomenclature it does. In the FireWire scheme, a device with a single FireWire port is a leaf. A
device with two ports is called a pass-through, and a device with three ports is called a
branch or hub. Pass-through and branch nodes operate as repeaters, reconstituting the digital signal for the next hop. Each FireWire system also has a single root, which is the foundation around which the rest of the system organizes itself.
You can daisy-chain devices with up to 16 links to the chain. After that, the delays in
relaying the signals from device to device go beyond those set in the standard. Accommodating larger numbers of devices requires using branches to create parallel data paths.
Under the current standard (1394a), FireWire allows a maximum cable length of 4.5
meters (about 15 feet). With 16 links to a daisy-chain, two FireWire devices could be
separated by as much as 72 meters (almost 200 feet).
Each FireWire cable contains two active connections for a full-duplex design (signals
travel both ways simultaneously in the cable on different wire pairs). Connectors at each
end of the cable are the same, so wiring is easyyou just plug things together. Software
takes care of all the details of the connection. The exception is that the 1394a standard
also allows for a miniaturized connector to fit in tight places (such as a camcorder).
CHAPTER 11 PORTS
FireWire
FireWire also allows for engineers to use the same signaling system for backplane designs.
That is, FireWire could be used as an expansion bus inside a computer as well as the port
linking to external peripherals. Currently, however, FireWire is not used as a backplane
inside personal computers.
The protocol used by FireWire uses 64-bit addressing. The 63 device limitation per chain
results from only six bits being used for node identification. The rest of the addressing
bits provide for large networks and the use of direct memory addressing10 bits for network identifications and 48 bits for memory addresses (the same as the latest Intel microprocessor, enough for uniquely identifying 281TB of memory per device). A single device
may use multiple identifications.
Signaling
To minimize noise, data connections in FireWire use differential signals, which means it
uses two wires that carry the same signal but of opposite polarity. Receiving equipment
subtracts the signal on one wire from that on the other to find the data as the difference
between the two signals. The benefit of this scheme is that any noise gets picked up by the
wires equally. When the receiving equipment subtracts the signals on the two wires, the
noise gets eliminatedthe equal noise signals subtracted from each other equals zero.
The original FireWire standard used a patented form of signal coding called data strobe
coding, using two differential wire pairs to carry a single data stream. One pair carried the
actual data; the second pair, called the strobe lines, complimented the state of the data pair
so that one and only one of the pairs changed polarity every clock cycle. For example, if
the data line carried two sequential bits of the same value, the strobe line reversed polarity
to mark the transition between them. If a sequence of two bits changed the polarity of the
data lines (a one followed by a zero, or zero followed by a one), the strobe line did not
change polarity. Summing the data and strobe lines together exactly reconstructed the
clock signal of the sending system, allowing the sending and receiving devices to precisely
lock up.
FireWire operates as a two-way channel, with different pairs of wire used for sending and
receiving. When one pair is sending data, the other operates as its strobe signal. In receiving data, the pair used for strobe in sending contains the data, and the other (data in sending) carries the receive strobe. In other words, as a device shifts from sending and
receiving, it shifts which wire pairs it uses for data and strobe.
The 1394b specification alters the data coding to use a system termed 8B/10B coding,
developed by IBM. The scheme encodes eight-bit bytes in 10-bit symbols that guarantee a
sequence of more than five identical bits never occurs and that the number of ones and
zeros in the code balancea characteristic important to engineers because it results in no
shift of the direct current voltage in the system.
363
364
Configuration
FireWire allows you to connect multiple devices together and uses an addressing system so
that the signals sent through a common channel are recognized only by the proper target
device. The linked devices can independently communicate among themselves without the
intervention of your computer. Each device can communicate at its own speeda single
FireWire connection shifts between speeds to accommodate each device. Of course, a lowspeed device may not be able to pass through higher-speed signals, so some forethought is
required to put together a system in which all devices operate at their optimum speeds.
FireWire eliminates such concern about setting device identifications with its own automated configuration process. Whenever a new device gets plugged into a FireWire system
(or when the whole system gets turned on), the automatic configuration process begins.
By signaling through the various connections, each device determines how it fits into the
system, either as a root node, a branch, a pass-through, or a leaf. The node also sends out
a special clock signal. Once the connection hierarchy is set up, the FireWire devices
determine their own ID numbers from their location in the hierarchy and send identifying
information (ID and device type) to their host.
You can hot-plug devices into a FireWire tree. That is, you can plug in a new device to a
group of FireWire devices without switching off the power. When you plug in a new
device, the change triggers a bus reset that erases the systems stored memory of the previous set of devices. Then the entire chain of devices goes through the configuration
process again and is assigned an address. The devices then identify themselves to one
another and wait for data transfers to begin.
Arbitration
FireWire transfers data in packets, a block of data preceded by a header that specifies where
the data goes and its priority. In the basic cable-based FireWire system, each device sharing
a connection gets a chance to send one packet in an arbitration period thats called a fairness
interval. The various devices take turns until all have had a chance to use the bus. After each
packet gets sent, a brief time called the sub-action gap elapses, after which another device can
send its packet. If no devices start to transmit when the sub-action gap ends, all devices wait
a bit longer, stretching the time to an arbitration reset gap. After that time elapses, a new
fairness interval begins, and all devices get to send one more packet. The cycle continues.
To handle devices that need a constant stream of data for real-time display, such as video or
audio signals, FireWire uses a special isochronous mode. Every 125 microseconds, one
device in the FireWire that needs isochronous data sends out a special timing packet that signals that isochronous devices can transmit. Each takes a turn in order of its priority, leaving a
brief isochronous gap delay between their packets. When the isochronous gap delay stretches
out to the sub-action gap length, the devices using ordinary asynchronous transfers take over
until the end of the 125-microsecond cycle when the next isochronous period begins.
CHAPTER 11 PORTS
FireWire
The scheme guarantees that video and audio gear can move its data in real time with a
minimum of buffer memory. (Audio devices require only a byte of buffer; video may need
as many as six bytes.) The 125-microsecond period matches the sampling rate used by
digital telephone systems to help digital telephone services.
The new 1394b standard also brings a new arbitration system called Bus
Owner/Supervisor/Selector (BOSS). Under this scheme, a device takes control of the bus as
the BOSS by being the last device to acknowledge the receipt of a packet sent to it (rather
than broadcast over the entire tree) or by receiving a specific grant of control. The BOSS
takes full command of the tree, even selecting the next node to be the BOSS.
Connectors
In the original FireWire version only a single, small, six-pin connector was defined for all
purposes. Each cable had an identical connector on each end, and all FireWire ports were
the same. The contacts were arranged in two parallel rows on opposite sides inside the
metal-and-plastic shield of the connector. The asymmetrical D shape of the active end of
the connector ensured that you plugged it in properly. Figure 11.5 shows this connector.
FIGURE 11.5
A FireWire connector.
The revised 1394a standard added a miniaturized connector. To keep it compact, the
design omitted the two power contacts. This design is favored for personal electronic
devices such as camcorders. Figure 11.6 shows this connector.
FIGURE 11.6
A miniaturized
FireWire connector.
365
366
The 1394b standard will add two more connectors to the FireWire arsenal, each with
eight contacts. The Beta connector is meant for systems that use only the new 1394b signaling system and do not understand the earlier versions of the standard. In addition, the
new standard defines a bilingual connector, one that speaks both the old and new FireWire
standards. The new designs are keyed so that while both beta and bilingual connectors
will fit a bilingual port, only a beta connector will fit a beta-only port. Figure 11.7 shows
this keying system.
FIGURE 11.7
The IEEE 1394b
connector keying.
IEEE 1394B Beta jack
IEEE 1394B Bilingual jack
Cabling
In current form, FireWire uses ordinary copper wires in a special cable design. Two variations are allowedone with solely four signal wires and one with the four signal wires and
two power wires. In both implementations, data travels across two shielded twisted pairs
of AWG 28 gauge wire, with a nominal impedance of 110 ohms. In the six-wire version,
two AWG 22 gauge wires additionally carry power at 8 to 33 volts with up to 1.5 amperes
to power a number of peripherals. Another shield will cover the entire collection of conductors.
The upcoming IEEE 1394b standard also allows for two forms of fiber optical connectionglass and plasticas well as ordinary Category 5 twisted-pair network cable. The
maximum length of plastic fiber optical connections is 50 meters (about 160 feet); for glass
optical fiber, the maximum length is 100 meters (about 320 feet). Either style of optical
connection can operate at speeds of 100 or 200Mbps. Category 5 wire allows connections
of up to 100 meters but only at the lowest data rate sanctioned by the standard, which is
100Mbps.
All FireWire cables are crossover cables. That is, the signals that appear on pins 1 and 2
at one end of the cable cross over to pins 3 and 4 at the other end. This permits all
FireWire ports to be wired the same and serve both as inputs and outputs. The same
connector can be used at each end of the cable, and no keying is necessary, as is the case
with USB.
The FireWire wiring scheme depends on each of the devices that are connected together
to relay signals to the others. Pulling the plug to one device could potentially knock down
the entire connection system. To avoid such difficulties and dependencies, FireWire uses
its power connections to keep in operation the interface circuitry in otherwise inactive
devices. These power lines could also supply enough current to run entire devices. No
CHAPTER 11 PORTS
IrDA
device may draw more than three watts from the FireWire bus, although a single device
may supply up to 40 watts. The FireWire circuitry itself in each interface requires only
about two milliwatts.
The FireWire wiring standard allows for up to 16 hops of 4.5 meters (about 15 feet) each.
As with current communications ports, the standard allows you to connect and disconnect
peripherals without switching off power to them. You can daisy-chain FireWire devices or
branch the cable between them. When you make changes, the network of connected
devices will automatically reconfigure itself to reflect the alterations.
IrDA
The one thing you dont want with a portable computer is a cable to tether you down; yet
most of the time you have to plug into one thing or another. Even a simple and routine
chore like downloading files from your notebook machine into your desktop computer
gets tangled in cable trouble. Not only do you have to plug in both ends, reaching behind
your desktop machine is only a little more elegantly done than fishing into a catch basin
for a fallen quarterand, more likely than not, unplugging something else that youll
inevitably need later only to discover the dangling cordbut youve got to tote that
writhing cable along with you wherever you go. There has to be a better way.
There is. You can link your computer to other systems and components with a light beam.
On the rear panel of many notebook computers, youll find a clear LED or a dark red
window through which your system can send and receive invisible infrared light beams.
Although originally introduced to allow you to link portable computers to desktop
machines, the same technology can tie in peripherals such as modems and printers, all
without the hassle of plugging and unplugging cables.
History
On June 28, 1993, a group of about 120 representatives from 50 computer-related companies got together to take the first step in cutting the cord. Creating what has come to be
known as the Infrared Developers Association (IrDA), this group aimed at more than making
your computer more convenient to carry. It also saw a new versatility and, hardly incidentally, a way to trim its own costs.
The idea behind the get together was to create a standard for using infrared light to link
your computer to peripherals and other systems. The technology had already been long
established, not only in television remote controls but also in a number of notebook computers already on the market. Rather than build a new technology, the goal of the group
was to find common ground, a standard so that the products of all manufacturers could
communicate with the computer equivalent of sign language.
367
368
Hardly a year later, on June 30, 1994, the group approved its first standard. The original
specification, now known as IrDA version 1.0, essentially gave the standard RS-232C port
an optical counterpart, one with the same data structure and, alas, speed limit. In August
1995, IrDA took the next step and approved high-speed extensions that pushed the wireless data rate to 4Mbps.
Overview
More than a gimmicky cordless keyboard, IrDA holds an advantage that makes computer
manufacturersparticularly those developing low-cost machineseye it with interest. It
can cut several dollars from the cost of a complex system by eliminating some expensive
hardware, a connector or two, and a cable. Compared to the other wireless technology,
radio, infrared requires less space because it needs only a tiny LED instead of a larger and
more costly antenna. Moreover, infrared transmissions are not regulated by the FCC as
are radio transmissions. Nor do they cause interference to radios, televisions, pacemakers,
and airliners. The range of infrared is more limited than radio and restricted to the line of
sight over a narrow angle. However, these weaknesses can become strengths for those who
are security conscious.
The original design formulated by IrDA was for a replacement for serial cables. To make
the technology easy and inexpensive to implement with existing components, it was based
on the standard RS-232C port and its constituent components. The original IrDA standard used asynchronous communication using the same data frame legacy serial ports as
well as its data rates from 2400 to 115,200 bits per second.
To keep power needs low and prevent interference among multiple installations in a single
room, IrDA kept the range of the system low, about one meter (three feet). Similarly, the
IrDA system concentrates the infrared beam used to carry data because diffusing the beam
would require more power for a given range and be prone to causing greater interference
among competing units. The laser diodes used in the IrDA system consequently focus
their beams into a cone with a spread of about 30 degrees.
After the initial serial-port replacement design was in place, IrDA worked to make its
interface suitable for replacing parallel ports as well. That goal led to the creation of the
IrDA high-speed standards for transmissions at data rates of 0.576, 1.152, and 4.0Mbps.
The two higher speeds use a packet-based synchronous system that requires a special
hardware-based communication controller. This controller monitors and controls the flow
of information between the host computers bus and communications buffers.
Consequently, a watershed of differences separate low-speed and high-speed IrDA systems. Although IrDA designed the high-speed standard to be backward compatible with
old equipment, making the higher speeds work requires special hardware. In other words,
although high-speed IrDA devices can successfully communicate with lower-speed units,
CHAPTER 11 PORTS
IrDA
such communications are constrained to the speeds of the lower-speed units. Low-speed
units cannot operate at high speeds without their hardware being upgraded.
IrDA defines not only the hardware but also the data format used by its system. The
group has published six standards to cover these aspects of IrDA communications. The
hardware itself forms the physical layer. In addition, IrDA defines a link access protocol
termed IrLAP and a link management protocol called IrLMP that describe the data formats used to negotiate and maintain communications. All IrDA ports must follow these
standards. In addition, IrDA has defined an optional transport protocol and optional Plugand-Play extensions to allow for the smooth integration of the system into modern computers. The groups IrCOMM standard describes a standard way for infrared ports to
emulate conventional computer serial and parallel ports.
Infrared Light
Infrared light is invisible electromagnetic radiation that has a wavelength longer than that
of visible light. Whereas you can see light that ranges in wavelength from 400 angstroms
(deep violet) to 700 angstroms (dark red), infrared stretches from 700 angstroms to 1000
or more. IrDA specifies that the infrared signal used by computers for communication has
a wavelength between 850 and 900 angstroms.
Data Rates
All IrDA ports must be able to operate at one basic speed9600 bits per second. All other
speeds are optional.
The IrDA specification allows for all the usual speed increments used by conventional
serial ports, from 2400bps to 115,200bps. All these speeds use the default modulation
scheme, Return-to-Zero Inverted (RZI). High-speed IrDA version 1.1 adds three additional speeds, 576Kbps, 1.152Mbps, and 4.0Mbps, based on a pulse-position modulation
scheme.
Regardless of the speed range implemented by a system or used for communications,
IrDA devices first establish communications at the mandatory 9600bps speed using the
Link Access Protocol. Once the two devices establish a common speed for communicating, they switch to it and use it for the balance of their transmissions.
Pulse Width
The infrared cell of an IrDA transmitter sends out its data in pulses, each lasting only a
fraction of the basic clock period or bit-cell. The relatively wide spacing between pulses
makes each pulse easier for the optical receiver to distinguish.
At speeds up to and including 115,200 bits per second, each infrared pulse must be at least
1.41 microseconds long. Each IrDA data pulse nominally lasts just 3/16th of the length of
a bit-cell, although pulse widths a bit more than 10 percent greater remain acceptable. For
369
370
example, each bit cell of a 9600bps signal would occupy 104.2 microseconds (that is, one
second divided by 9600). A typical IrDA pulse at that data rate would last 3/16th that
period, or 19.53 microseconds.
At higher speeds, the minimum pulse length is reduced to 295.2 nanoseconds at 576Kbps
and to only 115 nanoseconds at 4.0Mbps. At these higher speeds, the nominal pulse width
is one-quarter of the character cell. For example, at 4.0Mbps, each pulse is only 125
nanoseconds long. Again, pulses about 10 percent longer remain permissible. Table 11.4
summarizes the speeds and pulse lengths.
TA B L E 1 1 . 4
IrDA Speeds and Modulation
Signaling Rate
Modulation
Pulse Duration
2.4Kbps
RZI
78.13 microseconds
9.6Kbps
RZI
19.53 microseconds
19.2Kbps
RZI
9.77 microseconds
38.4Kbps
RZI
4.88 microseconds
57.6Kbps
RZI
3.26 microseconds
115.2Kbps
RZI
1.63 microseconds
0.576Mbps
RZI
434.0 nanoseconds
1.152Mbps
RZI
217.0 nanoseconds
4.0Mbps
4PPM, single pulse
125 nanoseconds
4.0Mbps
4PPM, double pulse
250.0 nanoseconds
Modulation
Depending on the speed at which a link operates, it may use one of two forms of modulation. At speeds lower than 4.0Mbps, the system employs Return-to-Zero Invert (RZI) modulation.
At the 4.0Mbps data rate, the IrDA system shifts to pulse position modulation. Because the
IrDA system involves four discrete pulse positions, it is abbreviated 4PPM.
IrDA requires data to be transmitted only in eight-bit format. In terms of conventional
serial-port parameters, a data frame for IrDA comprises a start bit, eight data bits, no parity bits, and a stop bit, for a total of 10 bits per character. Note, however, that zero insertion may increase the length of a transmitted byte of data. Any inserted zeroes are
removed automatically by the receiver and do not enter the data stream. No matter the
form of modulation used by the IrDA system, all byte values are transmitted with the least
significant bit first.
CHAPTER 11 PORTS
Bluetooth
Note that with RZI modulation, long sequences of logical ones will suppress pulses for
the entire duration of the sequence. To prevent such a lengthy gap from appearing in the
signal and causing a loss of sync, moderate speed IrDA systems add extra pulses to the signal with bit-stuffing (as discussed in Chapter 8).
Format
The IrDA system doesnt deal with data at the bit or byte level but instead arranges the
data transmitted through it in the form of packets, which the IrDA specification also
terms frames. A single frame can stretch from 5 to 2050 bytes (and sometimes more) in
length. As with other packetized systems, an IrDA frame includes address information,
data, and error correction, the last of which is applied at the frame level. The format of
the frame is rigidly defined by the IrDA Link Access Protocol standard, discussed later.
Aborted Frames
Whenever a receiver detects a string of seven or more consecutive logical onesthat is,
an absence of optical pulsesit immediately terminates the frame in progress and disregards the data it received (which is classed as invalid because of the lack of error-correction data). The receiver then awaits the next valid frame, signified by a start-of-frame flag,
address field, and control field. Any frame that ends in this summary manner is termed an
aborted frame.
A transmitter may intentionally abort a frame or a frame may be aborted because of an
interruption in the infrared signal. Anything that blocks the light path will stop infrared
pulses from reaching the receiver and, if long enough, abort the frame being transmitted.
Interference Suppression
High-speed systems automatically mute lower-speed systems that are operating in the
same environment to prevent interference. To stop the lower-speed link from transmitting, the high-speed system sends out a special Serial Infrared Interaction Pulse (SIP) at
intervals no longer than half a second. The SIP is a pulse 1.6 microseconds long, followed
by 7.1 microseconds of darkness, parameters exactly equal to a packet start pulse. When
the low-speed system sees what it thinks is a start pulse, it automatically starts looking for
data at the lower rates, suppressing its own transmission for half a second. Before it has a
chance to start sending its own data (if any), another SIP quiets the low-speed system for
the next half second.
Bluetooth
Radio yields the most versatile connection system: no wires and no worries. Because radio
waves can slip through most office walls, desktop ornaments, office supplies, and even
employees, radio-based links eliminate the line-of-sight requirements of optical links
such as IrDA. Linking devices with radio waves consequently yields the most convenient
371
372
connection for workers to free their peripherals from the chains of interconnecting cables.
A radio link can provide a reliable, cord-free connection that eliminates the snarl on the
rear of every desktop PC. It also allows you to link your wireless devicesin particular,
your cell phoneto your PC and keep everything wireless. Hardly a novel idea, of course,
but one that has been a long time coming for practical connections.
Finally, a single standard may bring the wireless dream to life. To provide common
ground and a standard for radio-based connections between PCs, their peripherals, and
related communications equipment, several major corporations worked together to
develop the Bluetooth specification. They designed the standard for the utmost in convenience, coupled with low cost but sacrificing rangeBluetooth is a short-range system
suitable for linking devices in an office suite rather than across miles like a cell phone.
Originally conceived as a way to link cellular devices to PCs, the actual specification transcends its origins. Bluetooth makes possible not only cell phone connections but also
could allow you to use your keyboard or mouse without a physical connection to your PC
and without fretting about office debris blocking optical signals. But Bluetooth is more
than a simple interface. It can become a small wireless network of intercommunicating
devices, handling both voice and data with equal ease. Although not a rival to traditional
networking systemsits speed limitations alone see to thatBluetooth adds versatility
that combines cell phone and PC technology.
The Bluetooth promoters refer to its multidevice links as a piconet (smaller than even a
micronet), able to link up to eight devices. Bluetooth allows even greater assemblages of
equipment by linking piconets together and accommodating temporarily inactive equipment within its reach.
On the other hand, the data speed of the Bluetooth system is modest. At most, Bluetooth
can move bits at a claimed rate of about 723Kbps asymmetricallythat is, the high rate is
in one direction; the return channel is slower, about one-fifth that rate. Moreover,
Bluetooth slows to accommodate bidirectional data and phone conversations. Despite the
modest data rate, however, the Bluetooth bit-rate is high enough to handle three simultaneous telephone conversations or a combination of voice and data simultaneously.
Good as it sounds, Bluetooth currently has a number of handicaps to overcome. It is not
supported by any version of Microsoft Windows in current release (including the initial
release of Windows XP). Microsoft, however, promises to add its own native support for
Bluetooth in subsequent releases of Windows.
History
Certainly Bluetooth was not the first attempt at creating radio-based data links for computers. Wireless schemes for exchanging data have been around longer than personal
computers. But Bluetooth differs from any previous radio-based data-exchange system in
CHAPTER 11 PORTS
Bluetooth
that it was conceived as an open standard for the computer and communications industries
to facilitate the design of compatible wireless hardware.
As with so many modern standards, Bluetooth represents the work of an industry consortium. In May 1998, representatives from five major corporations involved with PCs, office
equipment, and cellular telephones jointly conceived the idea of Bluetooth and began working toward creating the standard. The five founders were Ericsson (Telefonaktiebolaget LM
Ericsson), International Business Machines Corporation, Intel Corporation, Nokia
Corporation, and Toshiba Corporation. Together they formed the Bluetooth Special
Interest Group (SIG) and started work on the standard and the technologies needed to
make it a reality. The SIG released the first version of the specification, Bluetooth 1.0, on
July 24, 1999. A slightly revised version was released in December 1999.
Membership in the Bluetooth SIG grew to nine on December 1, 1999, when 3Com
Technologies, Lucent Technologies, Microsoft Corporation, and Motorola, Inc., joined
the group. In addition, over 1,200 individuals and companies have adopted the technology
by entering an agreement with the SIG that allows them to use the standard and share the
intellectual property required to implement it.
Although support for Bluetooth has been slow in coming, manufacturers have adapted the
technology for low-speed computer peripherals (such as wireless keyboards and mice).
Owing to the success of other wireless technologies, most successful applications of
Bluetooth are in communications products.
Overview
Bluetooth is a wireless packetized communications system that allows multiple devices to
share data in a small network. Heir to both cell phone and digital technologies, it nestles
between several existing standards, embracing them. It can link to your PC using a USB
connection, and it shares logical layers with IrDA. It not only handles data like a traditional serial port but also can carry more than 60 RS-232C connections.
In theory, a Bluetooth system operates entirely transparently. Devices link themselves
together without you having to do anything. All you need to do is turn on your Bluetooth
devices and bring them within range of one another. For example, available devices should
automatically pop up on your Windows desktopat least once Windows gains Bluetooth
support. You can then drag files to and from the device as if it were a local Windows
resource.
Behind the scenes, however, things arent quite so simple. The Bluetooth system must
accommodate a variety of device and data types. It needs to keep in constant contact with
each device. It must be able to detect when a new device appears and when other devices
get switched off or venture out of range. It has to moderate the conversations between
units, ensuring they dont all try to talk at the same time and interfere with one another.
373
374
Software Side
As an advanced interface, Bluetooth heavily processes the raw data it transmits. It repackages serial data bits into packets with built-in error control. It then combines a series of
packets of related serial data into links. It further processes the links through multiplexing
so that several serial streams can simultaneously share a single Bluetooth connection.
Bluetooth packetizes data, breaking a serial input stream into small pieces, each containing
address and optional error-correction information. A series of packets that starts as a single data stream and is later reconstructed into a replica of that stream is a link.
The Bluetooth standard supports two kinds of data links: synchronous and asynchronous.
Synchronous data is typically voice information, such as audio from telephone conversations. Asynchronous data is typically computer data. The chief difference is that synchronous data is time dependent, so synchronous packets get transmitted once without regard
to their reception. If a synchronous packet gets lost during transmission, it is forever lost.
Synchronous links between Bluetooth devices provide a full-duplex channel with an effective data rate of 64Kbps in each direction. In effect, a synchronous link is a standard digital telephone channel with eight-bit resolution and an 8-KHz sampling rate. The
Bluetooth standard allows for two devices to simultaneously share three such synchronous
links, the equivalent of three real-time telephone conversations. All links begin asynchronously because commands can only be sent in asynchronous packets. After the link is
established, the master and slave can negotiate to switch over to a synchronous link for
voice transfers or to move data asynchronously.
Each piconet has a single master and, potentially, multiple slaves. Each piconet shares a
single communications channel with all the devices (master and slave) locked together on
a common frequency and using a common clock, as discussed later. The single channel is
subdivided into one or more linksasynchronous and/or synchronous.
To handle contention between multiple links on its single channel, Bluetooth uses timedivision multiplexing. That is, each separate packet of a link gets a time period for transmission. The standard divides the communications channel into time slots, each 625
microseconds long. The shortest single packet fits a single slot with room to spare,
although Bluetooth allows packets to stretch out for up to five slots. Bluetooth allows a
maximum length of single-slot packets of 366 microseconds. The system accommodates
larger packets by letting them extend through up to five slots, filling the entire time of
four of the slots and part of the fifth.
In the Bluetooth system, each packet also defines a hop. That is, after each packet is sent,
the Bluetooth system switches to (or hops to) another carrier frequency. As noted later,
frequency-hopping helps ensure the integrity of Bluetooth transmissions. The minimum
hop length corresponds to a single slot, although a hop can last for up to five slots to
accommodate a long packet.
CHAPTER 11 PORTS
Bluetooth
The time division duplexing of the Bluetooth system works by assigning even-numbered
slots to the master and odd-numbered slots to the slaves. Masters can begin their transmissions only in even-numbered slots. If a packet lasts for an even number of slots (two or
four), no slave can begin until the next odd-numbered slot. In effect, then, packets use an
odd number of slots even if they use only a shorter, even number of slots.
Hardware Side
Bluetooth hardware provides the connection that carries the processed and packetized
data. Although in that Bluetooth makes a wireless connection, its hardware is essentially
invisiblethe system requires a collection of circuits to transmit and receive the data
properly.
As with all radio systems, Bluetooth starts with a carrier wave and modulates it with data.
Unlike most common radio systems, however, Bluetooth does not use a single fixed carrier frequency but rather hops to different frequencies more than a thousand times each
second. As a serial transmission system, time is important to Bluetooth to sort out data
bits. Each Bluetooth device maintains a clock that helps it determine when each bit in its
serial stream appears. Bluetooth cleverly combines these necessary elements to make a
wireless communications network.
Clocks
Each Bluetooth device has its own internal clock that paces its communications. The clock
of each device operates independently at approximately the necessary rate.
For the Bluetooth signals to be effectively demodulated, the clocks of the master and slaves
must be synchronized. The master device sets the frequency for all the slaves with which it
communicates. The slaves determine the exact frequency of the clock from the packet data.
The preamble of each packet contains a predetermined pattern of several cycles, which the
slaves can use to achieve synchrony. The Bluetooth system does not alter the operation of
the clock of the slaves, however. Instead, it stores the difference between the master and
slave clocks and uses this difference value to maintain its lock on the master.
When another master takes control during a communication session, each slave readjusts
the stored difference value to maintain its precise frequency coordination.
Topology
Bluetooth is designed to handle a variety of connection types. The basic link is point-topoint, two devices communicating only with one another. In such a link, one device operates as the master and the other as the slave. In a piconet configuration, a single master can
communicate with up to seven active slaves (a total of eight devices intercommunicating).
In addition, other slaves may lock on to the masters signal and be ready to communicate
without sending out active signals. Such inactive slaves are said to be in a parked state.
375
376
The master in the piconet determines which devices can communicate (that is, which
slaves are active or parked). In addition, several piconets can be linked together into a scatternet, with the master of one piconet communicating to a master or slave in another.
Frequencies
Bluetooth operates at radio frequencies assigned to industrial, scientific, and medical
devices; a range termed the ISM band. This range of frequencies in the UHF (Ultra High
Frequency) band has been set aside throughout most of the world for unlicensed, lowpower electronic equipment. Near the top of the UHF range, the ISM band uses frequencies about twice that of the highest UHF television channel.
The exact frequencies available vary somewhat in North America, Europe, and Japan. In
addition, France and Spain differ from the rest of Europe (although both countries are
working on moving to the standards used throughout the rest of Europe).
Bluetooth uses channels one megahertz wide for its signals. Rather than operating on a
single channel, a Bluetooth system uses them all. It uses the channels one at a time but
switches between them to help minimize interference and fading. It can also help keep
communications secure. Only the devices participating in a piconet know which channel
they will hop to next.
In Europe (except France and Spain) and North America, the Bluetooth system can hop
between 79 different channels. Elsewhere, the choices are limited to 23 channels. The
available frequencies and number of channels available are summarized in the Table 11.5.
TA B L E 1 1 . 5
Frequencies and Channels Available to Bluetooth
Location
Frequency Range
Channels Available
North America
2.400 to 2.4835GHz
79
Europe (except
Spain and France)
2.400 to 2.4835GHz
79
Spain
2.445 to 2.475GHz
23
France
2.4465 to 2.4835GHz
23
Japan
2.471 to 2.497GHz
23
A given Bluetooth system does not operate on one frequency but rather uses them all,
hopping from one channel to another, up to 1,600 times per second. If a given asynchronous packet does not get through on one frequency due to interference (and is
therefore not acknowledged), the next hop will send out a duplicate packet at a different
frequency.
CHAPTER 11 PORTS
Bluetooth
Unfortunately, Bluetooth does not have the entire 2.4GHz band to itself. The IEEE
802.11 wireless-networking standard currently uses the same frequencies, and interference
between the two systems (where both are active) is inevitable. Although in the long term
IEEE 802.11 will migrate to the 5GHz frequency range, at present the only way to
entirely prevent interference between the two systems is to use one or the other, not both.
Power
The Bluetooth specification defines three classes of equipment based on transmitter
power. Class 1 devices are the most powerful and can transmit with up to 100 milliwatts of
output power. Class 3 devices transmit with less than 1 milliwatt. Table 11.6 lists the maximum and minimum output powers for each power class.
TA B L E 1 1 . 6
Powers Assigned to Bluetooth Power Classes
Power Class
Maximum Output Power
Minimum Output Power
100 mW
1 mW
2.5 mW
0.25 mW
1 mW
Not specified
As with any radio-based system, greater power increases the coverage area, so a Class 1
device will have greater range than a Class 3 device (about 10 times greater because radio
propagation follows an inverse-square law). On the downside, greater output power means
the need for greater input power, which directly translates into battery drain. That 100
mW of output power will require about 100 times the battery power as a 1-mW device.
Fortunately, even Class 1 devices are modest power consumers compared to other facets
of notebook computers. For example, the power needs of a Class 1 device are less than
one-tenth the demand of a typical display screen.
Modulation
Bluetooth uses Gaussian frequency shift keying (FSK)that is, the presence of a data bit
alters (or shifts) the frequency of the carrier wave. Bluetooth specifies the polarity of the
FSK modulation. It represents a binary one with a positive deviation of the carrier wave
and a binary zero with a negative deviation. The raw data rate is 1Mbps (one million symbols per second).
Because of how the digital code affects the frequency shift keying modulation, the information content of the modulation affects the deviation of the signal. The Bluetooth standard specifies that the minimum deviation should never be smaller than 115KHz. The
maximum deviation will be between 140 and 175KHz.
377
378
Components
Bluetooth architecture builds a system from three parts: a radio unit, a link control unit,
and a support unit that provides link management and the host terminal interface. These
are functional divisions, and all will be integrated into most handheld Bluetooth devices.
In your PC, all three will likely reside on a Bluetooth interface card that installs like any
other expansion board in a standard PCI slot.
The radio unit implements the hardware aspects of Bluetooth described earlier. It determines the power and coverage of the Bluetooth device, and its circuitry creates the carrier
wave (altering its frequency for each hop), modulates it, amplifies it, and radiates it
through an antenna. The ultra-high frequencies used by the Bluetooth system have a
short wavelength that allows the antenna to be integrated invisibly into the cases of many
mobile devices.
The link control unit is the mastermind of the Bluetooth system. It implements the various control and management protocols for setting up and maintaining the wireless connection. It searches out and identifies new devices wanting to join the piconet, tracks the
frequency hopping, and controls the operating state of the device.
The support unit provides the actual interface between the logic of the host device and
the Bluetooth connection. It adapts the signals of the host to match the Bluetooth system,
both electrically and logically. For example, in a PC-based Bluetooth interface card, the
support unit adapts the parallel bus signals of the PCI connection into the packetized serial form used by Bluetooth. It also checks data coming in from the wireless connection for
errors and requests retransmission when necessary.

The Bluetooth Special Interest Group promulgates the Bluetooth specifications. It also
facilitates the licensing of Bluetooth intellectual property. You can obtain the complete
specification from the SIG at www.bluetooth.com.
RS-232C Serial Ports

The day you win the vmega-lottery and instantly climb into wealth and social status, you
may be tempted to leave your old friends to belch and scratch while drinking someone
elses beer. But its hard to leave old friends behind, particularly when you need someone
to watch the house during your round-the-world cruise. So it is with the classic RS-232C
port. Its got so many bad habits its hard to talk about in polite company, but its just too
dang useful to forget about.
The serial port is truly the old codger of computer interfaces, a true child of the 60s.
An industry trade group, the Electronics Industry Association (EIA) hammered out the
CHAPTER 11 PORTS
official RS-232C specification in 1969, but the port had been in use for years at the time.
It found ready acceptance on the first personal computer because no other electronic connection for data equipment was so widely used. The ports survive today because some
folks still want to connect gear they bought in 1969 to their new computers.
Electrical Operation
RS-232C ports are asynchronous. They operate without a clock signal. But in order for
two devices to communicate they need at least a general idea of what rate to expect data.
Consequently, you must set the speed of each RS-232C port before you begin communicating, and the speeds of any two connected ports must match.
You have quite a wide variety to choose from. The serial ports in computers generally
operate at any speed in the odd-looking sequence that runs 150, 300, 600, 1200, 2400,
4800, 9600, 19,200, 38,400, 57,600, and 115,200 bits per second.
The RS-232C moves data one byte at a time. To suit its asynchronous nature, each byte
requires its own packing into a serial frame. In this form the typical serial port takes about
a dozen bits to move a bytea frame comprises two start bits, eight data bits, one parity
bit, and one stop bit to indicate the end of the frame. As a result, a serial port has overhead of about one-third of its potential peak data rate. A 9600 bit per second serial connection actually moves text at about 800 characters per second (6400bps).
A basic RS-232C connection needs only three connections: one for sending data, one for
receiving data, and a common ground. Most serial links also use hardware flow-control
signals. The most common serial port uses eight separate connections.
RS-232C port use single-ended signaling. Although this design simplifies the circuitry to
make the ports, it also limits the potential range of a connection. Long cable runs are apt
to pick up noise and blur high-data-rate signals. You can probably extend a 9600bps connection to a hundred feet or more. At a quarter mile, youll probably be down to 1200 or
300bps (slower than even cheap printers can type).
Because the RS-232C port originated in the data communications rather than computer
industry, some of its terminology is different from that used to describe other ports. For
example, the low and high logic states are termed space and mark in the RS-232C scheme.
Space is the absence of a bit, and mark is the presence of a bit. On the serial line, a space
is a positive voltage; a mark is a negative voltage.
In other words, when youre not sending data down a serial line, it has an overall positive
voltage on it. Data will appear as a serial of negative-going pulses. The original design of
the serial port specification called for the voltage to shift from a positive 12 volts to negative 12 volts. Because 12 volts is an uncommon potential in many computers, the serial
voltage often varies from positive 5 to negative 5 volts.
379
380
Connectors
The physical manifestation of a serial port is the connector that glowers on the rear panel
of your computer. It is where you plug your serial peripheral into your computer. And it
can be the root of all evilor so it will seem after a number of long evenings during
which you valiantly try to make your serial device work with your computer, only to have
text disappear like phantoms at sunrise. Again, the principal problem with serial ports is
the number of options they allow designers. Serial ports can use either of two styles of
connectors, each of which has two options in signal assignment. Worse, some manufacturers venture bravely in their own directions with the all-important flow-control signals.
Sorting out all these options is the most frustrating part of serial port configuration.
25-Pin
The basic serial port connector is called a 25-pin D-shell. It earns its name from having 25
connections arranged in two rows that are surrounded by a metal guide that roughly takes
the form of a letter D. The male variety of this connectorthe one that actually has pins
inside itis normally used on computers. Most, but hardly all, serial peripherals use the
female connector (the one with holes instead of pins) for their serial ports. Although both
serial and parallel ports use the same style 25-pin D-shell connectors, you can distinguish
serial ports from parallel ports because on most computers the latter use female connectors. Figure 11.8 shows the typical male serial port DB-25 connector that youll find on
the back of your computer.
FIGURE 11.8
The male DB-25 connector used by serial
ports on computers.
Although the serial connector allows for 25 discrete signals, only a few of them are ever
actually used. Serial systems may involve as few as three connections. At most, computer
serial ports use 10 different signals. Table 11.7 lists the names of these signals, their
mnemonics, and the pins to which they are assigned in the standard 25-pin serial
connector.
TA B L E 1 1 . 7
25-Pin Serial Connector Signal Assignments
Pin
Function
Mnemonic
Chassis ground
None
Transmit data
TXD
Receive data
RXD
CHAPTER 11 PORTS
Pin
Function
Mnemonic
Request to send
RTS
Clear to send
CTS
Data set ready
RTS
Signal ground
GND
Carrier detect
CD
20
Data terminal ready
DTR
22
Ring indicator
RI
Note that in the standard serial cable, the signal ground (which is the return line for the
data signals on pins 2 and 3) is separated from the chassis ground on pin 1. The chassis
ground pin is connected directly to the metal chassis or case of the equipment, much like
the extra prong of a three-wire AC power cable, and it provides the same protective function. It ensures that the case of the two devices linked by the serial cable are at the same
potential, which means you wont get a shock if you touch both at the same time. As wonderful as this connection sounds, it is often omitted from serial cables. On the other hand,
the signal ground is a necessary signal that the serial link cannot work without. You should
never connect the chassis ground to the signal ground.
Nine-Pin
If nothing else, using a 25-pin D-shell connector for a serial port is a waste of at least 15
pins. Most serial connections use fewer than the complete 10; some as few as four with
hardware handshaking, and three with software flow control. For the sake of standardization, the computer industry sacrificed the cost of the other unused pins for years until a
largeror smaller, depending on your point of viewproblem arose: space. A serial port
connector was too big to fit on the retaining brackets of expansion boards along with a
parallel connector. In that all the pins in the parallel connector had an assigned function,
the serial connector met its destiny and got miniaturized.
Moving to a nine-pin connector allowed engineers to put connectors for both a serial port
and a parallel port on the retaining bracket of a single expansion board. This was an
important concern because all ports in early computers were installed on expansion
boards. Computer makers could save the cost of an entire expansion board by putting two
ports on one card. Later, after most manufacturers moved to putting ports on computer
motherboards, the smaller port design persisted.
As with the 25-pin variety of serial connector, the nine-pin serial jack on the back of computers uses a male connector. Figure 11.9 shows the nine-pin male connector thats used
on some computers for serial ports.
381
382
FIGURE 11.9
The male DB-9 plug
used by AT-class serial
devices.
Nine-pin connectors necessarily have different pin assignments than 25-pin connectors.
Table 11.8 lists the signal assignments on the most common nine-pin implementation of
the RS-232C port.
TA B L E 1 1 . 8
Nine-Pin Serial Connector
Pin
Function
Mnemonic
Carrier detect
CD
Receive data
RXD
Transmit data
TXD
Data terminal ready
DTR
Signal Ground
GND
Data set ready
DSR
Request to send
RTS
Clear to send
CTS
Ring indicator
RI
Other than the rearrangement of signals, the nine-pin and 25-pin serial connectors are
essentially the same. All the signals behave identically, no matter the size of the connector
on which they appear.
Signals
Serial communications is an exchange of signals across the serial interface. These signals
involve not just data but also the flow-control signals that help keep the data flowing as
fast as possiblebut not too fast.
First, well look at the signals and their flow in the kind of communication system for
which the serial port was designedlinking a computer to a modem. Then well examine
how attaching a serial peripheral to a serial port complicates matters and what you can do
to make the connection work.
Definitions
As with space and mark, RS-232C ports use other odd terminology. Serial terminology
assumes that each end of a connection has a different type of equipment attached to it.
CHAPTER 11 PORTS
One end has a data terminal connected to it. In the old days when the serial port was
developed, a terminal was exactly thata keyboard and a screen that translated typing
into serial signals. Today, a terminal is usually a computer. For reasons known but to those
who revel in rolling their tongues across excess syllables, the term Data Terminal
Equipment is often substituted. To make matters even more complex, many discussions talk
about DTE devices, which means exactly the same thing as data terminals.
The other end of the connection has a data set, which corresponds to a modem. Often
engineers substitute the more formal name Data Communication Equipment or talk about
DCE devices.
The distinction between data terminals and data sets (or DTE and DCE devices) is
important. Serial communications were originally designed to take place between one
DTE and one DCE, and the signals used by the system are defined in those terms.
Moreover, the types of RS-232 serial devices you wish to connect determines the kind of
cable you must use. First, however, lets look at the signals; then well consider what kind
of cable you need to carry them.
Transmit Data
The serial data leaving the RS-232 port travels on what is called the Transmit Data line,
which is usually abbreviated TXD. The signal on it comprises the long sequence of pulses
generated by the UART in the serial port. The data terminal sends out this signal, and the
data set listens to it.
Receive Data
The stream of bits going the other directionthat is, coming in from a distant serial
portgoes through the Receive Data line (usually abbreviated RXD) to reach the input of
the serial ports UART. The data terminal listens on this line for the data signal coming
from the data set.
Data Terminal Read
When the data terminal is able to participate in communicationsthat is, it is turned on
and in the proper operating modeit signals its readiness to the data set by applying a
positive voltage to the Data Terminal Ready line, which is abbreviated DTR.
Data Set Ready
When the data set is able to receive datathat is, it is turned on and in the proper operating modeit signals its readiness by applying a positive voltage to the Data Set Ready
line, which is abbreviated DSR. Because serial communications must be two way, the
data terminal will not send out a data signal unless it sees the DSR signal coming from
the data set.
383
384
Request to Send
When the data terminal is on and capable of receiving transmissions, it puts a positive
voltage on its Request to Send line, usually abbreviated RTS. This signal tells the data set
that it can send data to the data terminal. The absence of an RTS signal across the serial
connection will prevent the data set from sending out serial data. This allows the data terminal to control the flow of the data set to it.
Clear to Send
The data set, too, needs to control the signal flow from the data terminal. The signal it
uses is called Clear to Send, which is abbreviated CTS. The presence of the CTS signal in
effect tells the data terminal that the coast is clear and the data terminal can blast data
down the line. The absence of a CTS signal across the serial connection will prevent the
data terminal from sending out serial data.
Carrier Detect
The serial interface standard shows its roots in the communication industry with the
Carrier Detect signal, which is usually abbreviated CD. This signal gives a modem, the typical data set, a means of signaling to the data terminal that it has made a connection with a
distant modem. The signal says that the modem or data set has detected the carrier wave
of another modem on the telephone line. In effect, the carrier detect signal gets sent to
the data terminal to tell it that communications are possible. In some systems, the data
terminal must see the carrier detect signal before it will engage in data exchange. Other
systems simply ignore this signal.
Ring Indicator
Sometimes a data terminal has to get ready to communicate even before the flow of
information begins. For example, you might want to switch your communications program into answer mode so that it can deal with an incoming call. The designers of the
serial port provided such an early warning in the Ring Indicator signal, which is usually
abbreviated RI. When a modem serving as a data set detects ringing voltagethe lowfrequency, high-voltage signal that makes telephone bells ringon the telephone line to
which it is connected, it activates the RI signal, which alerts the data terminal to whats
going on. Although useful in setting up modem communications, you can regard the
ring indicator signal as optional because its absence usually will not prevent the flow of
serial data.
Signal Ground
All the signals used in a serial port need a return path. The signal ground provides this
return path. The single ground signal is the common return for all other signals on the
serial interface. Its absence will prevent serial communications entirely.
CHAPTER 11 PORTS
Flow Control
Serial ports can use both hardware and software flow control. Hardware flow control
involves the use of special control lines that can be (but dont have to be) part of a serial
connection. Your computer signals whether it is ready to accept more data by sending a
signal down the appropriate wire. Software flow control involves the exchange of characters
between computer and serial peripherals. One character tells the computer your peripheral is ready, and another warns that it cant deal with more data. Both hardware and software flow control take more than one form. As a default, computer serial ports use
hardware flow control (or hardware handshaking). Most serial peripherals do, too. In general, hardware flow control uses the Carrier Detect, Clear to Send, and Data Set Ready
signals.
Software flow control requires your serial peripheral and computer to exchange characters
or tokens to indicate whether they should transfer data. The serial peripheral normally
sends out one character to indicate it can accept data and a different character to indicate
that it is busy and cannot accommodate more. Two pairs of characters are often used:
XON/XOFF and ETX/ACK.
Cables
The design of the standard RS-232 serial interface anticipates that you will connect a data
terminal to a data set. When you do, all the connections at one end of the cable that link
them are carried through to the other endpin for pin, connection for connection. The
definitions of the signals at each end of the cable are the same, and the function and direction of travel (whether from data terminal to data set or the other way around) of each are
well defined. Each signal goes straight through from one end to the other. Even the connectors are the same at either end. Consequently, a serial cable should be relatively easy to
fabricate.
In the real world, nothing is so easy. Serial cables are usually much less complicated or
much more complicated than this simple design. Unfortunately, if you plan to use a serial
connection for a printer or plotter, you have to suffer through the more complex design.
Straight-Through Cables
Serial cables are often simpler than pin-for-pin connections from one end to the other
because no serial link uses all 25 connector pins. Even with the complex handshaking
schemes used by modems, only nine signals need to travel from the data terminal to the
data set, computer to modem. (For signaling purposes, the two grounds are redundant
most serial cables do not connect the chassis ground.) Consequently, you need only make
these 10 connections to make virtually any data terminaltodata set link work. Assuming
you have a 25-pin D-shell connector at either end of your serial cable, the essential pins
that must be connected are 2 through 8, 20, and 22 on a 25-pin D-shell connector. This is
385
386
usually called a nine-wire serial cable because the connection to pin 7 uses the shield of the
cable rather than a wire inside. With nine-pin connectors at either end of your serial
cable, all nine connections are essential.
Not all systems use all the handshaking signals, so you can often get away with fewer connections in a serial cable. The minimal case is a system that uses software handshaking
only. In that case, you need only three connections: Transmit Data, Receive Data, and the
signal ground. In other words, you need only connect pins 2, 3, and 7 on a 25-pin connector or pins 2, 3, and 5 on a nine-pin serial connector (providing, of course, you have the
same size connector at each end of the cable).
Although cables with an intermediary number of connections are often available, they are
not sufficiently less expensive than the nine-wire cable to justify the risk and lack of versatility. Therefore, you should limit your choices to a nine-wire cable for systems that use
hardware handshaking or three-wire cables for those that youre certain use only software
flow control.
Manufacturers use a wide range of cable types for serial connections. For relatively low
data rates and reasonable lengths of serial connections, you can get away with just about
anything, including twisted-pair telephone wire. To ensure against interference, you
should use shielded cable, which wraps a wire braid or aluminum-coated plastic film
around inner conductors to prevent signals leaking out or in. The shield of the cable
should be connected to the signal ground. (Ideally, the signal ground should have its own
wire, and the shield should be connected to the chassis ground, but most folks just dont
bother.)
Adapter Cables
If you need a cable with a 25-pin connector at one end and a nine-pin connector at the
other, you cannot use a straight-through design, even when you want to link a data terminal to a data set. The different signal layouts of the two styles of connectors are incompatible. After all, you cant possibly link pin 22 on a 25-pin connector to a nonexistent pin 22
on a nine-pin connector.
This problem is not uncommon. Even though the nine-pin connector has become a de
facto standard on computers, most other equipment, including serial plotters, printers, and
modems, has stuck with the 25-pin standard. To get from one connector type to another,
you need an adapter. The adapter can take the form of a small assembly with a connector
on each end of an adapter cable, typically from six inches to six feet long.
Crossover Cables
As long as you want to connect a computer serial port that functions to a modem, you
should have no problem with serial communications. You will be connecting a data terminal to a data set, exactly what engineers designed the serial systems for. Simply sling a
CHAPTER 11 PORTS
cable with enough conductors to handle all the vital signals between the computer and
modem andvoilserial communications without a hitch. Try it, and youre likely to
wonder why so many people complain about the capricious nature of serial connections.
When you want to connect a plotter or printer to a computer through a serial port, however, you will immediately encounter a problem. The architects of the RS-232 serial system decided that both computers and the devices are data terminal (DTE) devices. The
designations actually made sense, at least at that time. You were just as likely to connect a
serial printer (such as a teletype) to a modem as you were a computer terminal. There was
no concern about connecting a printer to a computer because computers didnt even exist
back then.
When you connect a plotter or printer and your computeror any two DTE devices
together with an ordinary serial cable, you will not have a communication system at all.
Neither machine will know that the other one is even there. Each one will listen on the
serial port signal line that the other is listening to, and each one will talk on the line that
the other talks on. One device wont hear a bit of what the other is saying.
The obvious solution to the problem is to switch some wires around. Move the Transmit
Data wire from the computer to where the Receive Data wire goes on the plotter or
printer. Route the computers Receive Data wire to the Transmit Data wire of the plotter
or printer. A simple crossover cable does exactly that, switching the Transmit and Receive
signals at one end of the connection.
Many of the devices that you plug into a computer are classed as DTE (or data terminals),
just like the computer. All these will require a crossover cable. Table 11.9 lists many of the
devices you might connect to your computer and whether they function as data terminals
(DTE) or data sets (DCE).
TA B L E 1 1 . 9
Common Serial Device Types
Peripheral
Device Type
Cable Needed to Connect to PC
PC
DTE
Crossover
Modem
DCE
Straight-through
Mouse
DCE
Straight-through
Trackball
DCE
Straight-through
Digitizer
DCE
Straight-through
Scanner
DCE
Straight-through
Serial printer
DTE
Crossover
Serial plotter
DTE
Crossover
387
388
Note that some people call crossover cables null modem cables. This is not correct. A null
modem is a single connector used in testing serial ports. It connects the Transmit Data
line to the Receive Data line of a serial port as well as crosses the handshaking connections within the connector as described earlier. Correctly speaking, a null modem cable is
equipped with this kind of wiring at both ends. It will force both serial ports constantly on
and prevent any hardware flow control from functioning at all. Although such a cable can
be useful, it is not the same as a crossover cable. Substituting one for the other will lead to
some unpleasant surprisessuch as text dropping from sight from within documents as
mysteriously and irrecoverably as D. B. Cooper.
UARTs
A serial port has two jobs to perform. It must repackage parallel data into serial form, and
it must send power down a long wire with another circuit at the end, which is called driving the line.
Turning parallel data into serial is such a common electrical function that engineers created special integrated circuits that do exactly that. Called Universal Asynchronous
Receiver/Transmitter chips, or UARTs, these chips gulp down a byte or more of data and
stream it out a bit at a time. In addition, they add all the other accouterments of the serial
signalthe start, parity, and stop bits. Because every practical serial connection is bidirectional, the UART works both ways, sending and receiving, as its name implies.
Because the UART does all the work of serializing your computers data signals, its operation is one of the limits on the performance of serial data exchanges. Computers have used
three different generations of UARTs, each of which imposes its own constraints. Early
computers used 8250 UARTs, and later machines shifted to the higher-speed 16450
UART. Both of these chips had one-byte buffers that were unable to keep up with normal
communications when multitasking software came into use. The replacement was the
16550A UART (commonly listed as 16550AF and 16550AFN, with the last initials indicating the package and temperature rating of the chip), which has a 16-byte First In, First
Out (FIFO) buffer.
To maintain backward compatibility with the 16450, the 16550 ignores its internal buffer
until it is specifically switched on. Most communications programs activate the buffer
automatically. Physically, the 16550 and 16450 will fit and operate in the same sockets, so
you can easily upgrade the older chip to the newer one.
Modern computers do not use separate UARTs. However, all the UART circuitryusually exactly equivalent to that of the 16550Ais built into the circuitry of nearly all
chipsets. To your software, the chipset acts exactly as if your serial ports were on separate
expansion boards, just as they were in the first personal computers.
CHAPTER 11 PORTS
Register Function
The register at the base address assigned to each serial port is used for data communications. Bytes are moved to and from the UART using the microprocessors OUT and IN
instructions. The next six addresses are used by other serial port registers. They are, in
order, the Interrupt Enable register, the Interrupt Identification register, the Line Control
register, the Modem Control register, the Line Status register, and the Modem Status register. Another register, called the Divisor Latch, shares the base address used by the
Transmit and Receive registers and the next higher register used by the Interrupt Enable
register. It is accessed by toggling a setting in the Line Control register.
This latch stores the divisor that determines the operating speed of the serial port.
Whatever value is loaded into the latch is multiplied by 16. The resulting product is used
to divide down the clock signal supplied to the UART chip to determine the bit rate.
Because of the factor of 16 multiplication, the highest speed the serial port can operate at
is limited to 1/16th the supplied clock (which is 1.8432MHz). Setting the latch value to its
minimum, 1, results in a bit rate of 115,200.
Registers not only store the values used by the UART chip but also are used to report
back to your system how the serial conversation is progressing. For example, the line status register indicates whether a character that has been loaded to be transmitted has actually been sent. It also indicates when a new character has been received.
Logical Interface
Your computer controls the serial port UART through a set of seven registers built in to
the chip. Although your programs could send data and commands to the UART (and,
through it, to your serial device) by using the hardware address of the registers on the
chip, this strategy has disadvantages. It requires the designers of systems to allocate once
and forever the system resources used by the serial port. The designers of the original
IBM computer were loathe to make such a permanent commitment. Instead they devised
a more flexible system that allows your software to access ports by name. In addition, they
worked out a way that port names would be assigned properly and automatically, even if
you didnt install ports in some predetermined order.
Port Names
The number of serial ports you can use in a computer varies with the operating system.
Originally, personal computers could only use two, but in 1987 the designers of DOS
expanded the possible port repertory to include COM3 and COM4. Under Windows 3.1,
up to nine serial ports could be used. Windows versions since Windows 95 extend serial
port support to 128.
389
390
Without special drivers, Windows recognizes four serial ports. When Windows checks
your system hardware each time it boots up, it check the ranges of addresses normally
used by the UART chips for serial ports. Each UART has seven registers that control it,
and these are usually identified by a base address, the input/output port used by the first of
these registers. The usual computer design allows for four base addresses for a serial port.
Current Windows versions search the nominal base addresses for serial ports and assign
their serial port drivers to those that are active. Devices out of the normal rangeincluding the serial ports built in to internal modemsrequire their own drivers to match their
hardware.
Interrupts
Serial ports normally operate as interrupt-driven devices. That is, when they must perform an action immediately, they send a special signal called an interrupt to your computers microprocessor. In the traditional computer design, only two interrupts are
available for serial ports, as listed in Table 11.10.
TA B L E 1 1 . 1 0
Default Settings for DOS and Windows Serial Ports
Port Name
Base Address
Interrupt
COM1
03F8(hex)
COM2
02F8(hex)
COM3
03E8(hex)
COM4
02E8(hex)
Systems with more than two serial ports (or oddly assigned interrupts) have to share two
interrupts between these serial portsone port is often assigned to two ports. This
sometimes results in problems, particularly when a mouse is connected to a serial port.
However, because all new computers have a dedicated mouse port, this problem no
longer occurs.
Parallel Ports
The defining characteristic of the parallel port design is implicit in its name. The port is
parallel because it conducts its signals through eight separate wiresone for each bit of
a byte of datathat are enclosed together in a single cable. The signal wires literally run
in parallel from your computer to their destinationor at least they did. Better cables
twist the physical wires together but keep their signals straight (and parallel).
In theory, having eight wires means you can move data eight times as fast through a parallel connection than through a single wire. All else being equal, simple math would make
CHAPTER 11 PORTS
Parallel Ports
this statement true. Although a number of practical concerns make such extrapolations
impossible, throughout its life, the parallel port has been known for its speed. It beat its
original competitor, the RS-232 port, hands down, outrunning the serial ports 115.2Kbps
maximum by factors from two to five, even in early computers. The latest incarnations of
parallel technology put the data rate through the parallel connection at over 100 times
faster than the basic serial port rate.
In simple installations (for example, when used for its original purpose of linking a printer
to your computer), the parallel port is a model of installation elegance. Just plug in your
printer, and the odds are it will work flawlesslyor that whatever flaws appear wont have
anything to do with the interconnection.
Despite such rave reviews, parallel ports are not trouble-free. All parallel ports are not
created equal. A number of different designs have appeared during the brief history of the
computer. Although new computers usually incorporate the latest, most versatile, and
highest speed of these, some manufacturers skimp. Even when you buy a brand-new computer, you may end up with a simple printer port that steps back to the first generation of
computer design.
A suitable place to being this saga is to sort out this confusion of parallel port designs by
tracing the parallel ports origins. As it turns out, the history of the parallel port is a long
one, older than even the personal computer, although the name and our story begin with
its introduction.
History
Necessity isnt just the mother of invention. It also spawned the parallel port. As with
most great inventions, the parallel port arose with a problem that needed to be solved.
When IBM developed its first computer, its engineers looked for a simplified way to link
to a printer, something without the hassles and manufacturing costs of a serial port. The
simple parallel connection, already used in a similar form by some printers, was an elegant
solution. Consequently, IBMs slightly modified version became standard equipment on
the first computers. Because of its intended purpose, it quickly gained the printer port
epithet. Not only were printers easy to attach to a parallel port, they were the only thing
that you could connect to these first ports at the time.
In truth, the contribution of computer-makers to the first parallel port was minimal. They
added a new connector that better fit the space available on the computer. The actual port
design was already being used on computer printers at the time. Originally created by
printer-maker Centronics Data Computer Corporation and used by printers throughout
the 1960s and 1970s, the connection was electrically simple, even elegant. It took little
circuitry to add to a printer or computer, even in the days when designers had to use discrete components instead of custom-designed circuits. A few old-timers still cling to history and call the parallel port a Centronics port.
391
392
The computer parallel port is not identical to the exact Centronics design, however. In
adapting it to the computer, IBM substituted a smaller connector. The large jack used by
the Centronics design had 36 pins and was too large to put where IBM wanted itsharing
a card-retaining bracket with a video connector on the computers first Monochrome
Display Adapter. In addition, IBM added two new signals to give the computer more control over the printer and adjusted the timing of the signals traveling through the interface.
All that said, most Centronics-style printers worked just fine with the original computer.
At the time, the computer parallel port had few higher aspirations. It did its job, and did it
well, moving data in one direction (from computer to printer) at rates from 50 to 150Kbps.
The computer parallel port, or subtle variations of it, became ubiquitous if not universal.
Any printer worth connecting to a computer used a parallel port (or so it seemed).
In 1987, however, IBMs engineers pushed the parallel port in a new direction. The
motive behind the change was surprisingnot a desire to improve communication but
rather a band-aid solution for a temporary problem (for which it was hardly ever used).
The company decided to adopt the 3.5-inch floppy disk drives for its new line of PS/2
computers at a time when all the worlds computer data was mired on 5.25-inch diskettes.
The new computers made no provision for building in the bigger drives. Instead, IBM
believed that the entire world would instantly switch over to the new disk format. People
would need to transfer their data once and only once to the new disk format. To make the
transfer possible, the company released its Data Migration Facility, a fancy name for a
cable and a couple disks. You used the cable to connect your old computer to your new
PS/2 and software on the disks to move files through the parallel port from the old
machine and disks to the new.
Implicit in this design is the ability of the PS/2 parallel port to receive data as well as send
it out, as to a printer. The engineers tinkered with the port design and made it work both
ways, creating a bidirectional parallel port. Because of the designs intimate connection with
the PS/2, it is sometimes termed the PS/2 parallel port.
The Data Migration Facility proved to be an inspirational idea despite its singular shortcoming of working in only one direction. As notebook computers became popular, they
also needed a convenient means to move files between machines. The makers of file transfer programs such as Brooklyn Bridge and LapLink knew a good connection when they
saw it. By tinkering with parallel port signals, they discovered that they could make any
parallel port operate in both directions and move data to and from computers.
The key to making bidirectional transfers on the old-fashioned one-way ports was to
redefine signals. They redirected tours of the signals in the parallel connector that originally had been designed to convey status information back from the printer to your
computer. These signals already went in the correct direction. All that the software
CHAPTER 11 PORTS
Parallel Ports
mavens did was to take direct control of the port and monitor the signals under their
new definitions. Of course, four signals cant make a byte. They were limited to shifting
four bits through the port in the backward direction. Because four bits make a nibble,
the new parallel port operating mode soon earned the name nibble mode.
This four-bits-at-a-time scheme had greater implications than just a new name. Half as
many bits also means half the speed. Nibble mode operates at about half the normal parallel port ratestill faster than single-line serial ports but not full parallel speed.
If both sides of a parallel connection had bidirectional ports, however, data transfers ran at
full speed both ways. Unfortunately, as manufacturers began adapting higher-performance
peripherals to use the parallel port, what once was fast performance became agonizingly
slow. Although the bidirectional parallel port more than met the modest data transfer
needs of printers and floppy disk drives, it lagged behind other means of connecting hard
disks and networks to computers.
Engineers at network adapter maker Xircom Incorporated decided to do something about
parallel performance and banded together with notebook computer maker Zenith Data
Systems to find a better solution. Along the way, they added Intel Corporation and
formed a triumvirate called Enhanced Parallel Port Partnership. They explored two ways of
increasing the data throughput of a parallel port. They streamlined the logical interface so
that your computer would need less overhead to move each byte through the port. In
addition, they tightly defined the timing of the signals passing through the port, minimizing wasted time and helping ensure against timing errors. They called the result of their
efforts the Enhanced Parallel Port (EPP).
On August 10, 1991, the organization released its first description of what it thought the
next generation of parallel ports should be and do. The organization continued to work
on a specification until March 1992, when it submitted Release 1.7 to the Institute of
Electrical and Electronic Engineers (IEEE) to be considered as an industry standard.
Although the EPP version of the parallel port could increase its performance by nearly
tenfold, that wasnt enough to please everybody. The speed potential made some engineers see the old parallel port as an alternative to more complex expansion buses such as
the SCSI system. With this idea in mind, Hewlett-Packard joined with Microsoft to make
the parallel port into a universal expansion standard called the extended capabilities port
(ECP). In November 1992, the two companies released the first version of the ECP specification, aimed at computers that use the ISA expansion bus. This first implementation
added two new transfer modes to the EPP designa fast two-way communication mode
between a computer and its peripherals, and another two-way mode with performance
further enhanced by simple integral data compressionand defined a complete software
control system.
393
394
The heart of the ECP innovation was a protocol for exchanging data across a high-speed
parallel connection. The devices at the two ends of each ECP transfer negotiate the
speed and mode of data movement. Your computer can query any ECP device to determine its capabilities. For example, your computer can determine what language your
printer speaks and set up the proper printer driver accordingly. In addition, ECP devices
tell your computer the speed at which they can accept transmissions and the format of the
data they understand. To ensure the quality of all transmissions, the ECP specification
included error detection and device handshaking. It also allowed the use of data compression to further speed transfers.
On March 30, 1994, the IEEE Standards Board approved its parallel port standard, IEEE1284-1994. The standard included all the basic modes and parallel port designs, including
both ECP and EPP. It was submitted to the American National Standards Institute and
approved as a standard on September 2, 1994.
The IEEE 1284 standard marked a watershed in parallel port design and nomenclature.
The standard defined (or redefined) all aspects of the parallel connection, from the software interface in your computer to the control electronics in your printer. It divided the
world of parallel ports in two: IEEE 1284-compatible devices, which are those that will work
with the new interface, which in turn includes just about every parallel port and device
ever made; and IEEE 1284-compliant devices, which are those that understand and use the
new standard. This distinction is essentially between pre- and post-standardization ports.
You can consider IEEE 1284-compatible ports to be old technology and IEEE 1284-compliant ports to be new technology.
Before IEEE 1284, parallel ports could be divided into four types: standard parallel ports,
bidirectional parallel ports (also known as PS/2 parallel ports), enhanced parallel ports, and
extended capabilities ports. The IEEE specification redefined the differences in ports,
classifying them by the transfer mode they use. Although the terms are not exactly the
same, you can consider a standard parallel port one that is able to use only nibble-mode
transfers. A PS/2 or bidirectional parallel port from the old days is one that can also make
use of byte-mode transfers. EPP and ECP ports are those that use EPP and ECP modes,
as described by the IEEE 1284 specification.
EPP and ECP remain standards separate from IEEE 1284, although they have been
revised to depend on it. Both EPP and ECP rely on their respective modes as defined in
the IEEE specification for their physical connections and electrical signaling. In other
words, IEEE 1284 describes the physical and electrical characteristics of a variety of
parallel ports. The other standards describe how the ports operate and link to your
applications.
CHAPTER 11 PORTS
Parallel Ports
Connectors
The best place to begin any discussion of the function and operation of the parallel port
is the connector. After all, the connector is what puts the port to work. It is the physical
manifestation of the parallel port, the one part of the interface and standard you can
actually touch or hold in your hand. It is the only part of the interface that most people
will ever have to deal with. Once you know the ins and outs of parallel connectors, youll
be able to plug in the vast majority of computer printers and the myriad of other things
that now suck signals from what was once the printers port.
Unfortunately, as with the variety of operating modes, the parallel port connector itself is
not a single thing.
Parallel ports use three different connectors, called A, B, and C.
The A connector
The A connector appears only on computers as the output of a parallel port. Technically,
it is described as a female 25-pin D-shell connector. Engineers chose this particular connector pragmaticallyit was readily available and was the smallest connector that could
handle the signals required in a full parallel connection. After it became ubiquitous, the
IEEE adopted it as its 1284-A connector. Figure 11.10 shows a conceptual view of the A
connector.
FIGURE 11.10
The IEEE-1284 A
connector, a female
25-pin D-shell jack.
Of the 25 contacts on this parallel port connector, 17 are assigned individual signals for
data transfer and control. The remaining eight serve as ground returns. Under the IEEE
1284 specification, the definition of each signal on each pin is dependent on the operating
mode of the port. Only the definitions change; the physical wiring inside your computer
and inside the cables does not changeif it did, shifting modes would be far from trivial.
The altered definitions change the protocol, which is the signal handshaking that mediates
each transfer.
A single physical connector on the back of your computer can operate in any of these five
modes, and the signal definitions and their operation will change accordingly. Table 11.11
lists these five modes and their signal assignments.
395
396
TA B L E 1 1 . 1 1
IEEE 1284-A Connector Signal Assignments in All Modes
Pin
Compatibility
Mode
Nibble
Mode
Byte Mode
EPP
Mode
ECP
Mode
nStrobe
HostClk
HostClk
nWrite
HostClk
Data 1
Data 1
Data 1
AD1
Data 1
Data 2
Data 2
Data 2
AD2
Data 2
Data 3
Data 3
Data 3
AD3
Data 3
Data 4
Data 4
Data 4
AD4
Data 4
Data 5
Data 5
Data 5
AD5
Data 5
Data 6
Data 6
Data 6
AD6
Data 6
Data 7
Data 7
Data 7
AD7
Data 7
Data 8
Data 8
Data 8
AD8
Data8
10
nAck
PtrClk
PtrClk
Intr
PeriphClk
11
Busy
PtrBusy
PtrBusy
nWait
PeriphAck
12
PError
AckDataReq
AckDataReq
User defined 1
nAckReverse
13
Select
Xflag
Xflag
User defined 3
Xflag
14
nAutoFd
HostBusy
HostBusy
nDStrb
HostAck
15
nFault
nDataAvail
nDataAvail
User defined 2
nPeriphRequest
16
nInit
nInit
nInt
nInt
nReverseRequest
17
nSelectIn
1284 Active
1284 Active
nAStrb
1284 Active
18
Pin 1 (nStrobe)
ground return
19
Pins 2 and 3
(Data 1 and 2)
ground return
20
Pins 4 and 5
(Data 3 and 4)
ground return
21
Pins 6 and 7
(Data 5 and 6)
ground return
22
Pins 8 and 9
(Data 7 and 8)
ground return
CHAPTER 11 PORTS
Parallel Ports
Pin
Compatibility
Mode
23
Pins 11 and 15
ground return
24
Pins 10, 12, and 13

ground return
25
Pins 14, 16, and 17

ground return
Nibble
Mode
Byte Mode
EPP
Mode
ECP
Mode
Along with standardized signal assignments, IEEE 1284 also gives us a standard nomenclature for describing the signals. In Table 11.11, as well as all following tables that refer
to the standard, signal names prefaced with a lowercase n indicate that the signal goes
negative when active (that is, the absence of a voltage means the signal is present).
Mode changes are negotiated between your computer and the printer or other peripheral
connected to the parallel port. Consequently, both ends of the connection switch modes
together so that the signal assignments remain consistent at both ends of the connection.
For example, if you connect an older printer that only understands compatibility mode,
your computer cannot negotiate any other operating mode with the printer. It will not
activate its EPP or ECP mode, so your printer will never get signals it cannot understand.
This negotiation of the mode ensures backward compatibility among parallel devices.
The B Connector
The parallel port input to printers is quite a different connector from that on your computer. In fact, the design predates personal computers, having first been used by a printer
company, Centronics, which gave the connector and parallel port its alternate name (now
falling into disuse). The design is a 36-pin ribbon connector (the contacts take the form of
thin metal ribbons) in a D-shell. Figure 11.11 shows this connector as a jack that would
appear on the back of a printer.
FIGURE 11.11
The IEEE-1284 B
connector, a 36-pin
ribbon jack.
The assignment of signals to the individual pins of this connector has gone through three
stages. The first standard was set by Centronics for its printers. In 1981, IBM altered this
design somewhat by redefining several of the connections. Finally, in 1994, the IEEE
published its standard assignments, which (like those of the A-connector) vary with
operating mode.
397
398
The C Connector
Given a chance to start over with a clean slate and no installed base, engineers would
hardly come up with the confusion of two different connectors with an assortment of different, sometimes-compatible operating modes. The IEEE saw the creation of the 1284
standard as such an opportunity, one they were happy to exploit. To eliminate the confusion of two connectors and the intrinsic need for adapters to move between them, they
took the logical step: They created a third connector, IEEE 1284-C.
For the most part, the C connector is just the B connector with some of the air let out.
Figure 11.12 shows a jack that youd find on the back of equipment using C connectors.
FIGURE 11.12
Conceptual view of
the 1284-C parallel
port connector.
Adapters
The standard printer cable for computers is an adapter cable. It rearranges the signals of
the A connector to the scheme of the B connector. Ever since the introduction of the first
computer, you needed this sort of cable just to make your printer work. Over the years
they have become plentiful and cheap.
To cut costs, many makers of adapter cables group all the grounds together as a single
common line so that you need only 18 instead of 25 conductors in the connecting cable.
Cheap adapters, which do not meet the IEEE 1284 standard, use this approach.
A modern printer cable should contain a full 25 connections with the ground signals
divided up among separate pins. A true IEEE 1284 printer cable is equipped with an A connector on one end and a B connector on the other, with the full complement of connections in between.
As new peripherals with the 1284-C connector become available, youll need to plug them
into your computer. To attach your existing computer to a printer or other device using
the C connector, youll need an adapter cable to convert the A connector layout to the C
connector design. On the other hand, if your next computer or parallel adapter uses the C
connector and you plan to stick with your old printer, youll need another variety of
adapterone that translates the C connector layout to that of the B connector.
Cable
The high-speed modes of modern parallel ports make them finicky. When your parallel
port operates in EPP or ECP mode, cable quality becomes critical, even for short runs.
Signaling speed across one of these interfaces can be in the megahertz range. The
CHAPTER 11 PORTS
Parallel Ports
frequencies far exceed the reliable limits of even short runs of the dubious low-cost
printer cables. Consequently, the IEEE 1284 specification precisely details a special cable
for high-speed operation. Figure 11.13 offers a conceptual view of the construction of this
special parallel data cable.
FIGURE 11.13
IEEE 1284 cable
construction details.
Jacket
Wire braid
Twisted wire pairs
Optional filler
Aluminum/polyester foil film
Optional release paper
Unlike standard parallel wiring, the data lines in IEEE 1284 cables must be doubleshielded to prevent interference from affecting the signals. Each signal wire must be
twisted with its ground return. Even though the various standard connectors do not provide separate pins for each of these grounds, the ground wires must be present and run
the full length of the cable.
The differences between old-fashioned printer cables and those that conform to the
IEEE 1284 standard are substantial. Although you can plug in a printer with either a
printer or IEEE 1284compliant cable, devices that exploit the high-speed potentials of
the EPP or ECP designs may not operate properly with a noncompliant cable. Often,
even when a printer fails to operate properly, the cable may be at fault. Substituting a
truly IEEE 1284compliant cable will bring reluctant connections to life.
Electrical Operation
In each of its five modes, the IEEE 1284 parallel port operates as if it were some kind of
completely different electronic creation. When in compatibility mode, the IEEE 1284
port closely parallels the operation of the plain-vanilla printer port of bygone days. It
allows data to travel in one direction only, from computer to printer. Nibble mode gives
your printer (or more likely, another peripheral) a voice and allows it to talk back to your
computer. In nibble mode, data can move in either of two directions, although asymmetrically. Information flows faster to your printer than it does on the return trip. Byte mode
makes the journey fully symmetrical.
With the shift to EPP mode, the parallel port becomes a true expansion bus. A new way
of linking to your computers bus gives it increased bidirectional speed. Many systems can
run their parallel ports 10 times faster in EPP mode than in compatibility, nibble, or byte
mode. ECP mode takes the final step, giving control in addition to speed. ECP can do
just about anything any other expansion interface (including SCSI) can do.
399
400
Because of these significant differences, the best way to get to know the parallel port is by
considering each separately as if it were an interface unto itself. Our examination will follow from simple to complex, which also mirrors the history of the parallel port.
Note that IEEE 1284 deals only with the signals traveling through the connections of the
parallel interface. It establishes the relationship between signals and their timing. It concerns itself neither with the data that is actually transferred, with the command protocols
encoded in the data, nor with the control system that produces the signals. In other words,
IEEE 1284 provides an environment under which other standards such as EPP and ECP
operate. That is, ECP and EPP modes are not the ECP and EPP standards, although those
modes are meant to be used by the parallel ports operating under respective standards.
Compatibility Mode
The least common denominator among parallel ports is the classic design that IBM introduced with its first computer. It was conceived strictly as a interface for the one-way transfer of information. Your computer sends data to your printer and expects nothing in
return. After all, a printer neither stores information nor creates information on its own.
In conception, this port is like a conveyor that unloads ore from a bulk freighter or rolls
coal out of a mine. The raw material travels in one direction. The conveyor mindlessly
pushes out stuff and more stuff, perhaps creating a dangerously precarious pile, until its
operator wakes up and switches it off before the pile gets much higher than his waist.
If your printer had unlimited speed or an unlimited internal buffer, such a one-way design
would work. But like the coal yard, your printer has a limited capacity and may not be
able to cart off data as fast as the interface shoves it out. The printer needs some way of
sending a signal to your computer to warn about a potential data overflow. In electronic
terms, the interface needs feedback of some kindit needs to get information from the
printer that your computer can use to control the data flow.
To provide the necessary feedback for controlling the data flow, the original Centronics
port design and IBMs adaptation of it both included several control signals. These were
designed to allow your computer to monitor how things are going with your printer
whether data is piling up, whether the printer has sufficient paper or ribbon, and even
whether the printer is turned on. Your computer can use this information to moderate the
outflowing gush of data or to post a message warning you that something is wrong with
your printer. In addition, the original parallel port included control signals sent from your
computer to the printer to tell it when the computer wanted to transfer data and to tell the
printer to reset itself. The IEEE 1284 standard carries all these functions into compatibility mode.
Strictly speaking, then, even this basic parallel port is not truly a one-way connection,
although its feedback provisions were designed strictly for monitoring rather than data
CHAPTER 11 PORTS
Parallel Ports
flow. For the first half of its life, the parallel port kept to this design. Until the adoption of
IEEE 1284, this was the design you could expect for the port on your printer and, almost
as likely, those on your computer.
Each signal flowing through the parallel port in compatibility mode has its own function
in handling the transfer of data.
Data Lines
The eight data lines of the parallel interface convey data in all operating modes. In compatibility mode, they carry data from the host to the peripheral on connector pins 2
through 9. The higher numbered pins are the more significant to the digital code. To
send data to the peripheral, the host puts a pattern of digital voltages on the data lines.
Strobe Line
The presence of signals on the data lines does not, in itself, move information from host
to peripheral. As your computer gets its act together, the pattern of data bits may vary in
the process of loading the correct values. No hardware can ensure that all eight will
always pop to the correct values simultaneously. Moreover, without further instruction
your printer has no way of knowing whether the data lines represent a single character or
multiple repetitions of the same character.
To ensure reliable communications, the system requires a means of telling the peripheral
that the pattern on the data lines represents valid information to be transferred. The strobe
line does exactly that. Your computer pulses the strobe line to tell your printer that the
bit-pattern on the data lines is a single valid character that the printer should read and
accept. The strobe line gives its pulse only after the signals on the data lines have settled
down. Most parallel ports delay the strobe signal by about half a microsecond to ensure
that the data signals have settled. The strobe itself lasts for at least half a microsecond so
that your printer can recognize it. (The strobe signal can last up to 500 microseconds.)
The signals on the data lines must maintain a constant value during this period and
slightly afterward so that your printer has a chance to read them.
The strobe signal is negative going. That is, a positive voltage (+5VDC) stays on the
strobe line until your printer wants to send the actual strobe signal. Your computer then
drops the positive voltage to near zero for the duration of the strobe pulse. The IEEE
1284 specification calls this signal nStrobe.
Busy Line
Sending data to your printer is a continuous cycle of setting up the data lines, sending the
strobe signal, and putting new values on the data lines. The parallel port design typically
requires about two microseconds for each turn of this cycle, allowing a perfect parallel
port to dump out nearly half a million characters a second into your hapless printer. (As
you will see, the actual maximum throughput of a parallel port is much lower than this.)
401
402
For some printers, coping with that data rate is about as daunting as trying to catch
machine gun fire with your bare hands. Before your printer can accept a second character,
its circuitry must do something with the one it has just received. Typically, the printer will
need to move the character into its internal buffer. Although the character moves at electronic speeds, it does not travel instantaneously. Your printer needs to be able to tell your
computer to wait for the processing of the current character before sending the next.
The parallel ports busy line gives your printer the needed breathing room. Your printer
switches on the busy signal as soon as it detects the strobe signal and keeps the signal
active until it is ready to accept the next character. The busy signal can last for a fraction
of a second (even as short as a microsecond), or your printer could hold it on indefinitely
while it waits for you to correct some error. No matter how long the busy signal is on, it
keeps your computer from sending out more data through the parallel port. It functions as
the basic flow-control system.
Acknowledge Line
The final part of the flow-control system of the parallel port is the acknowledge line. It tells
your computer that everything has gone well with the printing of a character or its transfer to the internal buffer. In effect, it is the opposite of the busy signal, telling your computer that the printer is ready rather than unready. Whereas the busy line says, Whoa,
the acknowledge line says, Giddyap! The acknowledge signal is the opposite of the busy
signal in another way: It is negative going whereas the busy signal is positive going. The
IEEE 1284 specification calls this signal nAck.
When your printer sends out the acknowledge signal, it completes the cycle of sending a
character. Typically the acknowledge signal on a conventional parallel port lasts about
eight microseconds, stretching a single character cycle across the port to 10 microseconds.
(IEEE 1284 specifies the length of nAck to be between 0.5 and 10 microseconds.) If you
assume the typical length of this signal for a conventional parallel port, the maximum
speed of the port works out to about 100,000 characters per second.
Select
In addition to transferring data to the printer, the basic parallel port allows your printer to
send signals back to your computer so that your computer can monitor the operation of
the printer. The original IBM design of the parallel interface includes three such signals
that tell your computer when your printer is ready, willing, and able to do its job. In
effect, these signals give your computer the ability to remotely sense the condition of your
printer.
The most essential of these signals is select. The presence of this signal on the parallel
interface tells your computer that your printer is online (that is, the printer is switched on
and is in its online mode, ready to receive data from your computer). In effect, it is a
CHAPTER 11 PORTS
Parallel Ports
remote indicator for the online light on your printers control panel. If this signal is not
present, your computer assumes that nothing is connected to your parallel port and
doesnt bother with the rest of its signal repertory.
Because the rest state of a parallel port line is an absence of voltage (which would be the
case if nothing were connected to the port to supply the voltage), the select signal takes
the form of a positive signal (nominally +5VDC) that in compatibility mode under the IEEE
1284 specification stays active the entire period your printer is online.
Paper Empty
To print anything your printer needs paper, and the most common problem that prevents
your printer from doing its job is running out of paper. The paper empty signal warns your
computer when your printer runs out. The IEEE 1284 specification calls this signal
PError for paper error, although it serves exactly the same function.
Paper empty is an information signal. It is not required for flow control because the busy
signal more than suffices for that purpose. Most printers will assert their busy signals for
the duration of the period they are without paper. Paper empty tells your computer the
specific reason that your printer has stopped data flow. This signal allows your operating
system or application to flash a message on your monitor to warn you to load more paper.
Fault
The third printer-to-computer status signal is fault, a catchall for warning of any other
problems that your printer may developout of ink, paper jams, overheating, conflagrations, and other disasters. In operation, fault is actually a steady-state positive signal. It
dips low (or off) to indicate a problem. At the same time, your printer may issue its other
signals to halt the data flow, including busy and select. It never hurts to be extra sure.
Because this signal is negative going, the IEEE specification calls it nFault.
Initialize Printer
In addition to the three signals your printer uses to warn of its condition, the basic parallel
port provides three control signals that your computer can use to command your printer
without adding anything to the data stream. Each of these three provides its own hardwired connection for a specific purpose. These include one to initialize the printer,
another to switch it to online condition (if the printer allows a remote control status
change), and a final signal to tell the printer to feed the paper up one line.
The initialize printer signal helps your computer and printer keep in sync. Your printer can
send a raft of different commands to your printer to change its mode of operation, change
font, alter printing pitch, and so on. Each of your applications that share your printer
might send out its own favored set of commands. And many applications are like sloppy
in-laws that come for a visit and fail to clean up after themselves. The programs may leave
403
404
your printer in some strange condition, such as set to print underscored boldface characters in agate size type with a script typeface. The next program you run might assume
some other condition and blithely print out a paycheck in illegible characters.
Initialize printer tells your printer to step back to ground zero. Just as your computer
boots up fresh and predictably, so does your printer. When your computer sends your
printer the initialize printer command, it tells the printer to boot up (that is, reset itself
and load its default operating parameters with its startup configuration of fonts, pitches,
typefaces, and the like). The command has the same effect as you switching off the printer
and turning it back on and simply substitutes for adding a remote control arm on your
computer to duplicate your actions.
During normal operation, your computer puts a constant voltage on the initialize printer
line. Removing the voltage tells your printer to reset. The IEEE 1284 specification calls
this negative-going signal nInit.
Select Input
The signal that allows your computer to switch your printer online and offline is called
select input. The IEEE 1284 specification calls it nSelectIn. It is active, forcing your printer
online, when it is low or off. Switching it to high deselects your printer.
Not all printers obey this command. Some have no provisions for switching themselves
on- and offline. Others have setup functions (such as a DIP switch) that allow you to
defeat the action of this signal.
Auto Feed XT
At the time IBM imposed its print system design on the rest of the world, different printers interpreted the lowly carriage return in one of two ways. Some printers took it literally. Carriage return means to move the printhead carriage back to its starting position on
the left side of the platen. Other printers thought more like typewriters. Moving the
printhead full left also indicated the start of a new line, so they obediently advanced the
paper one line when they got a carriage return command. IBM, being a premiere typewriter-maker at the time, opted for this second definition.
To give printer developers flexibility, however, the IBM parallel port design included the
Auto Feed XT signal to give your computer command of the printers handling of carriage
returns. Under the IEEE 1284 specification, this signal is called nAutoFd. By holding this
signal low or off, your computer commands your printer to act in the IBM and typewriter
manner, adding a line feed to every carriage return. Making this signal high tells your
printer to interpret carriage returns literally and only move the printhead. Despite the
availability of this signal, most early computer printers ignored it and did whatever their
setup configuration told them to do with carriage returns.
CHAPTER 11 PORTS
Parallel Ports
Nibble Mode
Early parallel ports used unidirectional circuitry for their data lines. No one foresaw the
need for your computer to acquire data from your printer, so there was no need to add the
expense or complication of bidirectional buffers to the simple parallel port. This tradition
of single-direction design and operation continues to this day in the least expensive
(which, of course, also means cheapest) parallel ports.
Every parallel port does, however, have five signals that are meant to travel from the
printer to your computer. These include (as designated by the IEEE 1284 specification)
nAck, Busy, PError, Select, and nFault. If you could suspend the normal operation of
these signals temporarily, you could use four of them to carry data back from the printer
to your computer. Of course, the information would flow at half speed, four bits at a time.
This means of moving data is the basis of nibble mode, so called because the computer
community calls half a byte (the aforementioned four bits) a nibble. Using nibble mode,
any parallel port can operate bidirectionally full speed forward but half speed in reverse.
Nibble mode requires that your computer take explicit command and control the operation of your parallel port. The port itself merely monitors all its data and monitoring signals and then relays the data to your computer. Your computer determines whether to
regard your printers status signals as backward-moving data. Of course, this system also
requires that the device at the other end of the parallel port (your printer or whatever)
know that it has switched into nibble mode and understand what signals to put where and
when. The IEEE 1284 specification defines a protocol for switching into nibble mode and
how computer and peripherals handle the nibble-mode signals.
The process is complex, involving several steps. First, your computer must identify
whether the peripheral connected to it recognizes the IEEE standard. If not, all bets are
off for using the standard. Products created before IEEE 1284 was adopted needed special
drivers that matched the port to a specific peripheral. Because the two were already
matched, they knew everything they needed to know about each other without negotiation. The pair could work without understanding the negotiation process or even the
IEEE 1284 specification. Using the specification, however, allows your computer and
peripherals to do the matching without your intervention.
Once your computer and peripheral decide they can use nibble mode, your computer signals to the peripheral to switch to the mode. Before the IEEE 1284 standard, the protocol
was proprietary to the parallel port peripheral. The standard gives all devices a common
means of controlling the switchover.
After both your computer and parallel port peripheral have switched to nibble mode, the
signals on the interface get new definitions. In addition, nibble mode itself operates in two
modes or phases, and the signals on the various parallel port lines behave differently in
each mode. These modes include reverse idle phase and reverse data transfer phase.
405
406
In reverse idle phase, the PtrClk signal (nAck in compatibility mode) operates as an attention signal from the parallel port peripheral. Activating this signal tells the parallel port to
issue an interrupt inside your computer, signaling that the peripheral has data available to
be transferred. Your computer acknowledges the need for data and requests its transfer by
switching the HostBusy signal (nAutoFd in compatibility mode) to low or off. This
switches the system to reverse data transfer phase. Your computer switches the HostBusy
signal to high again after the completion of the transfer of a full data byte. When the
peripheral has mode data ready and your computer switches HostBusy back to low again,
another transfer begins. If it switches to low without the peripheral having data available
to send, the transition reengages reverse idle phase.
Because moving a byte from peripheral to computer requires two nibble transfers, each of
which requires the same time as one byte transfer from computer to peripheral, reverse
transfers in nibble mode operate at half speed at best. The only advantage of nibble mode
is its universal compatibility. Even before the IEEE 1284 specification, it allowed any parallel port to operate bidirectionally. Because of this speed penalty alone, if you have a
peripheral and parallel port that lets you choose the operating mode for bidirectional
transfers, nibble mode is your least attractive choice.
Byte Mode
Unlike nibble mode, byte mode requires special hardware. The basic design for bytemode circuitry was laid down when IBM developed its PS/2 line of computers and developed the Data Migration Facility. By incorporating bidirectional buffers in all eight of the
data lines of the parallel port, IBM enabled them to both send and receive information on
each end of the connection. Other than that change, the new design involved no other
modifications to signals, connector pin assignments, or the overall operation of the port.
Before the advent of the IEEE standard, these ports were known as PS/2 parallel ports or
bidirectional parallel ports.
IEEE 1284 does more than put an official industry imprimatur on the IBM design, however. The standard redefines the bidirectional signals and adds a universal protocol for
negotiating bidirectional transfers.
As with nibble mode, a peripheral in byte mode uses the PtrClk signal to trigger an interrupt in the host computer to advise that the peripheral has data available for transfer.
When the computer services the interrupt, it checks the port nDataAvail signal, a negative-going signal that indicates a byte is available for transfer when it goes low. The computer can then pulse off the HostBusy signal to trigger the transfer using the HostClk
(nStrobe) signal to read the data. The computer raises the HostBusy signal again to indicate the successful transfer of the data byte. The cycle can then repeat for as many bytes
as need to be sent.
CHAPTER 11 PORTS
Parallel Ports
Because byte mode is fully symmetrical, transfers occur at the same speed in either direction. The speed limit is set by the performance of the port hardware, the speed at which
the host computer handles the port overhead, and the length of timing cycles set in the
IEEE 1284 specification. Potentially the design could require as little as four microseconds for each byte transferred, but real-world systems peak at about the same rate as
conventional parallel ports (100,000 bytes per second).
Enhanced Parallel Port Mode

When it was introduced, the chief innovation of the Enhanced Parallel Port (EPP) was its
improved performance, thanks to a design that hastened the speed at which your computer could pack data into the port. The EPP design altered port hardware so that instead
of using byte-wide registers to send data through the port, your computer could dump a
full 32-bit word of data directly from its bus into the port. The port would then handle all
the conversion necessary to repackage the data into four-byte-wide transfers. The reduction in computer overhead and more efficient hardware design enabled a performance
improvement by a factor of 10 in practical systems. This speed increase required more
stringent specifications for printer cables. The IEEE 1284 specification does not get into
the nitty-gritty of linking the parallel port circuitry to your computer, so it does not guarantee that a port in EPP mode will deliver all this speed boost. Moreover, the IEEE 1284
cable specs are not as demanding as the earlier EPP specs.
EPP mode of the IEEE 1284 specification uses only six signals in addition to the eight
data lines for controlling data transfers. Three more connections in the interface are
reserved for use by individual manufacturers and are not defined under the standard.
A given cycle across the EPP mode interface performs one of four operations: writing an
address, reading an address, writing data, or reading data. The address corresponds to a
register on the peripheral. The data operations are targeted on that address. Multiple data
bytes may follow a single address signal as a form of burst mode.
nWrite
Data can travel both ways through an EPP connection. The nWrite signal tells whether
the contents of the data lines are being sent from your computer to a peripheral or from a
peripheral to your computer. When the nWrite signal is set to low, it indicates data is
bound for the peripheral. When set to high, it indicates data sent from the peripheral.
nDStrobe
Soundboards are heavy feeders when it comes to system resources. A single soundboard
may require multiple interrupts, a wide range of input/output ports, and a dedicated
address range in High DOS memory. Because of these extensive resource demands, the
need for numerous drivers, and often-poor documentation, soundboards are the most
407
408
frustrating expansion products to add to a computer. In fact, a soundboard may be the

perfect gift to surreptitiously gain revenge, letting you bury the hatchet with an estranged
friend without the friend knowing youve sliced solidly into his back.
As with other parallel port transfers, your system needs a signal to indicate when the bits
on the data lines are valid and accurate. EPP mode uses a negative-going signal called
nDStrobe for this function in making data operations. Although this signal serves the same
function as the strobe signal on a standard parallel port, it has been moved to a different
pin, that used by the nAutoFd signal in compatibility mode.
nAStrobe
To identify a valid address on the interface bus, the EPP system uses the nAStrobe signal.
This signal uses the same connection as nSelectIn during compatibility mode.
nWait
To acknowledge that a peripheral has properly received a transfer, it deactivates the negative-going nWait signal (making it a positive voltage on the bus). By holding the signal
positive, the peripheral signals the host computer to wait. Making the signal negative indicates that the peripheral is ready for another transfer.
Intr
A peripheral connected to the EPP interface can signal to the host computer that it
requires immediate service by sending out the Intr signal. The transition between low and
high states of this signal indicates a request for an interrupt (that is, the signal is edge
triggered). EPP mode does not allocate a signal to acknowledge that the interrupt
request was received.
nInit
The escape hatch for EPP mode is the nInit signal. When this signal is activated (making
it low), it forces the system out of EPP mode and back into compatibility mode.
Extended Capabilities Port Mode

When operating in ECP mode, the IEEE 1284 port uses seven signals to control the flow
of data through the standard eight data lines. ECP mode defines two data-transfer signaling protocolsone for forward transfers (from computer to peripheral) and one for
reverse transfers (peripheral to computer)and the transitions between them. Transfers
are moderated by closed-loop handshaking, which guarantees that all bytes get where they
are meant to go, even if the connection is temporarily disrupted.
Because all parallel ports start in compatibility mode, your computer and its peripherals must
first negotiate with one another to arrange to shift into ECP mode. Your computer and its
software initiate the negotiation (as well as manage all aspects of the data transfers). Following
a successful negotiation to enter into ECP mode, the connection enters its forward idle phase.
CHAPTER 11 PORTS
Parallel Ports
HostClk
To transfer information or commands across the interface, your computer starts from the
forward idle phase and puts the appropriate signals on the data line. To signal to your
printer or other peripheral that the values on the data lines are valid and should be transferred, your computer activates its HostClk signal, setting it to a logical high.
PeriphAck
The actual transfer does not take place until your printer or other peripheral acknowledges the HostClk signal by sending back the PeriphAck signal, setting it to a logical high.
In response, your computer switches the HostClk signal to low. Your printer or peripheral
then knows it should read the signals on the data lines. Once it finishes reading the data
signals, the peripheral switches the PeriphAck signal to low. This completes the data
transfer. Both HostClk and PeriphAck are back to their forward idle phase norms, ready
for another transfer.
nPeriphRequest
When a peripheral needs to transfer information back to the host computer or to another
peripheral, it makes a request by driving the nPeriphRequest signal low. The request is a
suggestion rather than a command because only the host computer can initiate or reverse
the flow of data. The nPeriphRequest signal typically causes an interrupt in the host computer to make this request known.
nReverseRequest
To allow a peripheral to send data back to the host or to another device connected to the
interface, the host computer activates the nReverseRequest signal by driving it low, essentially switching off the voltage that otherwise appears there. This signals to the peripheral
that the host computer will allow the transfer.
nAckReverse
To acknowledge that it has received the nReverseRequest signal and that it is ready for a
reverse-direction transfer, the peripheral asserts its nAckReverse signal, driving it low. The
peripheral can then send information and commands through the eight data lines and the
PeriphAck signal.
PeriphClk
To begin a reverse transfer from peripheral to computer, the peripheral first loads the
appropriate bits onto the data lines. It then signals to the host computer that it has data
ready to transfer by driving the PeriphClk signal low.
HostAck
Your computer responds to the PeriphClk signal by switching the HostAck signal from its
idle logical low to a logical high. The peripheral responds by driving PeriphClk high.
409
410
When the host accepts the data, it responds by driving the HostAck signal low. This completes the transfer and returns the interface to the reverse idle phase.
Data Lines
Although the parallel interface uses the same eight data lines to transfer information as do
other IEEE 1284 port modes, it supplements them with an additional signal to indicate
whether the data lines contain data or a command. The signal used to make this nine-bit
information system changes with the direction of information transfer. When ECP mode
transfers data from the computer host to a peripheral (that is, during a forward transfer), it
uses the HostAck signal to specify a command or data. When a peripheral originates the
data being transferred (a reverse transfer), it uses the PeriphAck signal to specify a command or data.
Logical Interface
Your computer controls each of its parallel ports through a set of three consecutive
input/output ports. The typical computer sets aside three sets of these ports for three parallel ports, although most systems provide the matching hardware only for one. The base
addresses used by parallel ports include 03BC(hex), 0378(hex), and 0278(hex).
When Windows boots up, it scans these addresses and assigns a logical name to each.
These names are LPT1, LPT2, and LPT3. The name is a contraction of Line Printer,
echoing the original purpose of the port. The port with the name LPT1 can also use the
alias PRN. You can use these names at the system command prompt to identify a parallel
port and the printer connected to it.
The computer printer port was designed to be controlled by a software driver. Your computers BIOS provides a rudimentary driver, but most advanced operating systems similarly take direct hardware control of the parallel port through their own software drivers.
Windows includes a parallel port driver of its own. You may need to install drivers for any
device that connects to your parallel port. For example, every printer requires it own driver (many of whichbut not allare built in to Windows).
Control
Even in its immense wisdom, a microprocessor cant fathom how to operate a parallel port
by itself. It needs someone to tell it how to move the signals around. Moreover, the minutiae of constantly taking care of the details of controlling a port would be a waste of the
microprocessors valuable time. Consequently, system designers created help systems for
your computers big brain. Driver software tells the microprocessor how to control the
port, and port hardware handles all the details of port operation.
CHAPTER 11 PORTS
Parallel Ports
As parallel ports have evolved, so have these aspects of their control. The software that
controls the traditional parallel port thats built in to the firmware of your computer has
given way to a complex system of drivers. The port hardware, too, has changed to both
simplify operation and to speed it up.
These changes dont follow the neat system of modes laid down by IEEE 1284. Instead,
they have undergone a period of evolution in reaching their current condition.
Traditional Parallel Ports

In the original computer, each of its parallel ports linked to the computers microprocessor
through three separate I/O ports, each controlling its own register. The address of the
first of these registers served as the base address of the parallel port. The other two
addresses are next higher in sequence. For example, when the first parallel port in a computer has a base address of 0378 (hex), the other two I/O ports assigned it have addresses
of 0379 (hex) and 037A (hex).
The register at the base address of the parallel port serves as a data latch, called the printer
data register, which temporarily holds the values passed along to it by your computers
microprocessor. Each of the eight bits of this port is tied to one of the data lines leading
out of the parallel port connector. The correspondence is exact. For example, the most
significant bit of the register connects to the most significant bit on the port connector.
When your computers microprocessor writes values to the base register of the port, the
register latches those values until your microprocessor sends newer values to the port.
Your computer uses the next register on the parallel port, corresponding to the next I/O
port, to monitor what the printer is doing. Termed the printer status register, the various
bits that your microprocessor can read at this I/O port carry messages from the printer
back to your computer. The five most significant bits of this register directly correspond
to five signals appearing in the parallel cable: Bit 7 indicates the condition of the busy signal; bit 6, acknowledge; bit 5, paper empty; bit 4, select; and bit 3, error. The remaining
three bits of this register (bits 2, 1, and 0the least significant bits) served no function in
the original computer parallel port.
To send commands to your printer, your computer uses the third I/O port, offset two
ports from the base address of the parallel port. The register there, called the printer control register, relays commands through its five least significant bits. Of these, four directly
control corresponding parallel port lines. Bit 0 commands the strobe line; bit 1, the Auto
Feed XT line; bit 2, the initialize line; and bit 3, the select line.
To enable your printer to send interrupts to command the microprocessors attention, your
computer uses bit 4 of the printer control register. Setting this bit to high causes the
acknowledge signal from the printer to trigger a printer interrupt. During normal operation
your printer, after it receives and processes a character, changes the acknowledge signal
411
412
from a logical high to a low. Set bit 4, and your system detects the change in the acknowledge line through the printer status register and executes the hardware interrupt assigned to
the port. In the normal course of things, this interrupt simply instructs the microprocessor
to send another character to the printer.
All the values sent to the printer data register and the printer control register are put in
place by your computers microprocessor, and the chip must read and react to all the values
packed into the printer status register. The printer gets its instructions for what to do from
firmware that is part of your systems ROM BIOS. The routines coded for interrupt vector
017 (hex) carry out most of these functions. In the normal course of things, your applications call interrupt 017 (hex) after loading appropriate values into your microprocessors
registers, and the microprocessor relays the values to your printer. These operations are
very microprocessor intensive. They can occupy a substantial fraction of the power of a
microprocessor (particularly that of older, slower chips) during print operations.
Enhanced Parallel Ports

Intel set the pattern for Enhanced Parallel Port (EPP) by integrating the design into the
386SL chipset (which comprised a microprocessor and a support chip, the 386SL itself,
and the 82360SL I/O subsystem chip, which together required only memory to make a
complete computer). The EPP was conceived as a superset of the standard and PS/2 parallel ports. As with those designs, compatible transfers require the use of the three parallel
port registers at consecutive I/O port addresses. However, EPP adds five new registers to
the basic three. Although designers are free to locate these registers wherever they want
because they are accessed using drivers, in the typical implementation, these registers
occupy the next five I/O port addresses in sequence.
EPP Address Register
The first new register (offset three from the base I/O port address) is called the EPP
address register. It provides a direct channel through which your computer can specify
addresses of devices linked through the EPP connection. By loading an address value in
this register, your computer could select from among multiple devices attached to a single
parallel port, at least once parallel devices using EPP addressing become available.
EPP Data Registers
The upper four ports of the EPP system interface (starting at offset four from the base
port) link to the EPP data registers, which provide a 32-bit channel for sending data to the
EPP data buffer. The EPP port circuitry takes the data from the buffer, breaks it into four
separate bytes, and then sends the bytes through the EPP data lines in sequence. Substituting four I/O ports for the one used by standard parallel ports moves the conversion
into the port hardware, relieving your system from the responsibility of formatting the
data. In addition, your computer can write to the four EPP data registers simultaneously
using a single 32-bit double-word in a single clock cycle in computers that have 32-bit
CHAPTER 11 PORTS
Parallel Ports
data buses. In lesser machines, the EPP specification also allows for byte-wide and wordwide (16-bit) write operations through to the EPP data registers.
Unlike standard parallel ports, which require your computers microprocessor to shepherd
data through the port, the Enhanced Parallel Port works automatically. It requires no other
signals from your microprocessor after it loads the data in order to carry out a data transfer.
The EPP circuitry itself generates the data strobe signal on the bus almost as soon as your
microprocessor writes to the EPP data registers. When your microprocessor reads data
from the EPP data registers, the port circuitry automatically triggers the data strobe signal
to tell whatever device thats sending data to the EPP connection that your computer is
ready to receive more data. The EPP port can consequently push data through to the data
lines with a minimum of transfer overhead. This streamlined design is one of the major
factors that enables the EPP to operate so much faster than standard ports.
Fast Parallel Port Control Register
To switch from standard parallel port to bidirectional to EPP operation requires only
plugging values into one of the registers. Although the manufacturers can use any design
they want, needing only to alter their drivers to match, most follow the pattern set in the
SL chips. Intel added a software-controllable fast parallel port control register as part of the
chipset. This corresponds to the unused bits of the standard parallel port printer control
register.
Setting the most significant bit (bit 7) of the fast parallel port control register to high
engages EPP operation. Setting this bit to low (the default) forces the port into standard
mode. Another bit controls bidirectional operation. Setting bit 6 of the fast parallel port
control register to high engages bidirectional operation. When low, bit 6 keeps the port
unidirectional.
In most computers, an EPP doesnt automatically spring to life. Simply plugging your
printer into EPP hardware wont guarantee fast transfers. Enabling the EPP requires a
software driver that provides the link between your software and the EPP hardware.
Extended Capabilities Ports

As with other variations on the basic parallel port design, your computer controls an
Extended Capabilities Port (ECP) through a set of registers. To maintain backward compatibility with products requiring access to a standard parallel port, the ECP design starts
with the same trio of basic registers. However, it redefines the parallel port data in each of
the ports different operating modes.
The ECP design supplements the basic trio of parallel port registers with an additional set
of registers offset at port addresses 0400 (hex) higher than the base registers. One of
these, the extended control register, controls the operating mode of the ECP port.
413
414
As with other improved parallel port designs, ECP behaves exactly like a standard parallel
port in its default mode. Your programs can write bytes to its data register (located at the
ports base address, just as with a standard parallel port) to send the bits through the data
lines of the parallel connection. Switch to EPP or ECP mode, and your programs can
write at high speed to a register as wide as 32 bits. The ECP design allows for transfers 8,
16, or 32 bits wide at the option of the hardware designer.
To allow multiple devices to share a single parallel connection, the ECP design incorporates its own addressing scheme that allows your computer to separately identify and send
data to up to 128 devices. When your computer wants to route a packet or data stream
through the parallel connection to a particular peripheral, it sends out a channel address
command through the parallel port. The command includes a device address. When an
ECP parallel device receives the command, it compares the address to its own assigned
address. If the two do not match, the device ignores the data traveling through the parallel
connection until your computer sends the next channel address command through the
port. When your computer fails to indicate a channel address, the data gets broadcast to
all devices linked to the parallel connection.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
12
Local Area Networking

Computers are conspirators. They owe most of their success to deviously
working together, subverting the established order like Bolsheviks plotting the overthrow of the Czar. Each computer by itself can be safely
contained and controlled in a home or office. Linked together, however,
and theres no telling what mischief they will get into. They might join
together to pass subversive messages (which we call email). They might
pool their resources so they can take advantage of powers beyond their
capabilities and budgets (such as sharing a printer or a high-speed
Internet connection). They can be each others eyes and ears, letting
them pry into affairs halfway around the world. They can gather information about youeven the most delicate personal matters, such as your
credit card numberand share it with other conspirators (I mean computers, of course) elsewhere in the world. They can just make your life
miserableoh wait, they did that when they werent connected, too.
This computer conspiracy is the key to the power of the local area network and the Internet. It gives your computer more reach, more power,
and more capabilities. It simply makes your life more convenient and
your computer more connected. In fact, at one time local area networking was called connectivity.
Fortunately, computers themselves dont plot against you (although the
people behind those computer may). Even when connected to a network,
computers remain their obedient selves, ready to do your bidding.
The challenge you face in linking your computer to others is the same as
faced by a child growing up with siblingsyour computer has to learn to
418
share. When kids share, you get more quiet, greater peace of mind, and less bloodshed.
When computers share, you get the convenience of using the same files and other
resources, centralized management (including the capability to back up all computers
from one location or use one computer to back up others), and improved communication
between workers in your business.
The drawback to connectivity is that computer networks are even more difficult to understand and manage than a platoon of teenagers. They have their own rules, their own value
systems, their own hardware needs, and even their own language. Just listening in on a
conversation between network pros is enough to make you suspect that an alien invasion
from the planet Oxy-10 has succeeded. To get even a glimmer of understanding, you need
to know your way around layers of standards, architectures, and protocols. Installing a
network operating system can take system managers days; deciphering its idiosyncrasies
can keep users and operators puzzled for weeks. Network host adapters often prove
incompatible with other computer hardware, with their required interrupts and I/O
addresses locking horns with SCSI boards, port controllers, and other peripherals. And
weaving the wiring for a network is like threading a needle while wearing boxing gloves
during a cyclone that has blown out the electricity, the candles, and your last rays of hope.
In fact, no one in his right mind would tangle with a network were not the benefits so
great. File sharing across the network alone eliminates a major source of data loss, which
is duplication of records and out-of-sync file updates. Better still, a network lets you get
organized. You can put all your important files in one central location where they are easier to protect, both from disaster and theft. Instead of worrying about backing up half a
dozen computers individually, you can easily handle the chore with one command.
Electronic mail can bring order to the chaos of tracking messages and appointments, even
in a small office. With network-based email, you can communicate with your coworkers
without scattering memo slips everywhere. Sharing a costly laser printer or large hard disk
(with some networks, even modems) can cut your capital cost of the computers equipment by thousands or tens of thousands of dollars. Instead of buying a flotilla of personal
laser printers, for example, you can serve everyones hard copy needs with just one
machine.
Concepts
Network designers get excited when someone expresses interest in their rather esoteric
field, which they invariable serve up as a layer cake. Far from a rich devils food with a
heavy chocolate frosting (with jimmies), they roll out a network model, something that
will definitely disappoint your taste buds because its entirely imaginary. They call their
cake the Open Systems Interconnection Reference Model (or the OSI Model for short), and its a
layer cake the way software is layered. It represents the structure of a typical network with
the most important functions each given its own layer.
CHAPTER 12 LOCAL AREA NETWORKING
Concepts
Although the discussion of networks usually begins with this serving of cake, the OSI
Model does not represent any particular network. In fact, networks are by no means
obliged to follow the model. Although most have layers of functions, they are often
stacked differently. And while the OSI Model preaches a seven-layer model, even the foremost networkers in the worldthe Institute of Electrical and Electronic Engineersslip
in one more layer for good measure. Table 12.1 highlights the equivalence of different
layers in several common networking systems, including the OSI model.
TA B L E 1 2 . 1
Approximate Equivalence of Network Layers
Layer
OSI Model
Internet Model
Windows Model
Real World
Physical
Not specified
Adapter
Twisted-pair
Data Link
Not specified
Network
Internet Protocol
Transport
Transport Control
Protocol (TCP)
Session
Presentation
FTP, HTTP
Service
Operating system
Application
Not specified
Client
Browser
NIC
Protocol
Network operating system
Network driver
What the OSI Model shows is how many different functions you need to make a network
and how those functions interrelate to one another. Its a good place to begin, and it adds
an academic aura to our network discussion, which just might make you believe youre
learning something useful.
OSI Model
In 1984, the International Standards Organization laid out a blueprint to bring order to
the nonsense of networking by publishing the Open Systems Interconnection Reference
Model. You can imagine the bickering between hundreds of engineers trying to refine the
right recipe for cooking up a network, then arguing over how many layers into which to
divide their batter. The final, seven-layer model looks as if it might be the one and only
way to build a network. These layers, ranging from the connecting wire to software applications, define functions and protocols that enable the wide variety of network hardware
and software to work together. It just seems so logical, so compelling, that it is taught in
most colleges and endorsed by major organizations such as IBM.
419
420
Regardless of where it fits in with your beliefs, philosophies, and diet, the OSI Model presents an excellent way to understand networks. The layering defined by the OSI
Reference Model illustrates how the various elements of a networkfrom the wire running through your office ceiling to the Windows menu of your mail programfit
together and interact. Although few actual networks or network products exactly fit the
model, the layers show how networks must be structured as well as the problems in building a network.
Physical
The first layer of the OSI Reference Model is the Physical layer. It defines the basic hardware of the network, which is the cable that conducts the flow of information between the
devices linked by the network, or even the lack of a cable in wireless networking designs.
This layer defines not only the type of wire (for example, coaxial cable, twisted-pair wire,
and so on) but also the possible lengths and connections of the wire, the signals on the
wire, and the interfaces of the cabling systemor the frequency and modulation of the
radio devices that carry wireless network signals. This is the level at which the device that
connects a computer to the network (the network host adapter) is defined.
Data Link
Layer 2 in a network is called the Data Link layer. It defines how information gains access
to the wiring system. The Data Link layer defines the basic protocol used in the local network. This is the method used for deciding which computer can send messages over the
cable at any given time, the form of the messages, and the transmission method of those
messages.
This level defines the structure of the data that is transferred across the network. All data
transmitted under a given protocol takes a common form called the packet or network data
frame. Each packet is a block of data that is strictly formatted and may include destination
and source identification as well as error-correction information. All network data transfers are divided into one or more packets, the lengths of which are carefully controlled.
Breaking network messages into multiple packets enables the network to be shared without interference and interminable waits for access. If you transfer a large file (say, a
bitmap) across the network in one piece, you might monopolize the entire network for the
duration of the transfer. Everyone would have to wait. By breaking all transfers into manageable pieces, everyone gets access in a relatively brief period, thus making the network
more responsive.
Network
Layer 3 in the OSI Reference Model is the Network layer, which defines how the network
moves information from one device to another. This layer corresponds to the hardwareinterface function of the BIOS of an individual computer because it provides a common
Concepts
software interface that hides differences in underlying hardware. Software of higher layers
can run on any lower-layer hardware because of the compatibility this layer affords.
Protocols that enable the exchange of packets between different networks operate at this
level.
Transport
Layer 4 controls data movement across the network. The Transport layer defines how
messages are handledparticularly how the network reacts to packets that become lost as
well as other errors that may occur.
Session
Layer 5 of the OSI Reference Model defines the interaction between applications and
hardware, much as a computer BIOS provides function calls for programs. By using functions defined at this Session layer, programmers can create software that will operate on
any of a wide variety of hardware. In other words, the Session layer provides the interface
for applications and the network. Among computers, the most common of these application interfaces is IBMs Network Basic Input/Output System (NetBIOS).
Presentation
Layer 6, the Presentation layer, provides the file interface between network devices and the
computer software. This layer defines the code and format conversions that must take
place so that applications running under a computer operating system, such as DOS,
OS/2, or Macintosh System 7, can understand files stored under the networks native
format.
In classic networking, this function would be served by the computers BIOS. The first
personal computers lacked any hint of network connectibility, but in 1984 IBM introduced a trend-setting system called the IBM PC Network, which has been the foundation
for small computer networking ever since. The critical addition was a set of new codes to
the system BIOS developed by Sytek. This set of new codesspecifically, the Interrupt
5C(hex) routineshas become known as the Network BIOS, or NetBIOS.
The NetBIOS serves as a low-level application program interface for the network, and its
limitations get passed on to networks built around it. In particular, the NetBIOS imposed
requires that each computer on the network wear a unique name up to 15 characters long.
This limits the NetBIOS to smaller networks. Todays operating systems use driver software that takes the place of the NetBIOS.
Application
Layer 7 is the part of the network that you deal with personally. The Application layer
includes the basic services you expect from any network, including the ability to deal with
files, send messages to other network users through the mail system, and control print
jobs.
421
422
Internet Model
The Internet does not neatly fit the seven-layer OSI Model for the simple reason that the
Internet is not meant to be a network. It describes a network of networks. Consequently,
the Internet standards do not care about, nor do they describe, a physical layer or an
application layer. (The World Wide Web standards arguably define a presentation
layer.) When you use the Internet, however, the computer network you use to tie into the
Internet becomes a physical layer, and your browser acts as an application layer.
Similarly, Ethernet does not conform to a seven-layer model because its concerns are at
only the Physical through Session layers. Note that Ethernet breaks the Data Link layer
into its own two-layer system.
Windows Model
When you work across a network using Windows, all the OSI layers come into play.
Windows sees networks as layered somewhere between the two-tier practical approach
and the seven layers of OSI. Microsoft assigns four levels of concern when configuring
your network, using your network or operating system software. Under Windows, these
levels include the adapter, protocol, service, and client software, as shown in the Windows
Select Network Component Type dialog box (see Figure 12.1).
FIGURE 12.1
The Windows Select
Network Component
Type dialog box.
In this model, the network has four componentsthe client, the adapter, the protocol,
and the service. Each has a distinct role to play in making the connection between computers.
Adapter
The adapter is the hardware that connects your computer to the network. The common
term network interface card, often abbreviated as NIC, is one form of host adapter, a board
you slide into an expansion slot. The host adapter function also can be integrated into the
circuitry of the motherboard, as it often is in business-oriented computers.
Concepts
No matter its form or name, the adapter is the foundation of the physical layer of the network, the actual hardware that makes the network work with your computer. It translates
the parallel bus signals inside your computer into a serial form that can skitter through the
network wiring. The design and standards that the adapter follows determine the form
and speed of the physical side of the network.
From a practical perspective, the network adapter is generally the part of the network that
you must buy if you want to add network capabilities to a computer lacking them. You
slide a network host adapter or NIC into an expansion slot in your computer to provide a
port for plugging into the network wire.
Protocol
The protocol is the music of the packets, the lyrics that control the harmony of the data
traffic through the network wiring. The protocol dictates not only the logical form of the
packetthe arrangement of address, control information, and data among its bytesbut
also the rules on how the network deals with the packets. The protocol determines how
the packet gets where it is going, what happens when it doesnt, and how to recover when
an error appears in the data as it crosses the network.
Support for the most popular and useful protocols for small networks is included with
todays operating systems. It takes the form of drivers you install to implement a particular
networking system. Windows, for example, includes several protocols in its basic package.
If you need to, you can add others as easily as installing a new software driver.
Service
The service of the network is the work the packets perform. The services are often several
and always useful. Network services include exchanging files between disk drives (or making a drive far away on the network appear to be local to any or every computer in the
network), sharing a printer resource so that all computers have access to a centralized
printer, and passing electronic mail from a centralized post office to individual machines.
Most networking software includes the more useful services as part of the basic package.
Windows includes file and printer sharing as its primary services. The basic operating system also includes e-mail support. Again, new services are as easy to add as new driver software.
Client
To the network, the client is not you but rather where the operating system of your computer and the network come together. Its yet another piece of software, the one that
brings you the network resources so that you can take advantage of the services. The
client software allows the network to recognize your computer and to exchange data packets with it.
423
424
Architecture
Now that were done with the stuff that college professors want you to know to understand networks, lets get back to the real world and look at how networks are put together.
Networks link computers, but a link can take many forms, which a visit to any sausage
shop will readily confirm.
With networks, computers link together in many ways. The physical configuration (something those professors call a topology), describes how computers are connected, like the
sausages that squirt out of the grinder. Sausages link in a straight line, one after another.
Networks are more versatile.
Moreover, by their nature, networks define hierarchies of computers. Some networks
make some computers more important than others. The terms describing these arrangements are often bandied about as if they mean something. The descriptions that follow
should help you keep up with the conversation.
Topologies
In mathematics, topology is an elusive word. Some mathematicians see it as the study of
geometry without shapes or distance. Some definitions call it the study of deformation. It
is the properties of an object that survive when you change its size or shape by stretching
or twisting it. A beach ball has the same topology with or without air. Concepts such as
inside and outside survive the deformation, although they change physical places. Its a rich,
rewarding, and fun part of mathematics.
Topology describes how the parts of an object relate to and connect with one another.
The same word is applied to networks to describe how network nodes connect together. It
describes the network without reference to distance or physical location. It shows the
paths that signals must travel from one node to another.
Designers have developed several topologies for computer networks. Most can be reduced
to one of four basic layouts: linear, ring, star, and tree. The names describe how the cables
run throughout an installation.
Linear
A network with linear cabling has a single backbone, one main cable that runs from one
end of the system to the other. Along the way, computers tap into this backbone to send
and receive signals. The computers link to the backbone with a single cable through
which they both send and receive. In effect, the network backbone functions as a data bus,
and this configuration is often called a bus topology. Its as close as a network comes to the
sausages streaming from the grinding machine. Figure 12.2 shows a network bus.
Architecture
FIGURE 12.2
Simple bus network
topology.
In the typical installation, a wire leads from the computer to the backbone, and a T-connector links the two. The network backbone has a definite beginning and end. In most
cases, these ends are terminated with a resistor matching the characteristic impedance of
the cable in the background. That is, a 61-ohm network cable will have a 61-ohm termination at either end. These terminations prevent signals from reflecting from the ends of
the cable, thus helping ensure signal integrity.
Ring
The ring topology looks like a linear network thats biting its own tail. The backbone is a
continuous loop, a ring, with no end. But the ring is not a single, continuous wire. Instead,
it is made of short segments daisy-chained from one computer to the next, the last connected, in turn, to the first. Each computer thus has two connections. One wire connects a
computer to the computer before it in the ring, and a second wire leads to the next computer in the ring. Signals must traverse through one computer to get to the next, and the
signals typically are listened to and analyzed along the way. You can envision it as a snake
eating its own tail (or the first sausage swallowing the last link). If neither image awakens
your imagination, Figure 12.3 also shows a network ring.
FIGURE 12.3
A simple network
ring.
Star
Just as rays blast out from the core of a star, in the star topology, connecting cables
emanate from a centralized location called a hub, and each cable links a single computer
to the network. A popular image for the star topology is an old-fashioned wagon wheel
the network hub is the hub of the wheel, the cables are the spokes, and the computers are
ignored in the analogy. Try visualizing them as clumps of mud clinging to the rim (which,
depending on your particular network situation, may be an apt metaphor). Figure 12.4
shows a small network using the star topology.
425
426
FIGURE 12.4
Simple star network
topology.
Star-style networks have become popular because their topology matches that of other
office wiring. In the typical office building, the most common wiring is used by telephones, and telephone wiring converges at the wiring closet, which is the Private Branch
Exchange, or PBX (the telephone switching equipment for a business). Star-style topologies require only a single cable and connection for each device to link to the central location where all cables converge into the network hub.
Tree
When multiple hubs are connected together, the result is a tree. More like a family tree
than some great oak, the tree spreads out and, at least potentially, connects many more
nodes than might a single hub. Figure 12.5 shows a small network tree.
FIGURE 12.5
A simple network
tree.
The typical home or small business network is a star, whereas large networks take the
form of trees. (And network users eat the sausages.)
Hierarchies
Computers have classesor at least a class systemin some networking schemes.
Whereas some networks treat all computers the same, others elevate particular computers
to a special, more important role as servers. Although the network performs many of the
Architecture
same functions in either case, these two hierarchical systems enforce a few differences in
how the network is used.
Client/Server
In the client/server system, the shared resources of the networkfiles, high-speed connections with the Internet, printers, and the email systemare centralized on one or more
powerful computers with very large disk storage capacity called servers. Individual workstation computers used by workers at the company are called clients.
Exactly what is on the server (or servers) and what is local to each client depends on the
choices of the network manager. The server may host as little as a single shared database.
The server may also host the data file used by each client. In extreme cases, programs
used by each client load from a disk in the server.
As the network places more reliance on the server, the load on both the network and
server increase. Because nearly every computer operation becomes a network operation,
normal day-to-day use swallows up network bandwidth, and a slowdown in the network or
the server slows work on every computer.
The strong point of the centralized client/server network is ease of administration. The
network and all its connected clients are easier to control. Centralizing all data files on the
server makes data and programs easier to secure and back up. Also, it helps ensure that
everyone on the network is using the same programs and the same data.
When you surf the Internet, your computer becomes the client, and the computer that
provides the Web pages you read is the server.
Peer-to-Peer
The client/server system is a royalist system, particularly if you view a nations leader as a
servant of the people rather than a profiteer. The opposite is the true democracy, in which
every computer is equal. Computers share files and other resources (such as printers)
among one another. They share equally, each as the peer of the others, so this scheme is
called peer-to-peer networking.
Peer-to-peer means that there is no dedicated file server. All computers have their own,
local storage, but each computer is (or can be) granted access to the disk drives and printers connected to the other computers. The peer-to-peer system is not centralized. In fact,
it probably has no center, only a perimeter.
When you use the Internet to share files (say, downloading an MP3 file from someone
elses collection), youre using peer-to-peer networking.
In a peer-to-peer network, no one computer needs to be particularly endowed with overwhelming mass storage or an incomprehensible network operating system. But all computers need not be equal. In fact, one peer may provide a network resource, such as a
427
428
connection to a printer or the Internet. It may even have a large disk thats used to back
up the files from other peers.
In other words, the line between client/server and peer-to-peer systems can be fuzzy,
indeed. There is no electrical difference between the two systems. A peer-to-peer network
and a client/server network may use exactly the same topology. In fact, the same network
may take different characterizations, depending on how it is used.
Standards
A network is a collection of ideas, hardware, and software. The software comprises both
the programs that make it work and the protocols that let everything work together. The
hardware involves the network adapters as well as the wires, hubs, concentrators, routers,
and even more exotic fauna. Getting it all to work together requires standardization.
Because of the layered design of most networks, these standards can appear at any level in
the hierarchy; and they do. Some cover a single layer; others span them all to create a
cohesive system.
Current technology makes the best small computer network a hub-based peer-to-peer
design, cabled with twisted-pair wiring and running the software built in to your operating system. The big choice you face is the hardware standard. In the last few years, networks have converged on two basic hardware standards: 10Base-T and 100Base-T. Both
are specific implementations of Ethernet.
Just as celebrities are people famous principally for being famous, 10Base-T and 100BaseT are popular because they are popular. They are well known and generally understood.
Components for either are widely available and inexpensive. Setting them up is easy and
support is widely available.
The distinguishing characteristic of network hardware is the medium used for connecting
nodes. Small networksthe kind you might use in your home, office, or small business
most commonly use one of three interconnection types. They may be wired together in
the classic style, typically using twisted-pair wires in a star-based topology. They may be
linked wirelessly using low-powered radio systems, or their signals may be piggybacked on
an already existing wiring system, such as telephone lines or utility power lines.
Ethernet
The elder statesman of networking is Ethernet. It still reigns as king of the wires, and
wireless systems have appropriated much of its technology. It shows the extreme foresight
of the engineers at Xerox Corporations Palo Alto research center who developed it in the
1970s for linking the companys early Alto workstations to laser printers. The invention of
Ethernet is usually credited to Robert Metcalf, who later went on to found 3Com
Standards
Corporation, an early major supplier of computer networking hardware and software.

During its first years, Ethernet was proprietary to Xerox, a technology without a purpose,
in a world in which the personal computer had not yet been invented.
In September, 1980, however, Xerox joined with minicomputer maker Digital Equipment
Corporation and semiconductor manufacturer Intel Corporation to publish the first
Ethernet specification, which later became known as E.SPEC VER.1. The original specification was followed in November, 1982, by a revision that has become todays widely
used standard, E.SPEC VER.2.
This specification is not what most people call Ethernet, however. In January, 1985, the
Institute of Electrical and Electronic Engineers published a networking system derived
from Ethernet but not identical to it. The result was the IEEE 802.3 specification.
Ethernet and IEEE 802.3 share many characteristicsphysically, they use the same wiring
and connection schemesbut each uses its own packet structure. Consequently, although
you can plug host adapters for true Ethernet and IEEE 802.3 together in the same cabling
system, the two standards will not be able to talk to one another. No matter. No one uses
real Ethernet anymore. They use 802.3 instead and call it Ethernet.
Under the OSI Model, Ethernet provides the Data Link and Physical layersalthough
Ethernet splits the Data Link layer into two layers of its own: Media Access Control and
the Logic Link Control layer (or the MAC-client layer). The Ethernet specifications
define the cable, connections, and signals as well as the packet structure and control on
the physical medium. Table 12.2 shows the structure of an IEEE 802.3 packet.
TA B L E 1 2 . 2
Structure of an IEEE 802.3 (Ethernet) Packet
Function
Length
Comments
Preamble
Seven bytes
Each byte has the pattern 10101010.
Start frame delimiter
One byte
10101011 (for synchronization).
Destination address
Six bytes
Indicates where the packet is going.
Source address
Six bytes
Three-byte vendor ID, three-byte NIC ID.
Payload length
Two bytes
Number of bytes of data in the packet.
Data
As indicated above
Minimum length is 64 bytes; padded if shorter.
Frame check sequence
Four bytes
Error checking.
Data Link Layer

The basis of Ethernet is a clever scheme for arbitrating access to the central bus of the
system. The protocol, formally described as Carrier Sense, Multiple Access with Collision
Detection (CSMA/CD) is often described as being like a party line. Its not. Its much more
429
430
like polite conversation. All the computers in the network patiently listen to everything
thats going on across the network backbone. Only when there is a pause in the conversation will a new computer begin to speak. And if two or more computers start to talk at the
same time, all become quiet. They will wait for a random interval (and because it is random, each will wait a different interval) and, after the wait, attempt to begin speaking
again. One will be lucky and win access to the network. The other, unlucky computers
hear the first computer blabbing away and wait for another pause.
Access to the network line is not guaranteed in any period by the Ethernet protocol. The
laws of probability guide the system, and they dictate that eventually every device that
desires access will get it. Consequently, Ethernet is described as a probabilistic access system. As a practical matter, when few devices (compared to the bandwidth of the system)
attempt to use the Ethernet system, delays are minimal because all of them trying to talk
at one time is unlikely. As demand approaches the capacity of the system, however, the
efficiency of the probability-based protocol plummets. The size limit of an Ethernet system is not set by the number of computers but by the amount of traffic; the more packets
computers send, the more contention, and the more frustrated attempts.
Physical Layer
The Ethernet packet protocol has many physical embodiments. These can embrace just
about any topology, type of cable, or speed. The IEEE 802.3 specification defines several
of these, and it assigns a code name and specification to each. Among todays Ethernet
implementations, the most basic operates at a raw speed of 10MHz. That is, the clock frequency of the signals on the Ethernet (or IEEE 802.3) wire is 10MHz. Actual throughput
is lower because packets cannot occupy the full bandwidth of the Ethernet system.
Moreover, every packet contains formatting and address information that steals space that
could be used for data.
Originally, Ethernet used coaxial cables, and two versions of the 10MHz IEEE version of
Ethernet have proved popular: 10Base-5 (which uses thick coaxial cable, about one-half
inch in diameter) and 10Base-2 (which uses a thin coaxial cable, about 2/10th inch in
diameter).
Twisted-pair wiring is now used almost universally in Ethernet systems, except in special
applications (for example, to link hubs together or to extend the range of connections).
Basically the same kind of wire is used for speed that spans two orders of magnitude.
The base level for twisted-pair Ethernet is 10Base-T, which transfers the same signals as
the coaxial systems onto two pairs of ordinary copper wires. Each wire pair carries a
10MHz balanced signalone carrying data to the device and the other carrying it away
(in other words, separate receive and transit channels) for full-duplex operation.
The next increment up is 100Base-T, which is likely todays most popular wired networking standard. The 100Base-T system operates at 100MHz, yielding higher performance
Standards
consistent with transferring multimedia and other data-intensive applications across the
network. Its speed has made it the system of choice in most new installations.
During its gestation, 100Base-T wasnt a single system but rather a family of siblings, each
designed for different wiring environments. 100Base-TX is the purest implementation and
the most enduringbut its also the most demanding. It requires Class 5 wiring, shielded
twisted-pair designed for data applications. In return for the cost of the high-class wiring,
it permits full-duplex operation so that any network node can both send and receive data
simultaneously. The signals on the cable actually operate at 125MHz but use a five-bit
encoding scheme for every four bits. The extra code groups are used primarily for error
control. The current 100Base-T standard is formalized in the IEEE 802.3u specification.
To make the transition from 10Base-T to higher speeds easier, 100Base-T4 was designed
to work with shielded or unshielded voice-grade wiring, but it only allows for half-duplex
operations across four wire pairs. In addition, 100Base-T2 uses sophisticated data coding
to squeeze a 100Mbps data rate on ordinary voice-grade cables using one two-wire pair.
Currently neither of these transitional formats is in general use.
At the highest-speed end, Gigabit Ethernet moves data at a billion bits per second. The
first implementations used fiber optic media, but the most popular format is using the
same Category 5 twisted-pair wires as slower versions of Ethernet. Commonly called
1000Base-T and officially sanctioned under the IEEE 802.3ab standard, Gigabit
Ethernets high speed is not a direct upward mapping of 100MHz technology. Category 5
cables wont support a 1GHz speed. Instead, 1000Base-T uses all four pairs in a standard
Cat 5 cable, each one operating at 125MHz. To create a 1000Gbps data rate, the
1000Base-T adapter splits incoming data into four streams and then uses a five-level voltage coding scheme to encode two bits in every clock cycle. The data code requires only
four levels. The fifth is used for forward error correction. The standard allows both halfduplex (four pairs) and full-duplex (eight pairs using two Cat 5 cables) operation.
Wireless
The problem with wiring together a network is the wiring, and the best way to eliminate
the headaches of network wiring is to eliminate the wires themselves. True wireless networking systems do exactly that, substituting radio signals for the data line. Once an
exotic technology, wireless networking is now mainstream with standardized components
both readily available and affordable.
Wireless networking has won great favor, both in homes and businesses, because it allows
mobility. Cutting the cable means you can go anywhere, which makes wireless perfect for
notebook computers (and why many new notebooks come with built-in wireless networking). If you want to tote your computer around your house and work anywhere, wireless is
the only way to go. Many airports and businesses also use standard wireless connections,
431
432
so if you equip your portable computer with wireless for your home, you may be able to
use it when youre on the move.
Moreover, wireless networks are easier to install, at least at the hardware end. You slide a
PC Card into your notebook computer, plug the hub into a wall outlet, and youre connected. If youre afraid of things technical, that makes wireless your first choice. Of
course, the hardware technology does nothing to make software installation easier, but
wireless doesnt suffer any handicaps beyond all other network technologies.
On the other hand, wireless has a limited range that varies with the standard you use.
Government rules restrict the power that wireless network systems can transmit with.
Although equipment-makers quote ranges of 300 meters and more, that distance applies
only in the open airgreat if youre computing on the football field but misleading
should you want to be connected indoors. In practical application, a single wireless network hub might not cover an entire large home.
All wireless networks are also open to security intrusions. Once you put your data on the
air, it is open for inspection by anyone capable of receiving the signals. Encryption, if
used, will keep your data reasonably secret (although researchers have demonstrated that
resourceful snoops can break the code in an hour). More worrisome is that others can tap
into your network and take advantage of your Internet connection. When you block them,
you sacrifice some of the convenience of wireless.
Wireless networks endured a long teething period during which they suffered a nearly
fatal malaisea lack of standardization. Wireless systems were, at one time, all proprietary. But wireless technology proved so useful that the IEEE standardized it and keeps on
adding new standards. Two are currently in wide deployment, and several others are nearing approval. Table 12.3 summarizes the current standards situation.
TA B L E 1 2 . 3
Wireless Networking Standards
Standard
Common Name
Frequency
Data Rate
Status
802.11a
WiFi-5
5.2GHz
54Mbps
Official
802.11b
WiFi
2.4GHz
11Mbps
Official
802.11b+
2.4GHz
22Mbps
Proprietary/Pending
802.11g
2.4GHz
54Mbps
Pending
The following discussion covers todays wireless networking standards and those likely to
become standards in order of adoption or introduction.
802.11b
Through the year 2001, IEEE 802.11b was the most popular wireless standard on the
market, popularized under the name Wireless Fidelity (WiFi), given to it by the industry
Standards
group, the Wireless Ethernet Compatibility Alliance. That group now uses the name
WiFi Alliance. WiFi equipment is also sometimes called 2.4MHz wireless networking
because of the frequency band in which it operates.
Until October 2001, WiFi was the only official wireless networking standard. Despite its
b designation, 802.11b was the first official wireless networking standard to be marketed. Its destiny was set from its conception. It was the one IEEE wireless standard with
acceptable performance that could readily be built.
The range of 802.11b is usually given as up to 300 metersnearly 1000 feeta claim akin
to saying that youre up to 150 feet tall. Anything that gets in the waya wall, a drapery,
or a catcuts down that range. Moreover, the figure is mostly theoretical. Actual equipment varies widely in its reach. With some gear (such as the bargain models I chose),
youll be lucky to reach the full length of your house.
Moreover, range isnt what you might think. 802.11b is designed to gracefully degrade.
What that engineering nonsense means is that the farther away you get, the slower your
connection becomes. When youre near the limits of your system, youll have a 2Mbps
connection.
Note that the range is set by the least powerful transmitter and weakest receiver. In most
cases, the least powerful transmitter is in the PC Card adapter in your notebook computer. Power in these is often purposely kept low to maintain battery life and to stay
within the engineering limits of the card housing. You can often achieve longer distances
with an external adapter, such as one that plugs into your computers USB port.
802.11b is less than ideal for a home wireless network for more reasons than speed.
Interference is chief among them. The 2.4GHz band within which 802.11b works is one
of the most popular among engineers and designers. In addition to network equipment, it
hosts cordless telephones and the Bluetooth interconnection system. Worse, it is close to
the frequency used by consumer microwave ovens. Because the band is unlicensed, you
can never be sure what youll be sharing it with. Although 802.11b uses spread-spectrum
modulation, which makes it relatively immune to interference, you can still lose network
connections when you switch on your cordless phone.
802.11a
The place to start with 802.11a is with the name. Back-stepping one notch in the alphabet
doesnt mean moving down the technological ladder, as it does with software revisions.
When the standards-makers at the IEEE put pen to paper for wireless standards, 802.11a
did come first, but it was the dream before the reality. The high speed of 802.11a was
what they wanted, but they amended their views to accommodate its slower sibling, which
they knew they could deliver faster. When manufacturers introduced equipment following
the 802.11a standard, they initially promoted it as WiFi-5 because of the similarity with
the earlier technology. The 5 designated the higher frequency band used by the newer
433
434
system. In October, 2002, however, the WiFi Alliance decided to drop the WiFi-5 term
mostly because people wondered whatever happened to WiFi-2, 3, and 4.
From the beginning, 802.11a was meant by its IEEE devisors as the big brother to
802.11b. It is designed to be faster and provide more channels to support a greater number of simultaneous users. The bit-rate of the system is 54Mbps, although in practical
application actual throughput tops out at about half that rate. In addition, as the distance
between the computers and access points increase, the 802.11a system is designed to
degrade in speed to ensure data integrity, slowing down to about 6Mbps at the farthest
reaches of its range.
802.11a fits into the UN-II (Unlicensed National Information Infrastructure) band thats
centered at about 5.2GHz, a band thats currently little used by other equipment and services. The design of 802.11a provides for eight channels, allowing that number of access
points to overlap in coverage without interference problems.
The higher frequency band has a big advantagebandwidth. Theres more room for multiple channels, which means more equipment can operate in a given space without interference.
But the UN-II band occupied by 802.11a isnt without its problems. The biggest is
physics. All engineering rules of thumb say the higher the frequency, the shorter the
range. The 5.2GHz operating frequency puts 802.11a at a severe disadvantage, at least in
theory.
Practically, however, the higher intrinsic speed of 802.11a not only wipes out the disadvantage but also puts the faster standard ahead. Because of the digital technology used by
the common wireless networking systems, weaker signals mean lower speed rather than an
abrupt loss of signal. The farther away from the access point a user ventures, the lower
the speed his connection is likely to support.
Although physics doesnt favor 802.11as signals, mathematics does. Because 802.11a starts
out faster, it maintains a performance edge no matter the distance. For example, push an
802.11b link to 50 feet, and you cant possibly get it to operate faster than 11Mbps.
Although that range causes some degradation at 802.11as higher frequency, under that
standard you can expect about 36Mbps, according to one manufacturer of the equipment.
Raw distance is a red herring in figuring how many people can connect to a wireless system. When the reach of a single access point is insufficient, the solution is to add another
access point rather than to try to push up the range. Networks with multiple access points
can serve any size campus. The eight channels available under 802.11a make adding more
access points easier.
The security weakness is part of the plan. 802.11a is designed to match 802.11b at the
software level.
Standards
802.11b+
There is no such thing as an 802.11b+ standard, at least a standard adopted by some official-sounding organization such as the IEEE. But youll find a wealth of networking gear
available that uses the technology thats masquerading under the name. Meant to fill the
gap between slow-but-cheap 802.11b and faster-but-pricier 802.11a, this technology adds
a different modulation system to ordinary 802.11b equipment to gain double the speed
and about 30 percent more range. Because the technology is, at heart, 802.11b, building
dual-mode equipment is a breeze, and most manufacturers do it. As a result, you get two
speeds for the price of one (often less, because the 802.11b+ chipset is cheaper than other
802.11b chipsets).
The heart of 802.11b+ is the packet binary convolutionary coding system developed by Texas
Instruments (TI) under the trademark PBCC-22. (A convolutionary code encodes a
stream of data in which the symbol encoding a given bit depends on previous bits.) The
22 in the name refers to the speed. The only previous version was PBCC-11, the forerunner of PBCC-22 and also developed by TI.
PBCC-22 is a more efficient coding system than that used by ordinary 802.11b. In addition, the TI system uses eight-phase shift keying instead of 802.11b quadrature phase shift
keying, but at the same data rate (11Mbps). The result is that the system fits twice the
data in the same bandwidth as 802.11b. At the same time, the signal stands out better
from background noise, which, according to TI, produces 30 percent more linear range or
coverage of 70 percent more area. The same technology could be used to increase the
speed of the 802.11b signaling system to 33MHz, although without benefit of the range
increase. This aspect of PBCC-22 is not part of the TI chipset now offered for 802.1b+
components and has not received FCC approval.
802.11b+ can be considered a superset of 802.11b. Adapters and hubs meant for the new
system also work with all ordinary 802.11b equipment, switching to their high speed when
they find a mate operating at the higher speed.
Texas Instruments makes the only chipset to use 802.11b+ technology, so all equipment
using the technology uses the TI chipset. TI often prices this chipset below the price of
ordinary 802.11b chipsets, so the higher speed of 802.11b+ often is less expensive than the
performance of equipment strictly adhering to the official 802.11b standard.
802.11g
As this is written, 802.11g is a proposal rather than an official standard, although its
approval is expected in 2003. In effect a greatest hits standard, 802.11g combines the
frequency band used by 802.11b with the modulation and speed of 802.11a, putting 54Mb
operation in the 2.4GHz Industrial-Scientific-Medical (ISM) band.
Because of the lower frequency of the ISM band, 802.11g automatically gets an edge in
range over 802.11a. At any given speed, 802.11g nearly doubles the coverage of 802.11a.
435
436
In addition, promoters cite that the ISM band is subject to fewer restrictions throughout
the world than the 5GHz band, part of which is parceled out for military use in many
countries. Of course, the opposite side of the coin is that the ISM band is more in demand
and subject to greater interference. It also does not allow for as many channels as the
5GHz band.
Many of its promoters see 802.11g as a stepping stone to 802.11a. It allows engineers to
create dual-speed equipment more easily because the two speeds (11Mbps and 54Mbps)
will be able to share the same radio circuits.
Dual-Standard Systems
Nothing restricts a network or its equipment to using only one standard. To ease people
who have already installed WiFi equipment into the higher-performance WiFi-5 standard,
many manufacturers offer dual-standard equipment, including hubs that can recognize
and use either standard to match the signals it receives. They also offer dual-standard network adapters capable of linking at the highest speed to whatever wireless signals are
available. Nothing in the WiFi and WiFi-5 standards requires this interoperability.
Other high-speed standards do, however, guarantee interoperability. The proprietary
802.11b+ system is built around and is compatible with standard WiFi. Equipment made
to one standard will operate with equipment made to the other, at the highest speed both
devices support.
Similarly, 802.11g promises interoperability. In fact, at the time this is written, 802.11b+
has been proposed as an addition to 802.11g, making the proprietary standard official.
The various 802.11 standards are promulgated by the IEEE but are promoted by the
WiFi Alliance. The organization maintains a Web site at www.weca.net.
HomePNA
One of the original design goals of the 10Base-T wiring scheme was for it to use the same
kind of wires as used by ordinary telephones. Naturally, some people thought it would be
a great idea to use the same wires as their telephones, but they were unwilling to give up
the use of their phones when they wanted to network their computers.
Clever engineers, however, realized that normal voice conversations use only a tiny part of
the bandwidth of telephone wiringand voices use an entirely different part of that bandwidth than do computer networks. By selectively blocking the network signals from telephones and the telephone signals from networking components, they could put a network
on the regular telephone wiring in a home or office while still using the same wiring for
ordinary telephones.
Standards
Of course, theres a big difference between wiring a network and wiring a telephone system. A network runs point to point, from a hub to a network adapter. A telephone system
runs every which way, connecting telephones through branches and loops and often leaving stubs of wire trailing off into nowhere. Telephone wiring is like an obstacle course to
network signals.
History
Tut Systems devised a way to make it all work by developing the necessary signal-blocking
adapters and reducing the data rate to about a megabit per second, to cope with the poor
quality of telephone wiring. Tut Systems called this system HomeRun.
In June, 1998, eleven companies, including Tut (the others being 3Com, AMD, AT&T
Wireless, Broadcom, Compaq, Conexant, Hewlett-Packard, IBM, Intel, and Lucent
Technologies) formed the Home Phoneline Networking Alliance (HomePNA) to promote
telephone-line networking. The alliance chose HomeRun as the foundation for its first
standard, HomePNA version 1.0
In September, 2000, the alliance (which had grown to over 150 companies) kicked up the
data rate to about 10Mbps with a new HomePNA versionVersion 2.0.
On November 12, 2001, HomePNA released market (not technical) specifications for the
next generation of telephone-line networking, HomePNA Version 3.0, with a target of
100Mbps speed. The new standard will be backward compatible with HomePNA 1.0 and
2.0 and wont interfere with normal and advanced telephone services, including POTS,
ISDN, and xDSL. A new feature, Voice-over-HomePNA, will allow you to send up to
eight high-quality telephone conversations through a single pair of ordinary telephone
wires within your home. According to the association, the HomePNA Version 3.0 specification will be released by the end of 2002 and was not available as this is being written.
Version 1.0
Version 1.0 is a fairly straightforward adaptation of standard Ethernet, with only the data
rate slowed to accommodate the vagaries of the unpredictable telephone wiring environment. The system does not adapt in any way to the telephone line. It provides its standard
signal and thats that. Low speed is the only assurance of signal integrity. When a packet
doesnt get through, the sending device simply sends it again. The whole system is therefore simple, straightforward, and slow.
The design of Version 1.0 allows for modest-sized networks. Its addressing protocol
allows for a maximum of 25 nodes networked together. The signals system has a range of
about 150 meters (almost 500 feet), although its range will vary with the actual installationsome telephone systems are more complex (and less forgiving) than others.
437
438
Version 2.0
Although HomePNA Version 1.0 is capable of most routine business chores, such as sharing a printer or DSL connection, it lacks the performance required for the audio and
video applications that are becoming popular in home networking systems. Version 2.0
combines several technologies to achieve the necessary speed. It has two designated signaling rates: 2 million baud and 4 million baud (thats 2MHz and 4MHz, respectively).
The signal uses quadrature amplitude modulation (the same technology of 28.8Kbps
modems), but it can use several coding schemes, from 4 to 256 symbols per baud. As a
result, the standard allows for actual data rates from 4Mbps (2 Mbaud carrier with 4 QAM
modulation) to 32Mbps (4 Mbaud at 256 QAM modulation). The devices negotiate the
actual speed.
Compared to ordinary 10Base-T Ethernet systems, HomePNA Version 2.0 suffers
severely from transmission overhead, cutting its top speed well below its potential. To
ensure reliability, HomePNA uses its highest speeds only for the data part of its packets (if
at all). The system sends the header and tailer of each data frame, 84 bytes of each frame,
at the lowest rate (4Mbps) to ensure that control information is error free. As a consequence, in every frame, overhead eats almost one-third of the potential bandwidth at the
32Mbps data rate, about 80 percent at the 16Mbps data rate, and correspondingly less at
lower rates.
Equipment
HomePNA is a peer-to-peer networking system. It does not use a special hub or server.
Each computer on the network plugs into a HomePNA adapter (usually through a USB
port), which then plugs into a telephone wall jack. If you do not have a wall jack near your
computer, you can use extension telephone cables to lengthen the reach of the adapter.
No other hardware is required. Once all your computers are plugged in, you set up the
network software as you would for any wired network.
The HomePNA specification is maintained by the Home Phoneline Networking Alliance.
The group maintains a Web site at www.homepna.org.
HomePlug
The underlying assumption of the HomePlug networking system is that if you want to
network computers in your home, you likely already have power wiring installed. If not,
you dont likely need a network because you dont have any electricity to run your computers. Those big, fat, juicy electrical wires whisking dozens of amps of house current
throughout your house can easily carry an extra milliwatt or two of network signal. In that
the power lines inside your walls run to every room of your house, you should be able to
plug your network adapter into an outlet and transport the signals everywhere you need
them.
Standards
Youll want to be careful if you try such a thing. Unlike computer signals, house current
can kill youand the circuitry of your network adapter. House current is entirely different stuff from network signals. Thats bad if you stick your fingers into a light socket, but
its good if you want to piggyback network signals on power lines. The big difference
between the two means that its relatively easy for a special adapter (not an ordinary network adapter!) to separate the power from the data.
On the other hand, although those heavy copper wires are adept at conducting high-current power throughout your home, they are a morass for multimegahertz data signals. At
the frequencies needed for practical network operation, home wiring is literally a maze,
with multiple paths, most of which lead into dead-ends. Some frequencies can squeak
through the maze unscathed, whereas others disappear entirely, never to be seen again.
Motors, fluorescent lights, even fish-tank pumps add noise and interference to your power
lines, not to mention stuff that sneaks in from the outside, such as bursts of lightning.
Radio frequency signals like those a network uses bounce through this booby-trapped
electrical maze like a hand grenade caught in a pinball machine.
Transmission Method
The HomePlug solution is to try, try again. Rather than using a single technology to bust
through the maze, HomePlug uses several. It is adaptive. It tests the wiring between network nodes and checks what sort of signals will work best to transport data between them.
It then switches to those signals to make the transfer. Just in case, everything that
HomePlug carries is doused with a good measure of error correction. The result is a single networking standard built using several technologies that seamlessly and invisibly
mate, so you can treat your home wiring as a robust data communications channel.
The basic transmission technique in the HomePlug system is called orthogonal frequency
division multiplexing (OFDM), similar to that used in some telephone Digital Subscriber
Lines. The OFDM system splits the incoming data stream into several with lower bit
rates. Each stream then gets its own carrier to modulate, with the set of carriers carefully
chosen so that the spacing between them is inversely proportional to the bit-rate of each
carrier (this is the principle of orthogonality). Combining the modulated carriers of different frequencies together results in the frequency division multiplexing. In the resulting
signal, it is easy to separate out the data on each carrier because the data on one carrier is
easily mathematically isolated from the others.
Each isolated carrier can use a separate modulation scheme. Under HomePlug, differential
quarternary phase shift keying (DQPSK) is the preferred method. It differs from conventional QPSK in that the change in the phase of the carrier, not the absolute phase of the
carrier, encodes the data. Alternately, the system can shift to differential binary phase shift
keying (DBPSK) for lower speed but greater reliability when the connection is poor.
DQPSK shifts between four phases, whereas DBPSK shifts between two. The data
439
440
streams use a convolutionary code as forward error correction. That is, they add extra bits
to the digital code, which help the receiver sort out errors.
In the ideal case, with all carriers operating at top speed, the HomePlug system delivers a
raw data rate of about 20Mbps. Once all the overhead from data preparation and error
correction gets stripped out, it yields a clear communication channel for carrying data
packets with an effective bit-rate of about 14Mbps.
If two devices cannot communicate with the full potential bandwidth of the HomePlug
system, they can negotiate slower transmissions. To best match line conditions, they can
selectively drop out error-plagued carrier frequencies, shift from DQPSK to more reliable
DBPSK modulation, or increase the error correction applied to the data.
The HomePlug system uses a special robust mode for setting up communication between
channels. Robust mode activates all channels but uses DBPSK modulation and heavy
error correction. Using all channels ensures that all devices can hear the signals, and the
error correction ensures messages get through, if slowly. Once devices have negotiated the
connection starting with robust mode, they can shift to higher-speed transmissions.
The HomePlug system selects its carriers in the range between frequencies of 4.5 and
21MHz, operating with reduced power in the bands in which it might cause interference
with amateur radio transmissions. In addition, HomePlug transmits in bursts rather than
maintaining a constant signal on the line.
Protocol
The HomePlug system moves packets between nodes much like the Ethernet system with
a Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) protocol. That is, each
device listens to the power line and starts to transmit only when it does not hear another
device transmitting. If two devices accidentally start transmitting at the same time (resulting in a collision), they immediately stop, and each waits a random time before trying
again. HomePlug modifies the Ethernet strategy to add priority classes, greater fairness,
and faster access.
All packets sent from one device to another must be acknowledged, thus ensuring they
have been properly received. If a packet is not acknowledged, the transmitting device acts
as if its packet had collided with another, waiting a random period before retransmitting.
An acknowledgement sent back also signals to other devices that the transmission is complete and that they can attempt to send their messages.
According to the HomePlug alliance, the protocol used by the system results in little
enough delay that the system can handle time-critical applications such as streaming
media and Voice-over-Internet telephone conversations.
To prevent your neighbors from listening in on your network transmissions or from
sneaking into your network and stealing Internet service, the HomePlug system uses a
Hardware
56-bit encryption algorithm. The code is applied to packets at the Protocol layer. All
devices in a given networkthat is, the HomePlug adapters for all of your computers
use the same key, so all can communicate. Intruders wont know the key and wont be recognized by the network.
With HomePlug, such intrusions are more than theoretical. Anyone sharing the same
utility transformer with youtypically about six households tap into the same utility
transformerwill share signals on the power line. Your HomePlug networking signals will
consequently course through part of your neighborhood. Thats bad if you want to keep
secrets, but it could allow you and a neighbor share a single high-speed Internet connection (if your Internet Service Provider allows you to).
Equipment
HomePlug is a peer-to-peer networking system. It does not use a special hub or server.
Each computer on the network plugs into a HomePlug adapter (usually through a USB
port), which then plugs into a wall outlet. No other hardware is required. Once all your
computers are plugged in, you set up the network software as you would for any wired
network.
The specifications for the HomePlug networking system are maintained by the
HomePlug Powerline Alliance.
HomePlug Powerline Alliance, Inc.
2694 Bishop Drive, Suite 275
San Ramon, CA 94583
Homeplug.org
Phone: 925-275-6630
Fax: 925-275-6691
Hardware
To make a network operate under any standard requires actual hardwarepieces of equipment that plug into each other to make the physical network. Most of todays networking
systems require three different kinds of hardware. Network interface cards link your computer to the network wiring system (or wireless signaling system). A hub or access point
brings the signals from all the computers and other devices in the network together. And a
medium (such as wires or radio waves) links these two together. You build a network using
these three components from one network standard.
441
442
Network Interface Cards

The network interface card (NIC) and its associated driver software have the most challenging job in the network. The NIC takes the raw data from your computer, converts it into
the proper format for the network youre using, and then converts the electrical signals to
a format compatible with the rest of the network. If it doesnt work properly, your data
would be forever trapped in your computer.
Because it converts the stream of data between two different worlds (inside your computer
and the outside world of the network), it needs to match two standards: a connection system used by your computer and the connection system used by the network.
Traditionally, a network adapter has plugged into the expansion bus of the host computer.
No other connection could provide the speed required for network use. Desktop computers used internal NICs that slid into expansion slots. Portable computers used NICs in the
form of PC Cards.
With the advent of USB, however, that situation has changed. Although expansion
busbased NICs remain important, many networks are shifting to NICs that plug into
USB ports. The primary reason for this is that USB-based NICs are easier to install. Plug
in one connector, and youre finished. Although early USB-based NICs gave up speed for
this convenience, a USB 2.0 port can handle the data rates required even by 100Base-T
networking.
The simplified piggyback networking systems HomePlug and HomePNA rely on USBbased NICs for easy installation. In wireless networking, USB-based NICs often have
greater range than their PC Cardbased peers because they can operate at higher power
and with more sophisticated antennae. Compared to installing a PC Card adapter in your
desktop computer and then sliding a wireless NIC into the adapter, the USB-based NIC
makes more sense, both for the added range as well as the greater convenience and lower
cost.
The NIC you choose must match the standard and speed used by the rest of your network. The exception is, of course, that many NICs operate at more than one speed. For
example, nearly all of todays 100Base-T NICs slow down to accommodate 10Base-T networks. Dual-speed 802.11b+ wireless NICs will accommodate either 11 or 22MHz operation.
Most dual-speed NICs are autosensing. That is, they detect the speed on the network wire
to which they are connected and adjust their own operating speed to match. Autosensing
makes a NIC easier to set up, particularly if you dont know (or care) the speed at which
your network operates. You can just plug in the network wire and let your hardware worry
about the details.
Hardware
Some network adapters allow for optional boot ROMs, which allow computers to boot up
using a remote disk drive. However, this feature is more applicable to larger businesses
with dedicated network servers rather than a home or small business network.
Hubs and Access Points

A network hub passes signals from one computer to the next. A network access point is the
radio base station for a wireless network. Piggyback networks such as HomePlug and
HomePNA do not use dedicated hubs, although one of the computers connected to the
network acts like a hub.
Hubs
The most basic hub for a wired network has two functions. It provides a place to plug in
the network wire from each computer, and it regenerates the signals to ensure against
errors.
The design of Ethernet requires all the signals in the network loop to be shared. Every
computer in the loopthat is, every computer connected to a single hubsees exactly the
same signals. The easiest way to do this would be to short all the wires together.
Electrically, such a connection would be an anathema.
The circuitry of the hub prevents such disasters. It mixes all the signals it receives
together and then sends out the final mixture in its output.
To make the cabling for the system easy, the jacks on a hub are wired the opposite of the
jacks on NICs. That is, the send and receive connections are reservedthe connections
the NIC uses for sending, the hub uses for receiving.
Inexpensive hubs are distinguished primarily by the number and nature of the ports they
offer. You need one port on your hub for each computer or other device (such as a network printer or DSL router) in your network. You may want to have a few extra ports to
allow for growth, but more ports cost more, too. Some hubs include a crossover jack or
coaxial connection that serves as an uplink to tie additional hubs into your network.
Expensive hubs differ from the economic models chiefly by their management capabilitiesthings such as remote monitoring and reconfiguration, which are mostly irrelevant
to a small network.
As with NICs, hubs can operate at multiple speeds. Although some require each network
connection to be individually configured for a single speed (some are even physically set
only at one speed), most modern dual-speed hubs are autosensing, much like NICs.
Coupling a low-cost autosensing NIC and low-cost autosensing hub can sometimes be
problematic with marginal wiring. The hubs sense may negotiate the highest-speed signals on the network, even though many packets fail to negotiate the wiring at higher
speeds. In such instances, performance can fall below that of a low-speed-only network.
443
444
Some hubs are called switches; others have internal switches. A switch divides a network
into segments and shifts data between them. Most dual-speed hubs actually act as switches
when passing packets between 10Base-T and 100Base-T devices.
When used on a single-speed network, a switch can provide faster connections because
the full bandwidth of the network gets switched to service each NIC. When several NICs
contend for time on the network, however, the switch must arbitrate between them, and
in the end, they share time as they would on a hub-based network.
Switches usually have internal buffers to make it easier to convert between speeds and
arbitrate between NICs. In general, the larger the buffer the better, because a large buffer
helps the switch operate at a higher speed more of the time.
Access Points
Access points earn their name because they provide access to the airwaves. Although in
theory you could have a wireless-only hub to link together a small network of computers,
such a network would shortchange you. You could not access the Internet or any devices
that have built-in wired-networking adapters, such a network printers. In other words,
most wireless networks also require a wired connection. The access point provides access
to the airwaves for those wired connections.
The centerpiece of the access point is a combination of radio transmitter and receiver
(which engineers often talk about simply as the radio). Any adjustment you might possibly
make to the radio is hidden from you, so you dont have to be a radio engineer to use one.
The access points radio is both functionally and physically a sealed box, permanently set
to the frequencies used by the standard used by the networking system supported by the
access point. Even the power at which the access point operates is fixed. All access points
use about the same power because the maximum limit is set by government regulations.
Access points differ in several ways, however. Because every access point has a wired connection, they differ just as wired network dothey may support different speed standards.
In fact, most access points have multiple wired connections, so they are effectively wired
hubs, too. The same considerations that apply to wired hubs therefore apply to the wired
circuitry of access points.
Access points also differ in the interface between their radios and the airwavesnamely,
their antennae. Lower-cost access points have a single, permanently attached antenna.
More expensive hubs may have two or (rarely) more antennae. By using diversity techniques, multiple-antenna access points can select which of the antennae has the better signal and use it for individual transmissions. Consequently, dual-antennae access points
often give better coverage.
Some access points make their antennae removable. You can replace a removable antenna
with a different one. Some manufacturers offer directional antennae, which increase the
Hardware
range of an access point in one direction by sacrificing the range in all other directions. In
other words, they concentrate the radio energy in one direction. Directional antennae are
particularly useful in making an access point into a radio link between two nearby buildings.
Routers
When you want to link more than one network togetherfor example, you want to have
both wired and wireless networks in your officethe device that provides the necessary
bridge is called a router because it gives the network signals a route from one system to
another. When you want to link a home network to the Internet through a DSL line, you
need a router. Routers are more complex than hubs because they must match not only the
physical medium but also the protocols used across the different media.
Sometimes what you buy as a hub actually functions as a router. Most wireless access
points include 10Base-T or 100Base-T connections to link with a wired network. Many
DSL adapters are sold as hubs and have multiple network connections (or even a wireless
access point), even though they are, at heart, routers.
Cabling
Network wires must carry what are essentially radio signals between NICs and hubs, and
they have to do this without letting the signals actually become radio wavesbroadcasting
as interferenceor finding and mixing with companion radio waves, thus creating errorcausing noise on the network line. Two strategies are commonly used to combat noise and
interference in network wiring: shielding the signals with coaxial cable and preventing the
problem with balanced signaling on twisted-pair wiring.
Coaxial Cables
Coaxial cables get their name because they all have a central conductor surrounded by one
or more shields that may be a continuous braid or metalized plastic film. Each shield
amounts to a long, thin tube, and each shares the same longitudinal axisthe central conductor. The surrounding shield typically operates at ground potential, which shields stray
signals from leaking out of the central conductor or prevents noise from seeping in.
Figure 12.6 shows the construction of a simple coaxial cable.
Coaxial cables are tuned as transmission linessignals propagate down the wire and are
completely absorbed by circuitry at the other endwhich allows them to carry signals for
long distances without degradation. For this reason, they are often used in networks for
connecting hubs, which may be widely separated.
Twisted-Pair Wires
The primary alternative is twisted-pair wiring, which earns its name from being made of
two identical insulated conducting wires that are twisted around one another in a loose
445
446
double-helix. The most common form of twisted-pair wiring lacks the shield of coaxial
cable and is often denoted by the acronym UTP, which stands unshielded twisted pair.
Figure 12.7 shows a simplified twisted-pair cable.
FIGURE 12.6
Plastic jacket
Components of a
coaxial cable.
Foam insulation
Central conductor
Braided shield
FIGURE 12.7
Overall plastic jacket
Components of a
twisted-pair wiring
cable.
Color-coded wires
Conductors twisted in pairs
Multiple pairs per cable
Manufacturers rate their twisted-pair cables by category, which defines the speed of the
network they are capable of carrying. Currently standards define six categories: 1 through
5 and 5e (which stands for 5 enhanced). Two higher levels, 6 and 7, exist on the periphery
(proposed but not official standards). As yet, they are unnecessary because Gigabit
Ethernet operates with Category 5 wiring. Note that, strictly speaking, until the standards
for Levels 6 and 7 get approved, wires of these quality levels are not termed Category 6
and Category 7, because the categories are not yet defined. Instead, they are termed Level 6
and Level 7.
Those in the know dont bother with the full word Category. They abbreviate it Cat, so
look for Cat 5 or Cat 5e wire when you shop. Table 12.4 lists the standards and applications of wiring categories.
TA B L E 1 2 . 4
Standard Wire Categories
Category
Frequency Rating
Typical Application
None
Telephone and general wiring
Low-speed (1MHz) data
10
10Base-T
20
16Mbps token-ring networks
100
100Base-T networks
5e
100
100Base-T networks
Hardware
Most UTP wiring is installed in the form of multipair cables with up to several hundred
pairs inside a single plastic sheath. The most common varieties have 4 to 25 twisted pairs
in a single cable. Standard Cat 5 wire for networking has four pairs, two of which are used
by most 10Base-T and 100Base-T systems.
The pairs inside the cable are distinguished from one another by color codes. The body of
the wiring is one color alternating with a thinner band of another color. In the two wires
of a given pair, the background and banding color are oppositesthat is, one wire will
have a white background with a blue band, and its mate will have a blue background with
a white band. Each pair has a different color code.
To minimize radiation and interference, most systems that are based on UTP use differential signals. For extra protection, some twisted-pair wiring is available with shielding. As
with coaxial cable, the shielding prevents interference from getting to the signal conductors.
The specifications for most UTP networking systems limit the separation between any
NIC and hub to no more than 100 meters (about 325 feet). Longer runs require repeaters
or some other cabling system.
447
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
13
Telecommunications
Until the invention of the telegraph, no reliable communications traveled long distance faster than via a human letter carrier. That didnt stop
people from trying, so in the pre-telegraph days a lot of exciting technologies promised to revolutionize communicationssmoke signals,
semaphore flags, and carrier pigeons were all attempts to break the bond
between messenger and message.
The telegraph was able to let the messenger rest his feet and put pigeons
back where they belong, atop the heads of dignitaries statues in the
park. With the advent of the telegraph, messages finally were able to
travel at the ultimate speed limit of the universethe speed of light.
Well, not quite. The telegraph only changed the speed of the messenger.
It let the pony express rider exchange his steed for Pegasus, but the content of those messages never reached the speed of light. They were also
slowed by the interface technologythe telegraph key in the hands of a
human being. With fast fingers and a good ear, a 19th-Century telegrapher might squeeze ten words a minute down the wire, equivalent to a
bit rate of less than ten per second. At that rate, sending a snapshot
home from summer camp might take a month.
Since the advent of the telegraph, technology has concentrated on this
other side of the speed issue, making the message move through the data
channel faster. Using essentially the same wires as Samuel Morse, you
can move data roughly 100,000 times faster through the telephone system and ten million times faster through your home network.
450
Getting there isnt always easy. The analog nature of your telephone line challenges data
signals and imposes strict limits on how fast information can move through the system.
The telephone line is an arcane world of strange technologies, all meant to get more
speed from the simple telephone wire. It has been a long, hard battle, but one that has
paid off. Todays modems now race ahead of the theoretical speed of which information
can move through a telephone line. (They can do that because theyre no longer really
modems, but were getting ahead of ourselves.)
You can do better. Digital services promise to deliver data 20 times faster than the best
modem. Although people arent switching over to the new, high-speed connections anywhere near as fast as once predicted (thats one of the big reasons the telecommunications
industry crashed in 2002), you can be sure someday you will have an all-digital connection.
Analog Services
A modem is standard equipment in every new computerand with good reason. Its what
most people use to connect to the Internet. Consider a computer without a modem, and
you might as well buy a vacuum cleaner.
Todays modem is a world apart from those of only a few years ago, and the difference is
speed. Nearly every new modem sold operates at a top speed of 56,000bpsfast enough
that most people refuse to pay the higher charges demanded for faster-still all-digital services. Todays modems are fast enough for e-mail and most surfing. They falter only when
you want an edge in playing online games, want to download streaming video, or want to
exchange photos with relatives and friends.
Background
A true modem is a necessary evil in todays world of telecommunications because we still
suffer from a telephone system that labors under standards devised even before electronics
were invented, at a time when solid-state digital circuitry lay undreamed, almost a hundred years off. The telephone system was designed to handle analog signals only because
thats all that speaking into a microphone creates. Over the years, the telephone system
has evolved into an elaborate international network capable of handling millions of these
analog signals simultaneously and switching them from one telephone set to another, anywhere in the world.
In the last couple decades, telephone companies have shifted nearly all their circuits to
digital. Most central office circuitry is digital. Nearly every long distance call is sent
between cities and countries digitally. In fact, the only analog part of most telephone connections is the local loop, the wires that reach out from the telephone exchange to your
CHAPTER 13 TELECOMMUNICATIONS
Analog Services
home or office (and likewise extend from a distant exchange to the telephone of
whomever youre calling).
The chief reason any analog circuitry remains in the telephone system is that there are
hundreds of millions of plain-old telephone sets (which the technologically astute call simply POTS) dangling on the ends of telephone wires across the country. As long as people
depend on POTS, they need an analog network to connect to.
Modem Technology
Of all modern communications systems, the one with the most severe limits on bandwidth
is the telephone channel. Instead of the full frequency range of a good-quality stereo system (from 20 to 20,000Hz), a telephone channel only allows frequencies between 300 and
3000Hz to freely pass. This very narrow bandwidth works well for telephones because frequencies below 300Hz contain most of the power of the human voice but little of its intelligibility. Frequencies above 3000Hz increase the crispness of the sound but dont add
appreciably to intelligibility.
What works well for voice is horrible for data. Although intelligibility is the primary concern with voice communications (most of the time), data transfer is principally oriented to
bandwidth. The comparatively narrow bandwidth of the standard telephone channel limits
the bandwidth of the modulated signal it can carry, which in turn limits the amount of
digital information that can be squeezed down the phone line by a modem. The result is
that dial-up telephone lines are poor data channelsbut too often they are the only ones
we have.
Shannons Limit
Engineers can use a variety of technologies to squeeze more data into a narrow-bandwidth
channel, but they still face an ultimate limit on the amount of data they can squeeze
through a tight channel such as an analog telephone line. This ultimate limit combines
the effects of the bandwidth of the channel and the noise level in the channel. The greater
the noise, the more likely it will be confused with the information that has to compete
with it. This theoretical maximum data rate for a communication channel is called
Shannons Limit. This fundamental law of data communications, discovered by Claude L.
Shannon working at Bell Labs in 1948, states that the maximum number of digital bits
that can be transmitted over a given communication path in one second can be determined from the bandwidth (W) and signal-to-noise ratio (S/N, expressed in decibels) by
using the following formula:
Maximum data rate = W log (1 + S/N)
In telephone circuits, the analog-to-digital converters used in telephone company central
offices contribute the noise that most limits modem bandwidth. In creating the digital
451
452
pulse-coded modulation (PCM) signal, they create quantization distortion, which produces
an effective signal-to-noise ratio of about 36 dB. Quantization distortion results from the
inability of the digital system with a discrete number of voltage steps (256 in the case of
telephone company A/D converters) to exactly represent an analog signal that has an infinite number of levels. At this noise level, Shannons Limit for analog data on telephone
lines is about 33,600 bits per second. Modern modems for dial-up data communications
can be faster only by sidestepping some of the bandwidth issues.
Channels
Communications are supposed to be a two-way street. Information is supposed to flow in
both directions. You should learn something from everyone you talk to, and everyone
should learn from you. Even if you disregard the potential for success of such two-way
communication, one effect is undeniable: It cuts the usable bandwidth of a data communication channel in one direction in half because the data going the other way requires its
own share of the bandwidth.
Duplex
With modems, such a two-way exchange of information is called duplex communications.
Often it is redundantly called full-duplex. A full-duplex modem is able to simultaneously
handle two signals, usually (but not necessarily) going in opposite directions, so it can
send and receive information at the same time. Duplex modems use two carriers to simultaneously transmit and receive data, each of which has half the bandwidth available to it
and its modulation.
Half-Duplex
The alternative to duplex communications is half-duplex. In half-duplex transmissions,
only one signal is used. To carry on a two-way conversation, a modem must alternately
send and receive signals. Half-duplex transmissions allow more of the channel bandwidth
to be put to use but slow data communications because often a modem must switch
between sending and receiving modes after every block of data crawls through the channel.
Echoplex
The term duplex is often mistakenly used by some communications programs for computers to describe echoplex operation. In echoplex mode, a modem sends a character down the
phone line, and the distant modem returns the same character, echoing it. The echoed
character is then displayed on the originating terminal as confirmation that the character
was sent correctly. Without echoplex, the host computer usually writes the transmitted
character directly to its monitor screen. Although a duplex modem generates echoplex signals most easily, the two terms are not interchangeable.
Analog Services
With early communications programs, echoplex was a critical setup parameter. Some terminal programs relied on modem echoplex to display your typing on the screen. If you
had echoplex off, you wouldnt see what you typed. Other terminal programs, however,
displayed every character that went through the modem, so switching echoplex on would
display two of every letter you typed, lliikkee tthhiiss. Web browsers dont bother you
with the need to select this feature. Most, however, work without echoplex.
Switching Modems
To push more signal through a telephone line, some modems attempt to mimic fullduplex operation while actually running in half-duplex mode. Switching modems are halfduplex modems that reverse the direction of the signal at each end of the line in response
to the need to send data. This kind of operation can masquerade as full-duplex because
most of the time communications go only in one direction. You enter commands into a
remote access system, and only after the commands are received does the remote system
respond with the information you seek. Although one end is sending, the other end is
more than likely to be completely idle.
On the positive side, switching modems are able to achieve a doubling of the data rate
without adding any complexity to their modulation. However, the switching process itself
is time-consuming and inevitably involves a delay because the modems must let each other
know that they are switching. Because transmission delays across long-distance circuits are
often a substantial fraction of a second (most connections take at least one trip up to a
satellite and back down, a 50,000 mile journey that takes about a quarter of a second even
at the speed of light), the process of switching can eat huge holes into transmission time.
Most software modem protocols require a confirmation for each block of data sent, meaning the modem must switch twice for each block. The smaller the block, the more often
the switch must occur. Just one trip to a satellite would limit a switching modem with an
infinitely fast data rate, using the 128-byte blocks of some early modem protocols, to 1024
bits per second at the two-switches-per-second rate.
Asymmetrical Channels
Because of this weakness of switching modems, asymmetrical modems cut the waiting by
maintaining a semblance of two-way duplex communications while optimizing speed in
one direction only. These modems shoehorn in a lower speed channel in addition to a
higher speed one, splitting the total bandwidth of the modem channel unequally.
Early asymmetrical modems were able to flip-flop the direction of the high-speed communications, relying on algorithms to determine which way is the best way. The modern
asymmetrical technologies have a much simpler algorithm. Designed for Internet communications, they assume you need a greater data rate downstream (to you) than upstream
(back to the server). This design is effective because most people download blocks of data
453
454
from the Internet (typically Web pages rife with graphics) while sending only a few commands back to the Web server.
The latest V.90 modems operate asymmetrically at their highest speed. Cable modems and
satellite connections to the Internet also use a variation on asymmetrical modem technology. These systems typically provide you with a wide bandwidth downlink from a satellite
or cable system to permit you to quickly browse pages but rely on a narrow-channel telephone linka conventional modem linkto relay your commands back to the network.
Connection-Enhancing Technologies
Getting the most from your modem requires making the best match between it and the
connection it makes to the distant modem with which you want to communicate.
Although you have no control over the routing your local phone company and long distance carrier give to a given call (or even whether the connection remains consistent during a given call), a modem can make the best of what it gets. Using line compensation, it can
ameliorate some problems with the connection. Fallback helps the modem get the most
from a substandard connection or one that loses quality during the modem link-up. Data
compression helps the modem move more data through any connection, and error correction
compensates for transitory problems that would result in minor mistakes in transmissions.
Line Compensation
Although a long-distance telephone connection may sound unchanging to your ear, its
electrical characteristics vary by the moment. Everything, from a wire swaying in the
Wichita wind to the phone companys automatic rerouting of the call through Bangkok
when the direct circuits fill up, can change the amplitude, frequency, and phase response
of the circuit. The modem then faces two challenges: not to interpret such changes as data
and to maintain the quality of the line to a high-enough standard to support its use for
high-speed transmission.
Under modern communications standards, modems compensate for variations in telephone lines by equalizing these lines. That is, two modems exchange tones at different
frequencies and observe how signal strength and phase shift with frequency changes. The
modems then change their signals to behave in the exact opposite way to cancel out the
variations in the phone line. The modems compensate for deficiencies in the phone line to
make signals behave the way they would have in absence of the problems. If, for example,
the modems observe that high frequencies are too weak on the phone line, they will compensate by boosting high frequencies before sending them.
Modern modems also use echo cancellation to eliminate the return of their own signals
from the distant end of the telephone line. To achieve this, a modem sends out a tone and
listens for its return. Once it determines how long the delay is before the return signal
occurs and how strong the return is, the modem can compensate by generating the opposite signal and mixing it into the incoming data stream.
Analog Services
Echo Cancellation
When satellite rather than fiber optic technology dominated the long-distance telephone
market, the half-second delay imposed by the long distance the signals traveled from earth
to satellite and back created annoying echoes on the line. You spoke, and a fraction of a
second later, you heard your own voice, delayed by the satellite hop, as it looped through
the entire connection.
Switching to fiber optic long-distance connections minimizes the problem, at least to your
ears. But modems can be confused even with the resulting short-delay echoes. A delayed
pulse or phase-shift can sound like good data to a modem. To prevent such problems,
modern modems use echo cancellation, which automatically subtracts the original signal
from what the modem hears after the echo delay, thus canceling out the echo. To properly
cancel the echo, the modem must be trainedduring the initial handshake with a distant
modem, it checks for echo and measures the delay. It later uses the discovered values for
its echo cancellation.
Fallback
Most modems use at most two carriers for duplex communications. These carriers are
usually modulated to fill the available bandwidth. Sometimes, however, the quality of the
telephone line is not sufficient to allow reliable communications over the full bandwidth
expected by the modem, even with line compensation. In such cases, most high-speed
modems incorporate fallback capabilities. When the top speed does not work, they
attempt to communicate at lower speeds that are less critical of telephone line quality. A
pair of modems might first try 56,000bps and be unsuccessful. They next might try 53,000
or switch from high-speed to conventional modem technology with a fallback to
33,600bps.
Most modems fall back and stick with the slower speed that proves itself reliable. Some
modems, however, constantly check the condition of the telephone connection to sense
for any deterioration or improvement. If the line improves, these modems can shift back
to a higher speed.
Multiple-Carrier Modems
Most modems rely on a relatively complex form of modulation on one or two carriers to
achieve high speed. However, one clever idea (now relegated to a historical footnote by
the latest modem standards) is the multiple-carrier modem, which uses relatively simple
modulation on several simultaneous carrier signals. One of the chief advantages of this
system comes into play when the quality of the telephone connection deteriorates. Instead
of dropping down to the next incremental communications rate, thus generally cutting
data speed in half, the multiple-carrier modems just stop using the carriers in the doubtful
regions of the bandwidth. The communication rate may fall off just a small percentage in
the adjustment. (Of course, it could dip by as much as a normal fallback modem as well.)
455
456
Data Compression
Although theres no way of increasing the number of bits that can cross a telephone line
beyond the capacity of the channel, the information-handling capability of the modem
circuit can be increased by making each bit more meaningful. Many of the bits that are
sent through the telecommunications channel are meaningless or redundantthey convey
no additional information. By eliminating those worthless bits, the information content of
the data stream is more intense, and each bit is more meaningful. The process of paring
the bits is called data compression.
The effectiveness of compression varies with the type of data thats being transmitted.
One of the most prevalent data-compression schemes encodes repetitive data. Eight
recurrences of the same byte value might be coded as two bytes, one signifying the value
and the second the number of repetitions. This form of compression is most effective on
graphics, which often have many blocks of repeating text. Other compression methods
may strip out start, stop, and parity bits.
At one time, many modem manufacturers had their own methods of compressing data so
that you needed two matched modems to take advantage of the potential throughput
increases. Today, however, most modems follow international compression standards so
that any two modems using the same standards can communicate with one another at
compressed-data speeds. The most efficient of these international standards is called V.44.
These advanced modems perform the data compression on the fly in their own circuitry
as you transmit your data. Alternately, you can precompress your data before sending it to
your modem. Sort of like dehydrating soup, precompression (also known as file compression)
removes the unnecessary or redundant parts of a file, yet allows the vital contents to be
easily stored and reconstituted when needed. This gives you two advantages: The files you
send and receive require less storage space because they are compressed, and your serial
port operates at a lower speed for a given data throughput.
Note that once a file is compressed, it usually cannot be further compressed. Therefore,
modems that use on-the-fly compression standards cannot increase the throughput of precompressed files. In fact, using one on-the-fly modem data-compression system (MNP5)
actually can increase the transmission time for compressed files as compared to not using
modem data compression.
Error Checking and Error Correction
Because all high-speed modems operate closer to the limits of the telephone channel, they
are naturally more prone to data errors. To better cope with such problems, nearly all
high-speed modems have their own built-in error-checking methods (which detect only
transmission errors) and error-correction methods (which detect data errors and correct
the mistakes before they get passed along to your computer). These error-checking and
error-correction systems work like communications protocols, grouping bytes into blocks
Analog Services
and sending cyclical redundancy checking information. They differ from the protocols
used by communications software in that they are implemented in the hardware instead of
your computers software. That means they dont load down your computer when its
straining at the limits of its serial ports.
It can also mean that software communications protocols are redundant and a waste of
time. As mentioned before, in the case of switching modems, using a software-based communications protocol can be counterproductive with many high-speed modems, slowing
the transfer rate to a crawl. Most makers of modems using built-in error-checking advise
against using such software protocols.
All modem error-detection and error-correction systems require that both ends of the
connection use the same error-handling protocol. In order that modems can talk to one
another, a number of standards have been developed. Today, the most popular are MNP4
and V.42. You may also see the abbreviations LAPB and LAPM describing error-handling
methods.
LAPB stands for Link Access Procedure, Balanced, an error-correction protocol designed for
X.25 packet-switched services such as Telebit and Tymnet. Some high-speed modem makers adapted this standard to their dial-up modem products before the V.42 standard
(described later) was agreed on. For example, the Hayes Smartmodem 9600 from Hayes
Microcomputer Products included LAPB error-control capabilities.
LAPM is an acronym for Link Access Procedure for Modems and is the error-correction protocol used by the CCITT V.42 standard, described later in the chapter.
Combining Voice and Data
Having but a single telephone line can be a problem when you need to talk as well as send
data. In the old days, the solution was to switch. Youd type a message to the person at the
other end of the connection, such as Go voice, pick up the telephone handset, and tell
your modem to switch back to command mode so you could talk without its constant
squeal.
In the early 1990s, several manufacturers developed the means of squeezing both data and
voice down a single telephone line at the same time using special modem hardware. Three
technologiesVoiceView, VoiceSpan, and DSVDvied for market dominance.
Internet technology has made all three irrelevant. Instead of combining data and voice in
modem hardware, the modern alternative is to combine them using software inside your
computer. Your computer captures your voice with a microphone and digitizes it. Web
software packages the voice information into packets that get sent to an ordinary modem
exactly like data packets. At the other end of the connection, the receiving computer converts the voice packets back to audio to play through the computers speakers. The
457
458
modem connection doesnt careor even knowwhether its passing along packets of data
or digitized audio.
Analog Standards
Neither men nor modems are islands. Above all, they must communicate and share their
ideas with others. One modem would do the world no good. It would just send data out
into the vast analog unknown, never to be seen (or heard) again.
But having two modems isnt automatically enough. Like people, modems must speak the
same language for the utterances of one to be understood by the other. Modulation is part
of the modem language. In addition, modems must be able to understand the error-correction features and data-compression routines used by one another. Unlike most human
beings, who speak any of a million languages and dialects, each somewhat ill-defined,
modems are much more precise in the languages they use. They have their own equivalent of the French Academy: standards organizations.
In the United States, the first standards were set long ago by the most powerful force in
the telecommunications industry, which was the telephone company. More specifically, the
American Telephone and Telegraph Company was the telephone company prior to is
breakup announced on January 8, 1982, which resulted in seven local operating companies
(in addition to AT&T). Before then, the Bell System created nearly all U.S. telephone
standards, including two of the historically most important modem standards, Bell 103
and Bell 212A.
With the globalization of business and technology, communication standards have become
international, and the onus to set new standards has moved to an international standards
organization thats part of the United Nations, the International Telecommunications
Union (ITU) Telecommunications Standards Sector, which was formerly the Comite
Consultatif International Telegraphique et Telephonique (in English, thats International
Telegraph and Telephone Consultative Committee). All the current standards for modems
fall under the aegis of the ITU. You can purchase copies of the ITU standards from the
organizations Web site at www.itu.org.
Standards are important when buying a modem because they are your best assurance that
a given modem can successfully connect with any other modem in the world. In addition,
the standards you choose will determine how fast your modem can transfer data and how
reliably it will work. The kind of communications you want to carry out will determine
what kind of modem you need. If youre just going to send files electronically between
offices, you can buy two nonstandard modems and get more speed for your investment.
But if you want to communicate with the rest of the world, you will want to get a modem
that meets the international standards. Table 13.1 summarizes major modem speed
standards.
Analog Services
TA B L E 1 3 . 1
Modem Speed Standards
Standard
Organization
Introduction
Baud Rate
Data Speed
Details
103
Bell
1963
300
300bps
FSK
V.21
CCITT
1965
300
300bps
International version of Bell 103

using different
carrier frequencies
212A
Bell
1979
600
1200bps
QDPSK
V.22
CCITT
1980
600
1200bps
QDPSK,
International version of Bell 212A
V.22bis
CCITT
1984
600
2400bps
QAM
V.32
CCITT
1984
2400
9600bps
QAM
V.32bis
CCITT
1991
2400
14,400bps
QAM
V.34
CCITT
1998
3429
33,600bps
QAM, highest
true analog speed
V.90
ITU
1999
8000
56,000bps up,
Level-sensitive
33,600bps down digital encoding
V.92
ITU
2000
8000
56,000bps up,
48,000 down
Level-sensitive
digital encoding
Nearly all new modems now sold operate at a top speed set by the V.92 modem standard,
the international standard for modem communications at 56,000 bits per second (or
56Kbps) across dial-up telephone lines. It is the highest speed modem standard in use
today and possibly the highest speed true modems will ever achieve. Note that this standard is essentially the same as ITU V.90. The new designation does not indicate an
increase in speed. The chief changes between V.90 and V.92 include improved connection
setup and handshaking, so it takes a modem less time to set up a V.92 connection. The
V.92 standard includes (and is compatible with) the V.90 standard, and modems matching
the two standards will interconnect using V.90 technology.
Strictly speaking, V.92 is not a modem standard because, at its top speeds, it involves no
modulation or demodulation. When line conditions are not favorable, however, it shifts
back to analog technology to cope.
It is an asymmetrical standard, with a maximum upstream data rate of 48,000bps and a
maximum downstream data rate of about 56,000bps. It uses switching technology and
cannot send and receive simultaneously. In theory, in many connections, its downstream
459
460
falls short of its maximum rate because of line conditions. In addition, telephone regulations at one time prohibited the V.92 top operating speed because the last bit of speed
pushed the power level on the telephone line above the allowed standards. Most sources
listed a top practical speed of 53,000 bits per second, but your actual connection speed is
now determined by line conditions and not law.
V.92 takes advantage of the digital technology already in use to shift voice calls across
long-distance digital lines. The telephone standard for voice is to sample analog signals
8000 times a second using an eight-bit digital code. V.92 translates the digital code into
voltage levels on the telephone line from your local phone companys digital switch to
your home or office. Encoding digital information as voltage levels is a technology called
pulse amplitude modulation (PAM). Because the V.92 standard requires modems to operate
at frequencies in excess of normal voice circuits, its range is limitedV.92 connections
require your modem to have no more than three miles of telephone wire between it and
your telephone companys switch.
It should take 256 voltage levels, which are technically called quantization levels, to encode
the eight-bit data stream. However, noise and line conditions can often mask changes of
1/256th of the voltage on an ordinary telephone line. To avoid problems with noise, the
V.92 system uses only 128 quantization levels to encode data, allowing a seven-bit digital
code with an 8000Hz sampling rate or a 56,000bps data rate.
As originally developed as the V.90 standard, modems used the PAM system only for
downstream communications (that is, from the telephone companys switch to your
modem). Upstream data (from your modem back to the phone company) relied on conventional modem technology under the V.34 standard, although limited to 31,200bps.
The underlying assumption was that the phone company was better able to control digital
signals and keep them within the required limits. Moreover, only the telephone company
had direct access to the digital connection with long-distance trunk lines.
With the advent of the V.92 standard, this situation changed. Under V.92, modems use a
technology called V.PCM upstream to let you use PAM to send data to other modems and
services at a rate of 48,000 bits per second. The only difference between upstream and
downstream is that the upstream signal is limited to only 64 quantization levels, thus
allowing a six-bit code. The fewer voltage levels means that modem manufacturing inconsistencies and installation differences are less likely to cause signals exceeding the levels
allowed by law.
To reduce the time needed for modems to connect, V.92 uses a technology called
QuickConnect that takes advantage of the digital nature of the telephone system from the
central office onward. The only part of the connection that needs line compensation is the
local phone loop between where you use your modem and the telephone companys central office. Unless youre traveling with a notebook computer, this connection doesnt
Analog Services
often change (and even when youre traveling, youre likely to use the same hotel phone
for several connections at a time). Consequently, QuickConnect checks and remembers
the settings for the local telephone loop and tries to reuse them if possible.
In truth, QuickConnect does not reduce the time required to determine the quality of a
connection and compensate for it. Rather, it remembers the quality of your last connection under the assumption that youll use the same telephone line for consecutive calls.
The QuickConnect system stores in nonvolatile memory the equalizer and echo-cancellation settings as well as the digital characteristics of the line. When you place a subsequent
call, your modem first examines the tone from the distant modem and compares it to the
setting in memory. In case of a match, the modem starts with the stored settings to make a
fast connection. If the modem detects a substantial change (such as youre using a notebook computer in a different hotel room), it walks through the standard V.90 handshaking
procedure. The first time you use a V.92 modem in a given location, it will use the standard V.90 handshaking, requiring 20 to 30 seconds to adjust for the local phone line.
Subsequent QuickConnect handshaking usually takes about half as long.
Recognizing how modems are used in normal telephone systems, V.92 explicitly recognizes the need to be able to pause modem communications so you can take another call. A
feature called Modem-on-Hold lets you suspend data transfer without looking at your
modem connection. In effect, you can put your modem on hold, take another call, and
then return to your modem connection.
The V.92 standard is regarded as a refinement of V.90, which evolved out of two competing 56Kbps systems. K56flex is a proprietary technology that was independently developed
and initially marketed as two different and incompatible systems by Rockwell and Lucent
Technologies. In November, 1996, the two companies agreed to combine their work into
a single standard. x2 is a proprietary technology developed by U.S. Robotics. Both
K56flex and x2 used the same PAM digital encoding as was adopted for V.90 and V.92.
They differed from each other (and the standard) only in the handshaking used to set up
the connection.
In addition to speed standards, several other standards have been used for data compression and error correction. Before the widespread adoption of international standards,
many U.S. companies used technologies developed by Microcom Corporation and formalized as Microcom Networking Protocol Standards Levels 1 through 9. Currently, however, the dominant modem compression and error-control standards are those set by the
ITU. These include V.42, a world-wide error-correction standard (which also incorporates
MNP4 as an alternative protocol), V.42bis (data compression that can yield compression
factors up to four, potentially quadrupling the speed of modem transmissions), and V.44,
an improved compression system.
461
462
Modem Hardware
A modem is a signal converter that mediates the communications between a computer and
the telephone network. In function, a modern computer modem has five elements: interface circuitry for linking with the host computer; circuits to prepare data for transmission
by adding the proper start, stop, and parity bits; modulator circuitry that makes the
modem compatible with the telephone line; a user interface that gives you command of
the modems operation; and the package that gives the modem its physical embodiment.
Computer Interface
For a modem to work with your computer, the modem needs a means to connect to your
computers logic circuits. At one time, all modems used a standard or enhanced serial port
to link to the computer. However, because the standard serial port tops out at a data rate
thats too slow to handle todays fastest modemsthe serial limit is 115,200 bits per second, whereas some modems accept data at double that ratemodem-makers have developed parallel-interfaced modems.
All modems, whether installed outside your computer, in one of its expansion slots, or in a
PCMCIA slot, make use of a serial or parallel communications port. In the case of an
internal computer modem, the port is embedded in the circuitry of the modem, and the
expansion bus of the computer itself becomes the interface.
With an external modem, this need for an interface (and the use of a port) is obvious
because you fill the ports jack with the plug of a cable running off to your modem. With
an internal modem, the loss is less obvious. You may not even detect it until something
doesnt work because both your modem and your mouse (or some other peripheral) try to
use the same port at the same time.
In the case of serial modems, this interface converts the parallel data of your computer
into a serial form suitable for transmission down a telephone line. Modern modems operate so fast that the choice of serial port circuitry (particularly the UART) becomes critical
to achieving the best possible performance.
The serial and parallel ports built into internal modems are just like dedicated ports of the
same type. They need an input/output address and an interrupt to operate properly. The
Plug-and-Play system assigns these values to the modem during its configuration process.
(Older modems required you to select these values with jumpers or switches.)
Ordinarily you dont need to know or bother with these values. Some communications
software, however, may not mesh perfectly with the Windows system. It may demand you
tell it the port used by your modem. The modems properties sheet lists this value. You
can check it under Windows by clicking the modem icon in Control Panel and then clicking the Properties tab.
Analog Services
Data Preparation
Modern modem communications require that the data you want to send be properly prepared for transmission. This pre-transmission preparation helps your modem deliver the
highest possible data throughput while preventing errors from creeping in.
Most modem standards change the code used by the serial stream of data from the computer interface into code thats more efficient (for example, stripping out data-framing
information for quicker synchronous transfers). The incoming code stream may also be
analyzed and compressed to strip out redundant information. The modem may also add
error-detection or error-correction codes to the data stream.
At the receiving end, the modem must debrief the data stream and undo the compression
and coding of the transmitting modem. A micro controller inside the modem performs
these functions based on the communications standard you choose to use. If you select a
modem by the communications standards it uses, you dont have to worry about the
details of what this micro controller does.
Modulator
The heart of the modem is the circuitry that actually converts the digital information
from your computer into analog-compatible form. Because this circuitry produces a modulated signal, it is called a modulator.
User Interface
The fourth element in the modem is what you see and hear. Most modems give you some
way of monitoring what they do either audibly with a speaker or visually through a light
display. These features dont affect the speed of the modem or how it works but can make
one modem easier to use than another. Indicator lights are particularly helpful when you
want to troubleshoot communication problems.
Line Interface
Finally, the modem needs circuitry to connect with the telephone system. This line interface circuitry (in telephone terminology, a data access arrangement) boosts the strength of
the modems internal logic-level signals to a level matching that of normal telephone service. At the same time, the line interface circuitry protects your modem and computer
from dangerous anomalies on the telephone line (say, a nearby lightning strike), and it
protects the telephone company from odd things that may originate from your computer
and modem (say, a pulse from your computer in its death throes).
From your perspective, the line interface of the modem is the telephone jack on its back
panel. Some modems have two jacks so that you can loop through a standard telephone.
By convention, the jack marked Line connects with your telephone line; the jack
marked Phone connects to your telephone.
463
464
Over the years, this basic five-part modem design has changed little. But the circuits
themselves, the signal-processing techniques that they use, and the standards they follow
have all evolved to the point that modern modems can move data as fast as the theoretical
limits that telephone transmission lines allow.
Packaging
Internal modems plug into an expansion slot in your computer. The connector in the slot
provides all the electrical connections necessary to link to your computer. To make the
modem work, you only need to plug in a telephone line. The internal modem draws
power from your computer, so it needs no power supply of its own. Nor does it need a
case. Consequently, the internal modem is usually the least expensive at a given speed.
Because internal modems plug into a computers expansion bus, a given modem is compatible only with computers using the bus for which it was designed. You cannot put a
computers internal modem in a Macintosh or workstation.
External modems are self-contained peripherals that accept signals from your computer
through a serial or parallel port and also plug into your telephone line. Most need an
external source of power, typically a small transformer that plugs into a wall outlet and
through a short, thin cableinto the modem. At a minimum, then, you need a tangle of
three cables to make the modem work. You have two incentives to put up with the cable
snarl. External modems can work with computers that use any architecture as long as the
computer has the right kind of port. In addition, external modems usually give you a full
array of indicators that can facilitate troubleshooting.
Pocket modems are compact external modems designed for use with notebook computers.
They are usually designed to plug directly into a port connector on your computer, eliminating one interface cable. Many eliminate the need for a power supply and cable by running from battery power or drawing power from your computer or the telephone line.
PC Card modems plug into that PCMCIA slots that are typically found in notebooks. They
combine the advantage of cable-free simplicity of internal modems with the interchangeability of external modems (the PCMCIA interface was designed to work with a variety of
computer architectures). The confines of the PCMCIA slot also force PC Card modems
to be even more compact than pocket modems. This miniaturization takes its toll in
higher prices, however, although the ability to quickly move one modem between your
desktop and portable computer can compensate for the extra cost.
The confines of a PCMCIA slot preclude manufacturers from putting a full-size modular
telephone jack on PC Card modems. Modem-makers use one of two workarounds for this
problem. Most PC Card modems use short adapter cables with thin connectors on one
end to plug into the modem and a standard modular jack on the other. Other PC Card
modems use the X-Jack design, developed and patented by Megahertz Corporation (now
part of 3Com). The X-Jack pops out of the modem to provide a skeletal phone connector
Analog Services
into which you can plug a modular telephone cable. The X-Jack design is more convenient because you dont have to carry a separate adapter with you when you travel. On the
other hand, the X-Jack makes the modem more vulnerable to carelessness. Yank on the
phone cable, and it can break the X-Jack and render the modem useless. Yanking on an
adapter cable will more likely pull the cable out or damage only the cable. On the other
hand, the connectors in the adapter cables are also prone to invisible damage that can lead
to unreliable connections.
Indicators
The principal functional difference between external (including pocket) and internal
(including PC Card) modems is that the former have indicator lights that allow you to
monitor the operation of the modem and the progress of a given call. Internal modems,
being locked inside your computer, cannot offer such displays. Some software lets you
simulate the lights on your monitor, and Windows will even put a tiny display of two of
these indicators on your task bar. These indicators can be useful in troubleshooting
modem communications, so many computer people prefer to have them available (hence,
they prefer external modems).
The number and function of these indicators on external modems vary with the particular
product and the philosophy of the modem-maker. Typically youll find from four to eight
indicators on the front panel of a modem, as shown in Figure 13.1.
FIGURE 13.1
Typical indicators on
an external modems
front panel.
The most active and useful of these indicators are Send Data and Receive Data. These
lights flash whenever the modem sends out or receives in data from the telephone line.
They let you know whats going on during a communications session. For example, if the
lights keep flashing away but nothing appears on your monitor screen, you know you are
suffering a local problem, either in your computer, its software, or the hardware connection with your modem. If the Send Data light flashes but the Receive Data light does not
flicker in response, you know that the distant host is not responding.
Carrier Detect indicates that your modem is linked to another modem across the telephone
connection. It allows you to rule out line trouble if your modem does not seem to be getting a response. This light glows throughout the period your modem is connected.
465
466
Off-Hook glows whenever your modem opens a connection on your telephone line. It
lights when your modem starts to make a connection and continues to glow through dialing, negotiations, and the entire connection.
Terminal Ready glows when the modem senses that your computer is ready to communicate with it. When this light is lit, it assures you that youve connected your modem to
your computer and that your computers communications software has properly taken
control of your serial port.
Modem Ready glows whenever your modem is ready to work. It should be lit whenever
your modem is powered up and not in its test state.
High Speed indicates that the modem is operating at its fastest possible speed. Some
modems have separate indicators for each speed increment they support. Others forego
speed indicators entirely.
Auto Answer lights to let you know that your modem is in its answer mode. If your telephone rings, your modem will answer the call (at least if its connected to the line that is
ringing).
Table 13.2 summarizes the mnemonics commonly used for modem indicators and their
functions.
TA B L E 1 3 . 2
Modem Indicator Abbreviations and Definitions
Mnemonic
Spelled Out
Meaning
HS
High Speed
Modem is operating at highest speed.
AA
Auto Answer
Modem will answer phone.
CD
Carrier Detect
Modem is in contact with remote system.
OH
Off Hook
Modem is off hook, using the phone line.
RD
Receive Data
Modem is receiving data.
SD
Send Data
Modem is transmitting data.
TR
Terminal Ready
PC is ready to communicate.
MR
Modem Ready
Modem is ready to communicate.
Digital Services
Today, all of the international telephone network is digital, with the exception of one
wirethe one running from the telephone company central office to your home. This
local loop, sometimes called the last mile, is the same stuff that Alexander Graham Bell
Digital Services
experimented with and uses the same technology he developeda pair of copper wires
designed to carry analog voice signals. Eventually, this local loop will be replaced with a
newer connection technology, such as coaxial cable or fiber optics carrying only digital
data.
Businesses already have this option available to them. Modern cable television systems
offer digital data communications capabilities as well, although dial-up voice circuits
remain the province of the telephone company (mostly for legal reasons). Even so, theres
little doubt youll eventually shift from analog to digital services for all of your telecommunications needs.
You can already get all-digital circuits for your data. The only question is who will provide
the connection. Three technologies can provide you with a high-speed all-digital link
telephone, cable, and satellite.
All three work. All three deliver speeds that make modem connections seem like they are
antiquated. In fact, the most important limiting factor is availability. Only satellite services
can promise a link in any part of the United States (and most of the world). The others
depend on your local telephone company or cable provider upgrading its facilities to handle digital subscriber services.
Telephone Services
One key player in the supply of digital telecommunications services is quite familiarthe
telephone company. Beyond traditional POTS, telephone companies have developed a
number of all-digital communication services. Some of these have been around for a
while, aimed at business users with heavy data needs. Several new, all-digital services are
aimed directly at you as an individual consumer.
The range of digital services supplied by telephone companies is wide and spans a range
of data rates. Table 13.3 lists many of these and their maximum data rates.
TA B L E 1 3 . 3
Maximum Data Rates of Digital Telecommunications Standards
Standard
Connection Type
Downstream Rate
Upstream Rate
V.34
Analog
33.6Kbps
33.6Kbps
V.90
Digital/analog
56Kbps
33.6Kbps
V.92
Digital
56Kbps
48Kbps
SDS 56
Digital
56Kbps
56Kbps
ISDN
Digital
128Kbps
128Kbps
continues
467
468
TA B L E 1 3 . 3
Continued
Standard
Connection Type
Downstream Rate
Upstream Rate
SDSL
Digital
1.544Mbps
1.544MBps
T1
Digital
1.544Mbps
1.544MBps
E1
Digital
2.048Mbps.
2.048Mbps
ADSL
Digital
9Mbps
640Kbps
VDSL
Digital
52Mbps
2Mbps
You will still talk on the telephone for ages to come (if your other family members give
you a chance, of course), but the nature of the connection may finally change. Eventually,
digital technology will take over your local telephone connection. In fact, in many parts of
America and the rest of the world, you can already order a special digital line from your
local telephone company and access all-digital switched systems. You get the equivalent of
a telephone line, one that allows you to choose any conversation mate whos connected to
the telephone network (with the capability of handling your digital data, of course), as easily as dialing a telephone.
T1
The basic high-speed service provided by the telephone company is called T1, and its
roots go back to the first days of digital telephony in the early 1960s. The first systems
developed by Bell Labs selected the now-familiar 8KHz rate to sample analog signals and
translate them into eight-bit digital values. The result was a 64Kbps digital data stream.
To multiplex these digital signals on a single connection, Bells engineers combined 24 of
these voice channels together to create a data frame 193 bits long, the extra bit length
defining the beginning of the frame. The result was a data stream with a bit rate of
1.544Mbps. Bell engineers called the resulting 24-line structure DS1. AT&T used this
basic structure throughout its system to multiply the voice capacity of its telephone system, primarily as trunk lines between exchanges.
As telephone demand and private business exchanges (PBXs) became popular with larger
businesses, the telephone company began to offer T1 service directly to businesses. As
digital applications grew, T1 became the standard digital business interconnect. Many
Web servers tie into the network with a T1 line.
A key feature of the DS1 format was that it was compatible with standard copper telephone lines, although requiring repeaters (booster amplifiers) about every mile. The signal itself is quite unlike normal analog telephone connections, however, and that creates a
problem. Its signal transmission method is called Alternate Mark Inversion (AMI), a formatting code for T1 transmissions over twisted-pair copper cable. T1 transmissions are in
Digital Services
bipolar form. AMI represents a zero (or space) by the absence of a voltage; a one (or
mark) is represented by a positive or negative pulse, depending on whether the preceding
one was negative or positive (that is, marks are inverted on an alternating basis). This
encoding system generates a signal with a bandwidth about equivalent to its data rate,
1.5MHz. This high-speed signal creates a great deal of interference, so much that two T1
lines cannot safely cohabitate in the 50-pair cables used to route normal telephone services to homes.
Outside of the United States, the equivalent of T1 services is called E1. Although based
on the same technology as T1, E1 combines 30 voice channels with 64Kbps bandwidth to
create a 2.048Mbps digital channel.
Serious Web surfers dream of having a dedicated T1 line. The cost, however, is prohibitive. Installation is often thousands of dollars and monthly charges may be a thousand dollars, sometime more. Typically, your Internet Service Provider has an T1 (or better)
connection and divides it up, giving each customer a single modem-slice.
T3 is a synonym for DS3 service, which is approximately 45Mbps, and OC-3 is approximately 155Mbps fiber interface.
The primary problem with T1 is the interference-causing modulation system it uses, one
based on 1960s technology. Using the latest modulation techniques, the telecommunications industry has developed a service called High data rate Digital Subscriber Line (HDSL)
that features the same data rate as T1 or E1 but requires a much narrower bandwidth,
from 80 to 240KHz. One basic trick to the bandwidth-reduction technique is splitting the
signal across multiple phone lines. For T1 data rate, the service uses two lines; for E1,
three. Besides reducing interference, the lower data rate allows longer links without
repeaters, as much as 12,000 feet.
HDSL delivers high-speed data networking, up to 1.544Mbps over two copper pairs and
up to 2.048Mbps over three pairs, at a maximum range of 20,000 feet (about 3.8 miles or
6.1 km) from a central office. It is similar to Symmetrical Digital Subscriber Line (discussed later) and has symmetrical transmission capabilities. Most T1 lines installed today
utilize this technology.
Unfortunately the subscriber in the name of the standard was not meant to correspond
to you as an individual. It fits into the phone company scheme of things in the same place
as T1linking businesses and telephone company facilities.
DSL
The service youre most likely to buy from your telephone company goes under the name
Digital Subscriber Line. It uses your ordinary telephone wires to carry high-speed digital
signals. In fact, todays technologies let you use both the analog phone service and the digital capacity of the same wires at the same time. The high frequencies of the digital signal
469
470
are easily split from the low frequencies of the analog signal, so one wire can do double
duty. When telephone companies first introduced this piggyback service, they called it
G.Lite, but it has become the standard for residential DSL.
G.Lite is a special case of a more generalized service called Asymmetrical Digital
Subscriber Line or ADSL. It is asymmetrical because it offers a higher downstream data
rate from the server compared to its upstream rates, from you back to the server.
When telephone companies first offered DSL services, they needed to send out a technician on each job to install the splitters required to separate the analog and digital signals.
The need for a technician added hundreds of dollars to the cost of setting up an ADSL
connection. To eliminate the need for the splitter and create a common consumer standard for ADSL, several manufacturers in the telecommunications industry banded
together to create the Universal ADSL Working Group, the organization that defined the
G.Lite standard under the formal designation G.992.2.
ADSL can move downstream data at speeds up to 8Mbps. Upstream, however, the maximum rate is about 640Kbps to 1Mbps. ADSL doesnt operate at a single rate as does T1.
Its speed is limited by distance, with longer distances imposing greater constraints. It can
push data downstream at the T1 rate for up to about 18,000 feet from the central office.
At half that distance, its downstream speed potential approaches 8.5Mbps. To distinguish
the more general form of ADSL from G.Lite, this full-fledged ADSL is sometimes called
ADSL Full Rate or G.dmt. Its official designation is G.992.1.
G.Lite is one variety of ADSL. G.Lite allows for a bandwidth downstream of up to
1.544Mbps. Upstream the asymmetrical system allows for a bandwidth of up to 512Kbps.
The maximum length of a G.Lite connection stretching between the central office and
your home or business is 18,000 feet.
Table 13.4 summarizes the downstream speeds and maximum distances possible with
ADSL technology.
TA B L E 1 3 . 4
ADSL Downstream Data Rates
Equivalent Service
Downstream Data Rate
Distance
G.Lite
1.544Mbps
18,000 feet
T1
1.544Mbps
18,000 feet
E1
2.048Mbps
16,000 feet
DS1
6.312Mbps
12,000 feet
ADSL
8.448Mbps
9000 feet
Digital Services
Another kind of DSL offers the same speed in both the upstream and downstream directions. Termed Symmetrical Digital Subscriber Line (SDSL), this kind of service is best suited
for companies offering Web services from their own servers. The data speed of the SDSL
system ranges from 160Kbps up to 1.544Mbps. The maximum usable rate depends on the
distance of the subscriber from the central office. The lowest speed occurs at a maximum
range of the SDSL system, which is 24,000 feet (about 4.5 miles or 7.2 km).
The acronym SDSL has been used in the past to describe another service called Single-line
Digital Subscriber Line, which was an alternative to multiline data services for high-speed
operation.
The modulation system used by all DSL services operates at frequencies above the baseband used by the ordinary telephone service, which is why a single line can carry highspeed digital signals and ordinary telephone signals simultaneously. In typical DSL
implementations, the DSL signals start at about 50KHz, leaving the lower frequencies for
carrying conventional voice signals.
The splitter that divides the signal between voice and data on the subscribers premises is
the equivalent of a stereo speakers crossover; the splitter combines a high-pass filter to
extract a data-only signal and a low-pass filter to extract the voice-only signal. Using the
G.Lite system, you install the required splitters (sometimes called filters) by simply plugging them in using modular plugs and jacks like ordinary telephone equipment.
VDSL
The next step above ADSL is the Very-high-data-rate Digital Subscriber Line (VDSL). A
proposal only, the service is designed to initially operate asymmetrically at speeds higher
than ADSL but for shorter distances, potentially as high as 51.84Mbps downstream for
distances shorter than about 1000 feet, falling to one-quarter that at about four times the
distance (12.86Mbps at 4500 feet). Proposed upstream rates range from 1.6Mbps to
2.3Mbps. In the long term, developers hope to make the service symmetrical. VDSL is
designed to work exclusively in an ATM network architecture. As with ADSL, VDSL can
share a pair of wires with an ordinary telephone connection or even ISDN service.
SDS 56
Switched Data Services 56 (sometimes shortened to Switched-56) is an archaic connection
system that yielded a single digital channel capable of a 56Kbps data ratethe same as
with a modem but with true digital signals. The Switched-56 signals traveled through
conventional copper twisted-pair wiring (the same old stuff that carries your telephone
conversations). For most telephone companies, it was an interim service to bridge the gap
between POTS and ISDN service areas.
471
472
With Switched-56 you needed special head-end equipmentthe equivalent of a modem

to link the wire to your computer. To take advantage of the connection, you also needed
to communicate with someone who also had SDS 56 service.
In some locales, SDS 56 was no more expensive than an ordinary business telephone line.
Installation costs, however, could be substantially higher (PacBell, for example, at one
time charged $500 for installation), and some telephone companies added extra monthly
maintenance charges in addition to the normal dial-up costs. With modern modems
promising the same speed with no extra charges, little wonder Switched-56 gets discussed
in the past tense.
ISDN
The initials stand for Integrated Services Digital Network, a first attempt at bringing true
digital communications to the home through existing telephone lines. Although the service is still available, its essentially irrelevant because DSL offers more speed at about the
same cost.
ISDN predates DSL. Its start came in November 1992 when AT&T, MCI, and Sprint
embraced a standard they called ISDN-1. Today, two versions of ISDN are generally available. The simplest is the Basic Rate Interface (BRI), which takes advantage of the copper
twisted-pair wiring thats already in place, linking homes and offices to telephone
exchanges. Instead of a single analog signal, an ISDN line uses what is called 2B1Q line
coding to carry three digital channels: two B (for Bearer) channels that can carry any kind
of data (digitally encoded voice, fax, text, and numbers) at 64,000bps, and a D (or Delta)
channel, operating at 16,000bps, that can carry control signals and serve as a third data
channel. The three channels can be independently routed to different destinations
through the ISDN system.
The maximum distance an ISDN line can stretch from the central office is 18,000 feet
(about 3.4 miles or 5.5 km). To accommodate longer runs, this distance can be doubled by
adding a repeater in the middle of the line. A repeater is an amplifier that regenerates the
digital signals, erasing the signal distortion that arises on long lines.
A single BRI wire enables you to transfer uncompressed data bidirectionally at the
64,000bps rate, exactly like a duplex modem today but with higher speed and error-free
transmission, thanks to its all-digital nature. Even during such high-speed dual-direction
connections, the D channel would still be available for other functions.
The more elaborate form of ISDN service is called the Primary Rate Interface (PRI). This
service delivers 23 B channels (each operating at 64,000 bits per second) and one D channel (at 16,000 bits per second). As with normal telephone service, ISDN service is billed
by time in use, not the amount of data transmitted or received.
Digital Services
The strength of BRI service is that it makes do with todays ordinary twisted-pair telephone wiring. Neither you nor the various telephone companies need to invest the billions of dollars required to rewire the nation for digital service. Instead, only the central
office switches that route calls between telephones (which today are mostly plug-in
printed circuit boards) need to be upgraded.
Cable Services
The chief performance limit on telephone service is the twisted-pair wire that runs from
the central office to your home or business. Breaking through its performance limits
would require stringing an entirely new set of wires throughout the telephone system.
Considering the billions of dollars invested in existing twisted-pair telephone wiring, the
likelihood of the telephone company moving into a new connection system tomorrow is
remote.
Over the past two decades, however, other organizations have been hanging wires from
poles and pulling them underground to connect between a third to a half of the homes in
the United Statescable companies. The coaxial cables used by most such services have
bandwidths a hundred or more times wider than twisted pair. They regularly deliver
microwave signals to homes many miles from their distribution center.
Tapping that bandwidth has intrigued cable operators for years, and the explosive growth
of the Internet has set them salivating. The advent of digital cable has made Web connections an option for most cable subscribers.
Architecture
The cable television system differs from the international telephone system in several
ways. Most cable television systems are local. They are meant to cover only a limited geographic range. Cable systems do not interconnect. Each cable operator plucks the signals
it needs from the air, either from distant broadcast stations or from satellite. There is no
great cable web that allows you to directly link with another cable user anywhere in the
world.
Moreover, cable systems are designed differently. Telephone systems are point to point,
caller to caller, one on one. Cable systems are designed for broadcast in a one-to-many
fashion. Cable systems send the same signals to each of their subscribers. The wiring for
telephone systems resembles a star with a center hub (the central office) and individual
nodes, each connected directly to the hub with its own (albeit low bandwidth) wire. The
cable system is a spineone big, wide bandwidth wire carrying signals for all subscribers,
each of which taps into the same spine for the same signals.
The essence of the cable design is that all users share the same bandwidth. The cable is
piled with as many as 500 television channelsthats a full 3GB of bandwidth-ignoring
473
474
guard bands. When operators put Internet signals on the cable system, users have to share
that bandwidth, too. How wide a slice of bandwidth each user gets depends primarily on
how many users are sharing. If youre the only one to log on to the Internet on your cable
system, you can get speeds that T1 users would envy. Log on when Microsoft offers free
downloads of a new version of Internet Explorer or a beer company runs an online bikini
competition, and youre apt to get more free beer than bandwidth through the cable (that
is, about zero). With a high-speed DSL telephone line, you are guaranteed bandwidth.
With cable, you share and take your chances. You could do betteror much, much worse.
In technical terms, the telephone system guarantees Quality of Service (QoS). The cable
system may not (although some operators are making QoS guarantees to lure you over to
coaxial cable).
Cable systems allow for individual addressing of subscribers equipment. Each cable box
has an electronic serial number that the cable operators equipment can address individually (for example, to alter the services you are authorized to receive). But this individual
addressing does not provide an individual channel to your home. The signals to control
your cable box are broadcast to all subscribers on the cable system. Only your box with
the correct serial number responds to commands sent to it.
Standards
During the first few years of cable-based data services, the one element lacking was standardization. Each cable operator used its own equipment designs to adapt data signals to
the cable medium. The adapters used by different cable operators were incompatible, as
were the data signals from the head-end, the cable companys equivalent of the telephone
companys central office. Although such proprietary designs gave cable companies a measure of securitythey helped thwart widespread hackingthey also made equipment
more expensive and employee training more difficult.
Several efforts at developing cable standards started in the mid1990s. The first major
effort started at the IEEE, which formed a new working group called the Cable TV Media
Access Control and Physical Protocol Working Group in May 1994, to define a standard for
cable modems. The IEEE group members had difficulty agreeing to any single standard,
however, and the group missed its original target of December 1995, for publishing a
specification.
An impatient cable industry got tired of waiting for engineers to come up with a standard,
so a group of cable television operators, including such major players as Comcast, Cox,
MSOs, TCI, and Time Warner, started their own standardization effort. The companies
formed an independent limited partnership called Multimedia Cable Network System
Partners, Ltd. (usually shortened to MCNS), which Continental Cablevision and Rogers
Cablesystems then joined. In little more than a year, the partnership sifted through 12
proposals to produce a draft standard that it published as the Data Over Cable Service
Digital Services
Interface Specification (DOCSIS) in March, 1997. The industry research organization

CableLabs took over the administration of the specification and developed a compliance
program. By March 1998, the group had developed an interoperability certification program and, in March 1998, the ITU endorsed the DOCSIS standard as an official standard
ITU J.112.
A revised version of the DOCSIS specification, version 1.1, was released in April 1999.
The new standard better defined signal parameters to guarantee the bandwidth of the signal and minimize delays. In addition, the revision added standards for new services,
including voice-over-Internet telephones and constant bit-rate services. A European version of DOCSIS, called EuroDOCSIS, is used with the different television standards prevailing there. Another update, to DOCSIS 2.0, is in the works. The complete DOCSIS
specification is available online at www.cablemodem.com/specification.
DOCSIS fits data onto cable by taking over a single television channel for downstream
data and a second channel for upstream data. Each television channel has a bandwidth of
6MB. In general, the system can use any channel in the VHF or UHF ranges (from 50 to
864MHz) for downstream data, but upstream data is restricted to lower frequencies (5 to
42MHz), which makes the adapters less expensive to manufacturer.
The DOCSIS standard allows cable operators flexibility in choosing either 64- or 256state quadrate amplitude modulation for the downstream data signals they provide.
Typically cable operators use 64-state QAM, with which a single 6MHz downstream television channel can carry data at a rate of about 27Mbps. Upstream, most cable operators
have a choice of 16-state QAM or quadrature phase-shift keying, a more robust but slower
modulation scheme most choose to ride over the higher noise levels prevalent at lower
frequencies on cable. In this asymmetrical system, cable operators often limit upstream
bandwidth from individual users, often to as little as 320Kbps, although the standard
allows upstream rates as high as 10Mbps. At low-demand times, you might score downstream bandwidth approaching the full 27Mbps rate. Typically, however, cable services
deliver downstream data at 1 to 3Mbps.
To allow multiple users to share the same bandwidth, DOCSIS uses time division multiple
access (TDMA) technology. That is, each user gets a fraction of the total bandwidth in
which to transmit and receive data. To gain access to the cable, DOCSIS uses the
Ethernet MAC (Media Access Control) layer.
DOCSIS includes provisions for each user to have a 14-bit subscriber ID, which allows
cable operators to individually tailor service much as they do premium cable television
channels. For example, individual users may be assigned different bandwidths.
Before the wide adoption of DOCSIS, some cable companies deployed telco-return
modems, which used the cable companys high-speed coaxial cable for downstream data but
475
476
made the upstream link through a conventional phone line (with all its bandwidth constraints). With the move to DOCSIS, most cable modems now use only a cable connection.
Because all signals ride across the same coaxial cable, all your neighbors have access to the
packets of data you send and receive. To maintain privacy, your cable modem automatically encrypts everything you send and receive using the Data Encryption Standard (DES)
algorithm.
Satellite Services
The same technology used by direct-broadcast satellite television, through which you can
grab viewable video directly from orbiting satellites, also works for data. Using a small
parabolic antenna pointed at a geosynchronous satellite orbiting about 24,000 miles away,
you can tap into the Internet at speeds well beyond the capabilities of dial-up telephone
connections. Instead of television signals, the satellite simply beams a stream of data down
to earth.
The leading satellite service, DirecPC from Hughes Electronics, initiated service in 1997.
It bills itself as the fastest Internet service available nationwide. Although slower than
either DSL or cable modems, it holds the advantage of availabilityanywhere you can see
the southern sky, you can make a connection to DirecPC.
Satellite systems are inherently asymmetrical. You dont transmit your needs to the satellitedoing so would require an uplink and a much larger antenna. Instead, you use the
satellite connection only for a downlink, to receive data. To send data, your computer
connects to the Web through a conventional dial-up modem. The satellite-based downlink operates at 400Kbps while your phone-based uplink struggles along at modem speed,
14.4 to 56Kbps.
Satellites best fit a broadcast model. That is, they dump out their signals across wide areas
for consumption by the multitudes rather than directly targeting individuals. By limiting
the downlink bandwidth to 400Kbps, they maximize the number of subscribers that can
share the system. In addition, DirecPC attempts to maximize the speed and usefulness of
its product using push technology. The system pushes out selected Web and newsgroup
information, and your computer captures it as it is sent out, spooling the data to disk.
When you want to access one of the pushed Web sites or newsgroups, you can read it
almost instantly from the cache. So your computer wont clog up by trying to cache the
entire Internet, the system allows you to choose which sites and groups to cache locally.
Digital Services
DirecPC uses a 21-inch elliptical antennae designed for roof mounting. The DirecPC
antenna is a single-purpose device and can be used only for data, not satellite television
reception. The system also requires a receiver (sometimes called a modem) that may be
installed as an expansion board inside your computer or as a standalone external peripheral.
477
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
14
Internet
Surfings a breezeand I do mean riding the Web and not a wave. Just
point and click, and you can visit any Web site, anywhere in the world,
in microseconds (or a few minutes, if you have a dial-up line). About the
only thing easier is being dead, and thats not nearly as much fun (or so
Ive been told).
The Web is easy to use because, believe it or not, it was designed that
way. Scientists working around huge particle acceleratorsthe kind that
smash the tiniest pieces of matter into one another at nearly the speed of
light to create huge research budget deficitsspent a bit of their spare
time developing the idea. Basically, they decided to put a pretty face on
the work of the military-educational complex that let college professors
in Berkeley play Pong with researchers at Princeton.
But whats easy for you to use isnt necessarily easy for your computer.
And its not just your computer. Beneath all the fun and games is a snarl
of cables more tangled than a planet-size bowl of spaghetti, millions of
computers, and an assortment of hardware stranger than the collection
of an alien zoo (routers, switchers, multiplexers, and demultiplexers), and
some things so esoteric no one knows what to call them. That easy
point-and-click interface of the Web covers up an international conspiracy of computers working together to put pop-up ads atop your every
click.
As with any conspiracy, getting to the bottom of the Internet requires
getting the answers to a couple of questions: What does your computer
know, and when does it know it? The answers tell how the complex web
of the Internet really operates.
480
Actually, getting to the bottom of the Internet is easy. Youre already there. The computer
sitting in front of you is considered the lowest of the low, a mere client to the millions of
servers on the Web. At the far end is a hallowed spaceactually a super-superstitious 13
of themhousing the true masterminds of the Web, the 13 root name servers, the
machines that are the ultimate organizers of the Web.
In between is the Internet. But its nothing you can get your hands on. Its nothing real.
Although the Internet is built using hardware, it is not hardware itself. Similarly, you need
hardware to connect to the Internet, but that hardware only serves as a means to access
what you really want: the information that the Internet can bring to your computer.
Without the right hardware, you could not connect to the Internet, but having the hardware alone wont make a new World Wide Web.
Despite its unitary name, there is no giant Internet in the sky or in some huge office complex somewhere. In fact, the Internet is the classic case of there is no there there, as
Gertrude Stein observed in her book Everybodys Autobiography (impress your friends with
that gem). Like an artichoke, if you slice off individual petals or pieces of the Internet,
youll soon have a pile of pieces and no Internet anywhere, and you wont find it among
the pieces. Rather, like the artichoke, the Internet is the overall combination of the pieces.
Those pieces are tied together both physically and logically. The physical aspect is a collection of wires, optical fibers, and microwave radio links that carry digital signals between
computers. The combination of connections forms a redundant network. Computers are
linked to one another in a web that provides multiple signal paths between any two
machines.
The logical side is a set of standards for the signals that travel through that network. The
Internet uses various protocols, depending on what kind of data is being transferred. The
chief protocol and the defining standard of the Internet is TCP/IP, discussed in this chapter.
To work properly, the TCP/IP system requires every computer (or device) connected to
the Internet have a unique address, the IP address. Although simple in conceptthe IP
address is nothing more than a 32-bit binary number, at least for nowwhat that number
means and the future of the entire addressing system are two of the most complex issues
regarding the Internet.
Of course, 32-bit binary numbers probably dont roll off your tongue, and remembering
one for each Web site you want to visit sounds about as fun as an overnight study session
with too much coffee and stress. Thankfully, the developers of the Web concocted the
Domain Name System (DNS), which assigns somewhat more memorable names to every
Web site. Making the DNS system work, however, is one of the great challenges of the
Internet.
CHAPTER 14 INTERNET
History
The most visible piece of the Web is content. After all, if there werent anything worth
surfing for, youd probably turn your attention to something elsesay, surfing for real
when the tsunami warnings go out. Every Web page is, surprisingly, a computer program
written in unique languages understood by your Web browser.
And finally, youve somehow got to make a connection with the Internet. Thats what you
pay for.
History
The place to begin is the beginningand with the Internet, we need to go way back.
Although the Web is the medium of the moment, the Internet has a long history, and
locating its origins depends on how primitive an ancestor you seek.
The thread of the development of the Internet stretches all the way back to 1958, if you
pull on it hard enough. The Internets motherthe organization that gave birth to it
was itself born in the contrail of Sputnik. In October, 1957, the USSR took the world by
surprise by launching the first artificial satellite and made the U.S. suddenly seem technologically backward. In response, President Dwight D. Eisenhower launched the Advanced
Research Project Agency (ARPA) as part of the Department of Defense in January, 1958.
Then, as now, ARPAs work involved a lot of data processing, much of it at various university campuses across the country. Each computer, like the college that hosted it, was a
world unto itself. To work on the computer, you had to be at the college. To share the
results of the work on the computer, you needed a letter carrier with biceps built up from
carrying stacks of nine-track tapes from campus to campus. Information flowed no faster
between computers than did the mail.
Bob Taylor, working at ARPA in 1967, developed the idea of linking together into a
redundant, packet-based network all the computers of major universities participating in
the agencys programs. In October, 1969, the first bytes crossed what was to become
ARPAnet in tests linking Stanford Research Institute and the University of California at
Los Angeles. By December, 1969, four nodes of the fledgling internetworking system
were working.
The system began to assume its current identity with the first use of the Transmission
Control Protocol in a network in July, 1977. As a demonstration, TCP was used to link
together a packet radio network, SATnet, and ARPAnet. Then, in early 1978, the
Transmission Control Protocol was split into a portion that broke messages into packets,
reassembled them after transmission, kept order among the packets, and controlled error
control, called TCP, and a second protocol that concerned itself with the routing of packets through the linkage of the network, called the Internet Protocol (IP). The two together
made TCP/IP, the fundamental protocol of todays Internet.
481
482
If the Internet actually has a birthday, its January 1, 1983, when ARPAnet switched over
from the Network Control Protocol to TCP/IP. (By that time, ARPAnet was only one of
many networks linked by TCP/IP.) To give a friendlier front end to communications with
distant computer systems, Tim Berners-Lee, working at CERN in Geneva in 1990,
invented the World Wide Web.
The final step in the development of todays Internet came in 1991. In that year the
National Science Foundation, which was overseeing the operation of the Internet, lifted
its previous restrictions on its commercial use. The free-market free-for-all began.
Addressing
If you want to go to a store, its useful to know where its at. You could cruise for hours or
days looking for the store you wantand you might never find it. But once you have an
address, you have an anchor. You know where to go, and you should be able to quickly
figure out how to get there.
Although mindless cruising is one of the delights of the Web (after all, isnt that what surfing really is?), when youre more directed in your search (when you know what youre
looking for), having an address can be helpful. Even if you surf, you may want to return to
the same page youve visited before, and knowing where to find it will shorten your journey back (but maybe make it less fun).
Just like physical stores, the Internet uses an address to anchor Web sites. In fact, not only
sites but every device that connects to the Internet gets its own, unique address. The
Internet Protocol itself defines the addressing scheme thats used, so the addresses are naturally called IP addresses.
IP Addresses
Under the Internet Protocol, an address is four bytes, a 32-bit binary number. Usually
youll find it expressed as a series of four numbers called octets. The value of each octet is
expressed in decimal notation, and the individual octets are separated by periods. Hence,
an Internet address looks like this:
162.233.12.148
Every device gets its own, unique address. Do the math, and youll see this scheme allows
for 4,294,967,296 unique devices (which is simply 2 to the 32nd power, exactly what a 32bit binary address means).
Dividing IP addresses into octets is more than a matter of readability. By design, the IP
address is structured to help make finding any given computer easier. The address has two
parts: The first identifies a network connected to the Internet, and the last part identifies a
specific computer attached to the network.
CHAPTER 14 INTERNET
Addressing
For example, the computer with the IP address of 192.168.132.1 actually is computer
number 1 attached to the network named 192.168.132.0.
Structuring IP addresses in this way makes routing packets and messages across the
Internet easier for the routers charged with the job. They only need to find the network
and dump the packets on it. The network then routes the packets to the computer designated in the IP address.
Subnet Masks
You may detect the one small flaw in this addressing systemthe IP address by itself does
not indicate where the split between the network address and the device address occurs.
Although it would be easy to define the last octet as the device address, that was too arbitrary and limiting for the folks who developed the Internet. Such a scheme would limit
any network to only 256 addresses. The Internet designers preferred to permit greater
versatility in configuring networks. After all, the Internet was specifically meant to allow
colleges to exchange information, and most colleges now have substantially more than 256
students, each with his or her own computer tied into the network and Internet.
Instead of fixing the division between network and device addresses, the Internet designers chose to use a second number that defined the boundary. This second number is the
subnet mask. Youll encounter it nearly every time you tangle with IP addresses.
The subnet mask takes the same form as an IP address, four octets in dotted-decimal format. Unlike IP addresses, however, which allow full variation to yield more than four billion distinct numbers, only 32 different subnet masks are allowed in the IP scheme of
things. Once you think about it, this number makes sense because there are only 32 places
to draw the line between network and device addresses. Table 14.1 lists all the valid subnet
masks.
TA B L E 1 4 . 1
Subnet Masks
Mask in Hex
Mask in Decimal
Number of Available Addresses
80.00.00.00
128.0.0.0
2,147,483,648
C0.00.00.00
192.0.0.0
1,073,741,824
E0.00.00.00
224.0.0.0
536,870,912
F0.00.00.00
240.0.0.0
268,435,456
F8.00.00.00
248.0.0.0
134,217,728
FC.00.00.00
252.0.0.0
67,108,864
continues
483
484
TA B L E 1 4 . 1
Continued
Mask in Hex
Mask in Decimal
Number of Available Addresses
FE.00.00.00
254.0.0.0
33,554,432
FF.00.00.00
255.0.0.0
16,777,216
FF.80.00.00
255.128.0.0
8,388,608
FF.C0.00.00
255.192.0.0
4,194,304
FF.E0.00.00
255.224.0.0
2,097,152
FF.F0.00.00
255.240.0.0
1,048,576
FF.F8.00.00
255.248.0.0
524,288
FF.FC.00.00
255.252.0.0
262,144
FF.FE.00.00
255.254.0.0
131,072
FF.FF.00.00
255.255.0.0
65,536
FF.FF.80.00
255.255.128.0
32,768
FF.FF.C0.00
255.255.192.0
16,384
FF.FF.E0.00
255.255.224.0
8192
FF.FF.F0.00
255.255.240.0
4096
FF.FF.F8.00
255.255.248.0
2048
FF.FF.FC.00
255.255.252.0
1024
FF.FF.FE.00
255.255.254.0
512
FF.FF.FF.00
255.255.255.0
256
FF.FF.FF.80
255.255.255.128
128
FF.FF.FF.C0
255.255.255.192
64
FF.FF.FF.E0
255.255.255.224
32
FF.FF.FF.F0
255.255.255.240
16
FF.FF.FF.F8
255.255.255.248
FF.FF.FF.FC
255.255.255.252
FF.FF.FF.FE
255.255.255.254
FF.FF.FF.FF
255.255.255.255
Understanding why these particular masks are the only ones allowed and why they were
chosen requires examining the IP address and subnet mask numbers in their native binary
form.
CHAPTER 14 INTERNET
Addressing
Although most network administrators look at IP addresses in the dotted-decimal format,

computer equipment sees them as a series of 32 ones and zeros. For example, what you
see as the IP address 192.168.132.1 looks like this to your computer:
11000000101010001000010000000001
Divided into octets, this number becomes the following:

11000000.10101000.10000100.00000001
In this format, the number of the subnet mask makes more sense, at least if you look at it
with an engineers eyes. The number allows subnet mask numbers resulting in a dotteddecimal format that is all ones on the left and zeros on the right. For example, the subnet
mask expressed as 255.255.255.128 can also be represented in binary as follows:
11111111.11111111.11111111.10000000
In this form, the columns filled with the ones represent the digits of the IP address that
designate the network. The columns filled with zeros are the digits of the valid computer
identification numbers.
Dont bother memorizing the table of subnet masks. If all youre going to do is home networking, you only need to deal with one subnet mask: 255.255.255.0. Youll find that this
is Microsofts default when you set up TCP/IP on your system, and it is both necessary
and sufficient for most home network setups.
Network Classes
Subnet masks move into prominence when you venture into serious networking. That is,
when you move your network onto the Internet and have your own galaxy of computers
linking into workgroups. To accommodate the really big kids with prodigious needs, the
Internet was designed to be divvied up among governments and businesses in chunks that
depended on need and, more likely, lobbying.
Some organizations need more Internet space than others. Some are able to demand more
space than others. And some are able to use whatever forms of legal blackmail are available to extort more network space than others. In the days that InterNIC (the organization charged with administering the Internet at that time) assigned blocks of Internet
addresses, it followed a classification scheme with five levels: Classes A through E.
Although this scheme is no longer used, the addresses assigned under it remain. You can
classify it as interesting Internet trivia that still creeps into our lives when we least
expect it.
In any case, in each of the first three classes defined under the InterNIC scheme, the
number of addresses available to an organization was defined by the subnet mask. In
485
486
addition, InterNIC defined two more classes for special purposes: multicasting (sending
packets to multiple computers but not all on the Internet) and experimental purposes.
These classes were assigned their own ranges of special IP addresses. The five classes are
as follows:
Class AThese Internet addresses use a subnet mask of 255.0.0.0. The first bit in
a Class A address is always zero, so Class A addresses always start with the first octet
in the range of 0 to 126, inclusive. This classification leaves seven bits to identify the
network and 24 bits to identify individual devices.
Class BThese Internet addresses use a subnet mask of 255.255.0.0. The first two
bits of a Class B address are a one followed by a zero, so the first octet of a Class B
address always falls in the range 128 to 191, inclusive. This classification leaves 14
bits to identify the network and 16 bits to identify individual devices.
Class CThese Internet addresses use a subnet mask of 255.255.255.0. The first
three bits of a Class C address are always two ones followed by a zero, so Class C
addresses always fall in the range with the first octet of 192 to 223, inclusive. This
classification leaves 21 bits to identify the network and eight bits to identify individual devices.
Class DThese addresses always start with binary addresses of three ones followed
by a zero, which translates into a first octet in dotted-decimal notation in the range
224 to 239, inclusive. The remaining 28 bits in the address identify the group of
computers for which the multicast is meant.
Class EThese addresses always start with a binary address of four ones, which
translates into a first octet in dotted-decimal notation in the range 240 to 255, inclusive. As with Class D, the remaining 28 bits in the address identify the group of
computers for which the multicast is meant.
CIDR
Subnet masks are cumbersome in everyday use on the Internet. To help make routing
messages between computer networks more efficient, Internet workers developed Classless
Inter-Domain Routing (CIDR) to provide more flexibility than was possible with the subnet
mask scheme. The CIDR system is now used by virtually every computer on the Internets
backbone to route messages to their destinations.
Basically, the CIDR system distills the four-byte subnet mask into a single number
appended to an IP address called a network prefix. The number in the network prefix
describes the number of bits in the address that constitute the network designation part of
the address, much as the subnet mask does. For example, in the CIDR network address
192.168.132.7/24
CHAPTER 14 INTERNET
Addressing
the first 24 bits indicate the address of a network, and the last eight bits identify an individual computer.
IPv6
Although four billion is a lot of computers for a network, the Internets administrators see
the reserve of IP addresses quickly disappearing. They fear that sometime soon the world
will run out of them and no new computer can connect.
They havent been asleep, however. They have developed a revised version of the Internet
Protocol to accommodate longer addresses to break through the 32-bit addressing limit.
Called Internet Protocol Version 6 (we used version 4 today), the revision allows for IP
addresses 128-bits long. The result is that IPv6 accommodates more addresses than it is
convenient to write down. Every person in the world could have four billion computers,
each with its own IPv6 address, and there would still be 16 sextillion times more addresses
available.
The revisions of IPv6 dont stop with longer addresses. Under IPv6, the packet header
allows messages to be identified as part of a particular flow, such as a stream of audio or
video. Properly identified, the packets can be routed to follow the same path to help them
get reconstructed as a real-time stream. The header also includes extensions for authentication, error control, and privacy.
The IPv6 spec is designed to allow compatibility with todays IP addressing system. The
lowest 32 bits of the IPv6 address can function as a current IP address and routed appropriately.
What Address to Use

When setting up a home or small office network, at some time or another you will be
confronted with the choice of IP addresses to useits one of those unwelcome choices
that is given every network administrator. As far as I can tell, no readily available source
even hints at what IP address you should use. But the choice is critical, and the people
who govern such things in the Internet publish the addresses you should use.
In fact, the Internet Assigned Numbers Authority (IANA, which you can find on the Web
at www.iana.org) reserves three blocks of IP addresses for use by private networksthat
is, those that dont intend on directly connecting to the Web. Because its likely you will
connect only through a gateway at your ISP, your home network falls into the private network class, and these reserved addresses are the ones you should choose from. Table 14.2
lists the addresses IANA reserves for private networks.
487
488
TA B L E 1 4 . 2
Address Ranges Reserved for Private Networks
Start Address
End Address
Addresses in Range
10.0.0.0
10.255.255.255
16,777,216
172.16.0.0
172.31.255.255
1,048,576
192.168.0.0
192.168.255.255
65,536
Clearly, any of these three ranges will have more than enough room for any conceivable
home network. Microsoft uses the last of these, the range starting at 192.168.0.0, for the
private networks it automatically sets up for home use.
Certainly youre not limited to these values for your own private network. You could simply create your own IP address. If youre not too clever for you own good, you may get
away with it. Coming up with a valid address is not difficultbut its not a good idea
either. Internet addresses are assigned, and the Internet governing bodies go to lengths to
be sure theres no conflicts.
Addresses You Cannot Use

The IP naming rules dictate that you cannot use certain addresses for computers on a network. These fall at the two ends of the number range in the fourth octet. That is,
addresses ending in zero, such as 192.168.155.0, and those ending in 255, such as
192.168.154.255, cannot be used as addresses for computers or other devices connected
to a network. These addresses have a specifically defined meaning in the IP system.
Addresses ending in a zero refer to the network itself rather than any specific computer or
device connected to it. Addresses ending in 255 are used to broadcast messages to all
devices in the network, so all devices in a network will receive packets with the network
address and a 255 at its end.
The rule is simple: Never use these addresses when assigning IP addresses to any device in
your network.
DHCP
If you find IP addresses confusing, youre not the only one. Keeping track of all the
addresses used by a network can be confusing, indeed. What you really need is someone
who excels at organization, who rigidly assigns addresses and keeps track of every detail as
if he were a machine. In fact, a machine such as a computer would be a good choice to
take over the job.
Using the Dynamic Host Configuration Protocol (DHCP), you can move the responsibility
for assigning and organizing the IP addresses your network uses to one of its servers.
CHAPTER 14 INTERNET
Domain Name System
DHCP is an automatic method for assigning addresses to devices. When a device wants to
join the network, it queries the DHCP server, and the server sends back a unique IP
address for the device.
Not just any address will do. In the Microsoft scheme of things, the addresses assigned by
a server are drawn from within a scope, a range of no more than 255 contiguous addresses.
All the devices in a workgroup must be within the same scope, although a network may
have many intercommunicating scopes.
Setting up a DHCP server usually is more work than most normal people want to do.
Thats why its usually left to network administrators. But if you buy an inexpensive router
in order to share a high-speed Internet connection, odds are it has a DHCP server built in
to it. When you log in to your network in preparation for sharing your connection, the
DHCP server automatically sends your computer its own unique IP address so it can join
the network.
Note that you should have only one DHCP server in a network. If you install a dedicated
server to act as your DHCP server and you install a router for Internet sharing, the two
DHCP servers may come into conflict, possibly preventing your network from operating,
or just preventing some computers from seeing others on the network. To avoid problems,
make sure you have only one DHCP server.
Domain Name System

Although DHCP does a good job of hiding your own computers IP address from you, it
does nothing to make IP addresses manageable on the Web. So that you dont have to
type into your browser the IP address of each computer you want to visit on the Web (you
can if you want, by the way), the Webs creators developed the Domain Name System
(DNS).
From your perspective, DNS works by assigning a structured name to every Web site, the
familiar whatever-dot-com you use every day. In the language of the Internet, that dotcom is called the domain name of the Web site.
Most domain names take the form of a word, a period, and another few letters. Those letters after the period comprise the top-level domain, the primary organizing structure of the
Web. You cant use just anything as a top-level domain. The organization charged with
administering Web names, the Internet Corporation for Assigned Names and Numbers
(ICANN), maintains tight control of top-level domains.
There are two kinds of top-level domains: organizational and national. Originally, only six
organizational top-level domains were allowed, but on November 16, 2000, ICANN
added seven more. Table 14.3 lists the currently recognized top-level domains.
489
490
TA B L E 1 4 . 3
Currently Recognized Top-Level Domains
TLD
Use
Original Top-Level Domains
.com
Commercial use
.edu
Educational institutions
.gov
Government
.mil
Armed forces
.net
Service providers
.org
Organizations
Added November 18, 2000
.aero
Air-transport industry
.biz
Businesses
.coop
Cooperatives
.info
Anything
.museum
Museums
.name
Individuals
.pro
Professionals
In addition, each nation in the world is also given its own top-level domain, a twocharacter country code. These are listed in Table 14.4.
TA B L E 1 4 . 4
Internet Country Code Top-Level Domains
Country Code
Nation
.ad
Andorra
.ae
United Arab Emirates
.af
Afghanistan
.ag
Antigua and Barbuda
.ai
Anguilla
.al
Albania
.am
Armenia
.an
Netherlands Antilles
.ao
Angola
CHAPTER 14 INTERNET
Domain Name System
Country Code
Nation
.aq
Antarctica
.ar
Argentina
.as
American Samoa
.at
Austria
.au
Australia
.aw
Aruba
.az
Azerbaijan
.ba
Bosnia and Herzegovina
.bb
Barbados
.bd
Bangladesh
.be
Belgium
.bf
Burkina Faso
.bg
Bulgaria
.bh
Bahrain
.bi
Burundi
.bj
Benin
.bm
Bermuda
.bn
Brunei Darussalam
.bo
Bolivia
.br
Brazil
.bs
Bahamas
.bt
Bhutan
.bv
Bouvet Island
.bw
Botswana
.by
Belarus
.bz
Belize
.ca
Canada
.cc
Cocos (Keeling) Islands
.cd
Democratic Peoples Republic of Congo

continues
491
492
TA B L E 1 4 . 4
Continued
Country Code
Nation
.cf
Central African Republic
.cg
Republic of Congo
.ch
Switzerland
.ci
Cote dIvoire
.ck
Cook Islands
.cl
Chile
.cm
Cameroon
.cn
China
.co
Colombia
.cr
Costa Rica
.cu
Cuba
.cv
Cap Verde
.cx
Christmas Island
.cy
Cyprus
.cz
Czech Republic
.de
Germany
.dj
Djibouti
.dk
Denmark
.dm
Dominica
.do
Dominican Republic
.dz
Algeria
.ec
Ecuador
.ee
Estonia
.eg
Egypt
.eh
Western Sahara
.er
Eritrea
.es
Spain
.et
Ethiopia
.fi
Finland
CHAPTER 14 INTERNET
Domain Name System
Country Code
Nation
.fj
Fiji
.fk
Falkland Islands (Malvinas)
.fm
Federal State of Micronesia
.fo
Faeroe Islands
.fr
France
.ga
Gabon
.gd
Grenada
.ge
Georgia
.gf
French Guiana
.gg
Guernsey
.gh
Ghana
.gi
Gibraltar
.gl
Greenland
.gm
Gambia
.gn
Guinea
.gp
Guadeloupe
.gq
Equatorial Guinea
.gr
Greece
.gs
South Georgia and the South Sandwich Islands
.gt
Guatemala
.gu
Guam
.gw
Guinea-Bissau
.gy
Guyana
.hk
Hong Kong
.hm
Heard and McDonald Islands
.hn
Honduras
.hr
Croatia/Hrvatska
.ht
Haiti
.hu
Hungary
continues
493
494
TA B L E 1 4 . 4
Continued
Country Code
Nation
.id
Indonesia
.ie
Ireland
.il
Israel
.im
Isle of Man
.in
India
.io
British Indian Ocean Territory
.iq
Iraq
.ir
Iran (Islamic Republic of)
.is
Iceland
.it
Italy
.je
Jersey
.jm
Jamaica
.jo
Jordan
.jp
Japan
.ke
Kenya
.kg
Kyrgyzstan
.kh
Cambodia
.ki
Kiribati
.km
Comoros
.kn
Saint Kitts and Nevis
.kp
Democratic Peoples Republic of Korea
.kr
Republic of Korea
.kw
Kuwait
.ky
Cayman Islands
.kz
Kazakhstan
.la
Peoples Democratic Republic of Laos
.lb
Lebanon
.lc
Saint Lucia
.li
Liechtenstein
CHAPTER 14 INTERNET
Domain Name System
Country Code
Nation
.lk
Sri Lanka
.lr
Liberia
.ls
Lesotho
.lt
Lithuania
.lu
Luxembourg
.lv
Latvia
.ly
Libyan Arab Jamahiriya
.ma
Morocco
.mc
Monaco
.md
Republic of Moldova
.mg
Madagascar
.mh
Marshall Islands
.mk
Macedonia
.ml
Mali
.mm
Myanmar
.mn
Mongolia
.mo
Macau
.mp
Northern Mariana Islands
.mq
Martinique
.mr
Mauritania
.ms
Montserrat
.mt
Malta
.mu
Mauritius
.mv
Maldives
.mw
Malawi
.mx
Mexico
.my
Malaysia
.mz
Mozambique
.na
Namibia
continues
495
496
TA B L E 1 4 . 4
Continued
Country Code
Nation
.nc
New Caledonia
.ne
Niger
.nf
Norfolk Island
.ng
Nigeria
.ni
Nicaragua
.nl
Netherlands
.no
Norway
.np
Nepal
.nr
Nauru
.nu
Niue
.nz
New Zealand
.om
Oman
.pa
Panama
.pe
Peru
.pf
French Polynesia
.pg
Papua New Guinea
.ph
Philippines
.pk
Pakistan
.pl
Poland
.pm
St. Pierre and Miquelon
.pn
Pitcairn Island
.pr
Puerto Rico
.ps
Palestinian Territories
.pt
Portugal
.pw
Palau
.py
Paraguay
.qa
Qatar
.re
Reunion Island
.ro
Romania
CHAPTER 14 INTERNET
Domain Name System
Country Code
Nation
.ru
Russian Federation
.rw
Rwanda
.sa
Saudi Arabia
.sb
Solomon Islands
.sc
Seychelles
.sd
Sudan
.se
Sweden
.sg
Singapore
.sh
St. Helena
.si
Slovenia
.sj
Svalbard and Jan Mayen Islands
.sk
Slovak Republic
.sl
Sierra Leone
.sm
San Marino
.sn
Senegal
.so
Somalia
.sr
Suriname
.st
Sao Tome and Principe
.sv
El Salvador
.sy
Syrian Arab Republic
.sz
Swaziland
.tc
Turks and Caicos Islands
.td
Chad
.tf
French Southern Territories
.tg
Togo
.th
Thailand
.tj
Tajikistan
.tk
Tokelau
.tm
Turkmenistan
.tn
Tunisia
continues
497
498
TA B L E 1 4 . 4
Continued
Country Code
Nation
.to
Tonga
.tp
East Timor
.tr
Turkey
.tt
Trinidad and Tobago
.tv
Tuvalu
.tw
Taiwan
.tz
Tanzania
.ua
Ukraine
.ug
Uganda
.uk
United Kingdom
.um
US Minor Outlying Islands
.us
United States
.uy
Uruguay
.uz
Uzbekistan
.va
Holy See (City Vatican State)
.vc
Saint Vincent and the Grenadines
.ve
Venezuela
.vg
Virgin Islands (British)
.vi
Virgin Islands (USA)
.vn
Vietnam
.vu
Vanuatu
.wf
Wallis and Futuna Islands
.ws
Western Samoa
.ye
Yemen
.yt
Mayotte
.yu
Yugoslavia
.za
South Africa
.zm
Zambia
.zr
Zaire
.zw
Zimbabwe
CHAPTER 14 INTERNET
Domain Name System
To the left of the period is the name of the actual domain assigned the Web site through a
registry. If there is more than one period in the name, the leftmost portion is a subdomain
of the next domain name to the right. The DNS system allows for multiple subdomains.
Each subdomain (or domain, if there are no subdomains) specifies an actual server on the
network.
To the right of the top-level domain is the directory path to a particular file on the designated server containing a Web page or other data. In Internet lingo, this composite construction of the domain name and directory path is called a Uniform Resource Locator,
because the name itself holds all the information computers on the Web need in order to
find a particular page or file.
The best view of the Internet comes with following a packet sent from your computer.
When you log in to a Web site, you actually send a command to a distant server telling it
to download a page of data to your computer. Your Web browser packages that command
into a packet labeled with the address of the server storing the page that you want. Your
computer sends the packet to your modem (or terminal adapter), which transmits it across
your telephone or other connection to your Internet Service Provider (ISP).
How DNS Works

The instant you press your mouse button with the cursor pointing at a particularly juicy
image on your monitor, your computer drops everything to obey your command. The
mouse sends your microprocessor an interrupt to make it pay immediate attention. The
mouse driver checks to see whether a signal has come in saying youve pressed the mouse
button. The driver passes this vital information to your operating system, which takes a
peek into its private memory to see what location value it has stored for the location of
your mouses cursor on the screen. The operating system then checks to see whether the
mouse location corresponds to a hotspot on the Web page youre viewing, indicating to
your browser that there is a hyperlink instruction to send you to a new location on the
Web. When there is, the fun on the Web begins.
Your browser has no idea where to find the page you want on the Web. All it has is a
nameits sort of like finding an isolated name on a message pad when you awake from a
drunken stupor. You recognize it as a name, but you dont know why you wrote down the
name or how to get in touch with the person to find out.
You might start with the white pages of your telephone book, but your computer cant
even open a book, let alone look something up. The only thing it can do is send out an
electrical signal down the connection with the Internet. It doesnt send out just any signal.
It sends out the name of the Web site youre looking for in a special data packet. Rather
than the white pages, the name goes to a resolver.
499
500
A resolver is not a special machine. Rather, the term resolver defines a particular function
of a special kind of server on the network, called a name server. The resolver does what its
name saysit tries to resolve the address of a site on the Web. It looks at the name your
computer has sent to it in a standard form known as a Uniform Resource Locator (URL).
Like a mailman sorting mail and looking at the bottom of the address first (for the ZIP
Code and state), the resolver examines the last part of the URL firstthe part of the
name to the right of the rightmost periodthe familiar .com, .org, or .edu. This portion
of the name is the top-level domain, which tells the resolver how to find the location of the
Web site. In Internet terms, the top-level domain is about as general as describing a creature as being in the animal kingdom.
The top-level domain doesnt tell the resolver where to look for the Web site. Rather, it
tells the resolver where to look for a list of site names in the top-level domain. Resolvers
arent stupid. If theyve looked up a top-level domain before, they probably already know
where to look. If not, however, they call on one of the 13 root name servers to tell them
which servers store the information about each top-level domain.
The root name server, or, more likely, the resolver, passes your request for the Web site to
one of the name servers assigned to the top-level domain of the URL youre looking for.
Hundreds of thousands of servers may track this information. Its kept in multiple copies
for speed and reliability. Speed because one server is not burdened with finding every
requested URL in its domain, and reliability because if one server becomes unavailable,
there are hundreds of others that can take its place.
This server matches your requested URL with the domain name server (also abbreviated
DNS) that handles the Web site. The server sends the requested URL to the IP address
a block of four bytes of binary codeof the DNS. The DNS knows all the names of the
Web sites it serves. It passes the IP address of the Web site you want back to your computer so it can use this address to find the page you want.
When your computer signals to the IP address, it sends a request for the page listed in the
hotspot you clicked. The server at the Web site diligently finds the page and passes it back
to your computer. Your operating systems passes it to your browser, which formats the
page for the screen and passes it back to your operating system, which sends it, in turn, to
your display driver and then your monitor screen.
All these requests travel from server to server with light speed, so everything happens fast.
Your computer should know the IP address of the page you want and start loading the
page in a fraction of a second. Meanwhile, youve probably become impatient and clicked
on something else, starting the whole process over again.
CHAPTER 14 INTERNET
Domain Name System
Root Name Servers

The part of the Web in charge of identifying each site and getting its address to you is
called the Root Name Server System. The master plan that makes it work is the DNS protocol, which describes the packets that need to be exchanged and provides the roadmap for
them to follow. The root name servers hold the key to locating the indexes containing the
IP addresses you need. The root zone file is the index itself.
The root name servers are arguably the most important computers on the World Wide
Web. Only they store the official records of the locations of the registries for each top-level
domain, the rightmost part of each Web address. Because this information is so vital to the
operation of the Web, its stored not in triplicate but in 13 duplicate copies in separate
computers spread across the world.
Well, not quite. The 13 root name servers actually represent only six distinct geographic
locations. Six are clustered around Washington, D.C., two are co-located (and co-operated)
in Marina Del Ray, California, two are in the Silicon Valley area, one is in Japan, one is in
the U.K., and one is in Sweden. Table 14.5 lists the 13 root name servers.
TA B L E 1 4 . 5
The 13 Root Name Servers of the World Wide Web
Name Organization
City, State/Province
Country
URL
Network Solutions, Inc.
Herndon, VA
USA
http://www.netsol.com
Information Sciences
Institute, University of
Southern California
Marina Del Rey, CA
USA
http://www.isi.edu
PSINet
Herndon, VA
USA
http://www.psi.net
University of Maryland
College Park, MD
USA
http://www.umd.edu
National Aeronautics and

Space Administration
Mountain View, CA
USA
http://www.nasa.gov
Internet Software
Consortium
Palo Alto, CA
USA
http://www.isc.org
Defense Information
Agency Systems
Vienna, VA
USA
http://nic.mil
Army Research Laboratory
Aberdeen, MD
USA
http://www.arl.mil
NORDUNet
Stockholm
Sweden
http://www.nordu.net
Network Solutions, Inc.
Herndon, VA
USA
http://www.netsol.com
RIPE NCC (Rseaux IP

Europens)
London
UK
http://www.ripe.net
continues
501
502
TA B L E 1 4 . 5
Continued
Name Organization
City, State/Province
Country
URL
University of Southern
California
Marina Del Rey, CA
USA
http://www.isi.edu
WIDE
Tokyo
Japan
http://www.wide.ad.jp
WINS
Theres another piece to the IP naming system: the names your computers wear when you
access them over your own network using the networking capabilities of Windows. You
assign these names using the Windows Internet Name System (WINS). This system is
responsible for converting the names you assign to IP addresses for routing messages
through networks using the Internet Protocol. In effect, WINS works like DNS at the
local level.
In the WINS system, you assign your computer a name when you set up networking on
that machine. The computer then sends its name to the server, and the server stores the
names you assign in a database, which the server references to resolve requests for IP
addresses.
Routing
An address does you no good if you have no way of getting there. You need wheels (and
likely gas money).
The wheels of the Internet is the Transport Control Protocol/Internet Protocol, or as it is
more commonly known, TCP/IP.
TCP/IP sees everything in terms of packets. Instead of moving data in a long stream like
unraveling a roll of movie film, the protocol breaks it into pieces. Having a bunch of short
chunks automatically ensures that there will be breaks in the flow of data during which
other computers can negotiate for time to send their own packets. At the distant end of
the connection, the packets get reassembled to put the data back into its original form.
Each has a predefined structure, with a header that contains address, routing, and control
information and a payload of data. The payload moves through the network intact and
unexamined, so its content is irrelevant to the network. Packets might contain anything,
from program code for Unix computers, to bits of video images, to cream cheese on celery (if you could fit that into computer data).
Moving packets around the Internet is a lot like modern psychotherapy. It is nondirected.
The packets simply ramble around until they happen upon the place where they are
CHAPTER 14 INTERNET
Web Pages
going. They may follow any one of a near infinite number of paths between the two communicating computers. The Internet imposes no fixed structure. Thats one of its greatest
strengthsbecause messages dont have an assigned path, an interrupt of any path wont
stop data from flowing. It simply finds another path to its destination.
Of course, having packets floating all over the Internet is not the most efficient way of
moving information. Consequently, servers build tables of paths to send the messages
along, routing them according to IP address. When a message first goes to an IP address,
an Internet server checks the path it followed and can reuse the same path (or start the
packet along its way using the same path) for subsequent packets.
The wonderful thing about TCP/IP is that any computer system can use it.
Microprocessor types, architectures, and programming languages mean nothing to
TCP/IP. Think of TCP/IP packets as being the shipping containers of the Internet. A
computer at one end of the connection fills a packet up, and another machine at the far
end of the connection empties it out. It doesnt matter to the freight line whats inside
(although customs may take a peek if it crosses bordersthe Internets equivalent of customs is the firewall). Using TCP/IP, the Internet will carry anything. It doesnt matter
whether its useful or even compatible with the recipient system. Thats a matter left to the
two communicating systems. The network doesnt care.
Web Pages
The World Wide Web is the most visually complicated and compelling aspect of the
Internet. Despite its appearances, however, the Web is nothing more than another file
transfer protocol. When you call up a page from the Web, the remote server simply
downloads a file to your computer. Your Web browser then decodes the page, executing
commands embedded in it to alter the typeface and to display images at the appropriate
place. Most browsers cache several file pages (or even megabytes of them) so that when
you step back, you need not wait for the same page to download once again.
The commands for displaying text use their own language, called the Hypertext Markup
Language (HTML). As exotic and daunting as HTML sounds, its nothing more than a
coding system that combines formatting information in textual form with the readable text
of a document. Your browser reads the formatting commands, which are set off by a special prefix so that the browser knows they are commands, and organizes the text in accordance with them, arranging it on the page, selecting the appropriate font and emphasis,
and intermixing graphical elements. Writing in HTML is only a matter of knowing the
right codes and where to put them. Web authoring tools embed the proper commands
using menu-driven interfaces so that you dont have to do the memorization.
503
504
Service Providers
In truth, the Internet was not designed to link computers but rather to tie together computer
networks. As its name implies, the Internet allows data to flow between networks. Even if
you only have a single computer, when you connect with the Internet, you must run a network protocol the same as if you had slung miles of Ethernet cable through your home
and office. Whether you like it or not, you end up tangled in the web of networking when
you connect to the Internet.
The ISP actually operates as a message forwarder. At the ISP, your message gets combined
with those from other computers and sent through a higher-speed connection (at least you
should hope it is a high-speed connection) to yet another concentrator that eventually
sends your packet to one of five regional centers (located in New York, Chicago, San
Francisco, Los Angeles, and Maryland). There, the major Internet carriers exchange signals, routing the packets from your modem to the carrier that will haul them to their destination based on their Internet address.
Performance Limits
Linking to the Internet requires both hardware and software. The hardware runs a wide
gamut. Most people connect through standard telephone lines using a modem.
Okay, so your Internet access through your modem or digital connection isnt as fast as
youd like. Welcome to the club. As the Duchess of Windsor never said, You can never
be too rich or thin or have an Interconnection thats fast enough. Everyone would like
Web pages to download instantly. Barring that, theyd like them to load in a few seconds.
Barring that, theyd just like them to load before the next Ice Age.
The most tempting way to increase your Internet speed is to update your modemmove
from dial-up to a broadband service such as DSL, cable, or satellite. Once you do, you
may discover the dirty secret of the Internet: Youre working on the wrong bottleneck.
You may have a high-speed connection, but the server you want to download pages from
may be someones ten-year-old computer hogtied by a similar-vintage 9600bps modem.
Or a server with heady-duty equipment may be overwhelmed by more requests than it can
handle. Remember, your packets may get slowed anywhere along their way through the
Web.
You can easily check your Internet bottleneck and see what you can do about it. Pick a
large file and download it at your normal online time. Then, pry yourself out of bed early
and try downloading the same file at 6 a.m. EST or earlier when Internet traffic is likely
to be low. If you notice an appreciable difference in response and download times, a faster
modem wont likely make your online sessions substantially speedier. The constraints
arent in your computer but in the server and network itself.
CHAPTER 14 INTERNET
Service Providers
Another way to check is with one of the many services designed for checking DSL speed.
To find one, simply perform a Web search for DSL speed test. One choice is
www.dslreports.com.
Security
As originally conceived, the Internet is not just a means for moving messages between
computers. It was designed as a link between computer systems that allowed scientists to
share machines. One researcher in Boston could, for example, run programs on a computer system in San Francisco. Commands for computer systems move across wires just as
easily as words and images. To the computer and the Internet, they are all just data.
Much of the expense businesses put into connecting to the Internet involves undoing the
work of the original Internet creators. The first thing they install is a firewall, which
blocks outsiders from taking control of the businesss internal computer network. They
must remain constantly vigilant that some creative soul doesnt discover yet another flaw
in the security systems built into the Internet itself.
Can someone break into your computer through the Internet? Its certainly possible.
Truth be told, however, rummaging through someones computer is about as interesting as
burrowing into his sock drawer. Moreover, the number of computers out there makes it
statistically unlikely any given errant James Bond will commandeer your computer, particularly when theres stuff much more interesting (and challenging to break into) such as the
networks of multibillion dollar companies, colleges, government agencies, and the military.
The one weakness to this argument is that it assumes whoever would break into your
computer uses a degree of intelligence. Even a dull, uninteresting computer loaded with
naught but a two-edition-old copy of Office can be the target of the computer terrorist.
Generally, someone whose thinking process got stalled on issues of morality, the computer
terrorist doesnt target you as much as the rest of the world that causes him so much frustration or boredom. His digital equivalent of a bomb is the computer virus.
A computer virus is program code added to your computer without your permission. The
name, as a metaphor for human disease, is apt. As with a human virus, a computer virus
cannot reproduce by itselfit takes command of your computer and uses its resources to
duplicate itself. Computer viruses are contagious in that they can be passed along from
one machine to another. And computer viruses vary in their effects, from deadly (wiping
out the entire contents of your hard disk) to trivial (posting a message on your screen).
But computer viruses are nothing more than digital code, and they are machine specific.
Neither you nor your toaster nor your PDA can catch a computer virus from your computer.
505
506
Most computer viruses latch onto your computer and lie in wait. When a specific event
occursfor example, a key datethey swing into action, performing whatever dreadful
act their designers got a chuckle from. To continue infecting other computers, they also
clone themselves and copy themselves to whatever disks you use in your computer. In
general, viruses add their code to another program in your computer. They cant do anything until the program they attach themselves to begins running. Virus writers like to
attach their viruses to parts of the operating system so that the code will load every time
you run your computer. Because antivirus programs and operating system now readily
detect such viruses, the virus terrorists have developed other tactics. One of the latest is
the macro-virus, which runs as a macro to a program. In effect, the virus is written in a
higher-level language that escapes detection by the antivirus software.
Viruses get into your computer because you let them. They come through any connection
your computer has with the outside world, including floppy disks and going online.
Browsing Web pages ordinarily wont put you at risk because HTTP doesnt pass along
executable programs. Plug-ins may, however. Whenever you download a file, you run a
risk of bringing a virus with it. Software and drivers that you download are the most likely
carriers. Most Webmasters do their best to ensure that they dont pass along viruses.
However, you should always be wary when you download a program from a less reputable
site. The same warning applies to e-mail from unknown senders.
There is no such thing as a sub-band or sub-carrier virus that sneaks into your computer
through a sub-band of your modems transmissions. Even were it possible to fiddle with
the operation of a modem and add a new, invisible modulation to it, the information
encoded on it could never get to your computer. Every byte from an analog modem must
go through the UART in the modem or serial port, then be read by your computers
microprocessor. The modem has no facility to link a sideband signal (even if there were
such a thing) to that data stream.
PART 4
Storage
The difference between genius and mere intelligence is storage. The quick-witted react fast,
but the true genius can call upon memories, experiences, and knowledge to find real
answersthe difference between pressing a button fast and having the insight to know which
button to press.
Computers are no different. Without memory, a computer is nothing more than a switchboard. All its reactions would have to be hard-wired in. The machine could not read through
programs or retain data. It would be stuck in a persistent vegetative state, kept alive by electricity but able to react only autonomously.
The technology for remembering did not originate with computers. Ever since people had
thoughts they deemed worth preserving, they have used mechanical means for aiding their
memories. When early humans first took charcoal and ochre to sketch on the walls of their
ritual caves, they were making mechanical memories of hunts and ceremonies, of men, bison,
and mastodons. Primitive, perhaps, but even with all our modern technologies, we have yet to
make a record of ourselves that has lasted as long as cave drawings.
Perhaps inadvertently, those drawings have survived for the long term. But we, as thinking
people, rely on another kind of memoryshort term. When youre working on a problem,
you have to hold a piece of it in your mindsay, the one to carry when youre adding a column of numbers or the telephone number you looked up and need to dial.
Computers have the same need for two kinds of memoryboth long term and short term.
Long-term storage holds the stuff you hope your computer never forgetsyour operating
system, your programs, your data, and the MP3 music youve acquired. Your computer also
needs short-term memory while it works on your programs.
Unlike human beings, whose memory is not understood well enough to distinguish the different processes required for the long and short term, computers use widely different but
well-understood technologies for their long- and short-term storage. This part of the book
will examine how the two kinds of memory fit and work together. Well also look at the individual technologies involved in short-term and long-term storage.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
15
Principles: Freezing Time

The term memory covers a lot of territory, even when confined to the
computer field. Strictly speaking, memory is anything that holds data,
even a single bit. That memory can take a variety of forms. A binary
storage system, the kind used by todays computers, can be built from
marbles, marzipan, or metal-oxide semiconductors. Not all forms of
memory work with equal efficacy (as youll soon see), but the concept is
the same with all of thempreserving bits of information in a recognizable and usable form. Some forms of memory are just easier for an electronic microprocessor to recognize and manipulate. On the other hand,
other sorts of memory may roll or taste better.
When discussing the remembering capabilities of computers, engineers
usually distinguish between memory and storage. Although both let your
computer recall details and data, the two have different purposes and use
different technologiesand a raft of different terminologies. Although
the two concepts can be distinguished (and named) in many ways, the
most useful is as primary and secondary storage.
Defining Memory
The stuff that most people call computer memory in a specific sense
functions as your computers primary storage. That is, the contents of the
storage system are in a form that your computers microprocessor can
immediately access, ready to be used. It can be accessed using only electricity at the speed of electricity (which can be nearly the speed of light).
It is the memory used by your computers microprocessor to hold the
data and program code thats used during the active execution of programs, the microprocessors main job. For this reason, primary storage is
sometimes called working memory.
512
The immediacy of primary memory requires that your microprocessor be able to find any
given value without poring through huge blocks of data. The microprocessor must access
any value at random. Consequently, most people refer to the working memory in their
computers as random access memory, or RAM, although RAM has a more specific definition
when applied to memory technologies, as youll see later on.
No matter the name you use for it, primary storage is in effect the short-term memory of
your computer. Its easy to get at but tends to be limited in capacityat least compared to
other kinds of storage.
The alternate kind of storage is termed secondary storage. In most computers, disks and
tape systems serve as the secondary storage system. They function as the machines longterm memory. Not only does disk and tape memory maintain information that must be
kept for a long time, but it also holds the bulk of the information that the computer deals
with. Secondary storage may be tens, hundreds, or thousands of times larger than primary
storage. Secondary storage is often termed mass storage because of its voluminous capacity:
It stores a huge mass of data.
Secondary storage is one extra step away from your computers microprocessor. Your computer must transfer the information in secondary storage into its primary storage system
in order to work on it. Secondary storage also adds a complication to the hardware. Most
secondary storage is electromechanical. In addition to moving electrical signals, it also
involves physically moving a disk or tape to provide access to information. Because
mechanical things generally move slower than electrical signals (except in science fiction),
secondary storage is slower than primary storage, typically by a factor of a thousand or
more.
In other words, the most important aspect of the primary storage system in your computer is access speed, although you want to have as much of it as possible. The most
important aspect of secondary storage is capacity, although you want it to be as fast as possible.
Why does your computer need two kinds of memory? The role of secondary storage is
obvious. Its the only place that the computer can keep things without worrying about a
power glitch making its memory disappear. Its the only kind of storage that allows you to
freely exchange blocks of data, such as distributing programs.
The need for primary storage may not seem as obvious. The reason is purely speed. As a
microprocessor operates, it needs a constant stream of instructionsthe program that it
executes as it operatesand data that the instructions tell it to manipulate. That stuff has
to come from somewhere. If it were on disk, the microprocessor might have to wait for
each byte to be found before it could carry out its operation. On the average, that would
take about nine milliseconds per instruction. Todays microprocessors can run through
CHAPTER 15 PRINCIPLES: FREEZING TIME
Defining Memory
instructions and data about twenty million times faster than that. Looking stuff up on disk
would clearly slow things down.
Electronic memory bridges the gap. Your computer, under the direction of its operating
system, copies the instructions and data that are recorded on your disk to solid-state
memory. Your microprocessor can then operate at full-speed (or nearly so), millions of
times faster than if it had to move bytes between disk and its registers.
Your computer never knows where its next byte is coming from. The microprocessor
might need to read or write any byte in the program or the mass of data it is working
with. If youre running several programs and shift between them, the need for bytes can
be far ranging indeed. To prevent the million-fold slowdown of the microprocessor, modern computer systems are designed to keep all (or most) of each program and its data in
solid-state primary storage. Thats why you need dozens or hundreds of megabytes of primary storage in a modern computer equipped with Windows.
Volatility
In all all-too-human memories, one characteristic separates out short-term and long-term
memories. The former are fleeting. If a given fact or observation doesnt make it into your
long-term memory, youll quickly forget whatever it wasfor example, the name that
went with the face so quickly introduced to you at a party.
A computers primary storage is similar. The contents can be fleeting. With computers,
however, technology rather than attention determines what gets remembered and what is
forgotten. For computers, the reaction to an interruption in electrical supply defines the
difference between short- and long-term remembering capabilities. The technical term
used to describe the difference is memory volatility. Computer memory is classed either as
nonvolatile or volatile.
Volatile memory is, like worldly glory, transitory. It lasts not the three score years and ten of
human existence or the fifteen minutes of fame. It survives only as long as does its source
of power. Remove power from volatile memory, and its contents evaporate in microseconds. The main memory system in nearly every computer is volatile.
Nonvolatile memory is exactly what you expect memory to be, forever. Once you store
something in nonvolatile memory, it stays there until you change it. Neither rain, nor
sleet, nor dark of night, nor a power failure affects nonvolatile memory. Types of nonvolatile memory include magnetic storage (tape and disk drives) and special forms of
memory chips (read-only memory and Flash memory).
Nonvolatile memory can be simulated by providing backup power to volatile memory systemsa technology commonly used in the CMOS configuration memory systems used in
most computersbut this memory remains vulnerable to the vagaries of the battery.
513
514
Should the battery die or slip from its connection even momentarily, the contents of this
simulated nonvolatile memory may be lost.
Given the choice, youd of course want the memory of your computer to be nonvolatile.
The problem is that nearly all memory systems based solely on electricity and electronic
storage are volatile. Those all-electric memory systems that are nonvolatile are cursed
with a number of drawbacks. Most are substantially slower than volatile memory (rewriting a block of Flash memory can take seconds compared to the microseconds required by
most kinds of volatile memory). The common nonvolatile memory systems also have limited lives. Flash memory typically can be rewritten a few hundred-thousand times. Volatile
memory might get rewritten hundreds of thousands of times in less than a second.
Measurement
The basic form of computer memory is the cell, which holds a single bit of data. The term
bit is a contraction of binary digit. A bit is the smallest possible piece of information. It
doesnt hold much intelligence; it only indicates whether something is or isnt (on or off),
is up or down, is something (one) or nothing (zero). Its like the legal system: Everything
is in black and white, and there are no shades of gray (at least when the gavel comes
down).
When enough bits are taken collectively, they can code meaningful information. A pattern
of bits can encode more complex information. In their most elementary form, for example, five bits could store the number 5. Making the position of each bit in the code significant increases the amount of information a pattern with a given number of bits can
identify. (The increase follows the exponential increase of powers of twofor n bits, 2 to
the nth power of unique patterns can be identified.) By storing many bit-patterns in
duplicative memory units, a storage system can retain any amount of information.
Measuring Units
People dont remember the same way computers do. For us human beings, remembering
a complex symbol can be as easy as storing a single bit. Although two choices may be
enough for a machine, we prefer a multitude of selections. Our selection of symbols is as
broad as our imaginations. Fortunately for typewriter-makers, however, weve reserved
just a few characters as the symbol set for our language26 uppercase letters, a matching
number of lowercase letters, ten numerals, and enough punctuation marks to keep grammar teachers preoccupied for entire careers.
Representing these characters in binary form makes computers wonderfully useful, so
computer engineers tried to develop the most efficient bit-patterns for storing the diversity of symbols we finicky humans prefer. If you add together all those letters, numbers,
and punctuation marks, youll find that the lowest power of two that could code them all
is two to the seventh power, or 128. Computer engineers went one better: By using an
Defining Memory
eight-bit code, yielding a capacity of 256 symbols, they found that all the odd diacritical
marks of foreign languages could be represented by the same code. The usefulness of this
eight-bit code has made eight bits the standard unit of computer storage, a unit called the
byte.
Half a byte (a four-bit storage unit) is called a nibble because, at least in the beginning of
the personal computer revolution, engineers had senses of humor. Four bits can encode 16
symbolsenough for ten numerals and six operators (addition, subtraction, multiplication, division, exponents, and square roots), making the unit useful for numbers-only
devices such as handheld calculators.
The generalized term for a package of bits is the digital word, which can comprise any
number of bits that a computer might use as a group. In the world of Intel microprocessors, however, the term word has developed a more specific meaningtwo bytes of data,
or 16 bits.
In the Intel scheme, a double-word comprises two words, or 32 bits; a quad-word is four
words, eight bytes, or 64 bits.
The most recent Intel microprocessors are designed to handle data in larger gulps. To
improve performance, they feature wider internal buses between their integral caches and
processing circuitry. In the case of the current Intel microprocessors, this bus is 128-bits
wide. Intel calls a single bus-width gulp a line of memory.
Because the designations word and double-word sometimes vary with the register width of
microprocessors, the Institute of Electrical and Electronic Engineers (IEEE) developed a
nomenclature system thats unambiguous for multiple-byte widths: doublet for two bytes,
quadlet for four, and octlet for eight. Table 15.1 summarizes the common names and the
IEEE standard designations for the sizes of primary storage units.
TA B L E 1 5 . 1
Unit
Primary Intel Memory Storage Unit Designations
IEEE Standard Notation
Bit
Bits
Bytes
0.125
Nibble
Nibble
0.5
Byte
Byte
Word
Doublet
16
Double-word
Quadlet
32
Quad-word
Octlet
64
128
16
Line
515
516
The Multimedia Extensions (MMX) used by all the latest computer microprocessors
introduced four additional data types into computer parlance. These repackage groups of
smaller data units into the 64-bit registers used by the new microprocessors. The new
units are all termed packed because they fit (or pack) as many smaller units as possible into
the larger registers. These new units are named after the smaller units comprising them.
For example, when eight bytes are bunched together into one 64-bit block to fit an MMX
microprocessor register, the data is in packed byte form. Table 15.2 lists the names of these
new data types.
TA B L E 1 5 . 2
New 64-Bit MMX Storage Designations
Name
Basic Units
Number of Units
Packed byte
Byte (8 bits)
Packed word
Word (16 bits)
Packed double-word
Double-word (32 bits)
Quad-word
64 bits
Todays applications demand thousands and millions of bytes of memory. The basic measuring units for memory are consequently large multiples of the byte. Although they wear
common Greek prefixes shared by units of the metric system, the computer world has
adopted a slightly different measuring system. Although the Greek prefix kilo means thou10
sand, computer people assign a value of 1024 to it, the closest round number in binary, 2
(two to the tenth power). Larger units increase by a similar factor so that a megabyte is
20
30
actually 2 bytes and a gigabyte is 2 bytes. Table 15.3 summarizes the names and values
of these larger measuring units.
TA B L E 1 5 . 3
Names and Abbreviations of Large Storage Units
Unit
Abbreviation
Size in Units
Size in Bytes
Kilobyte
KB or K
1024 bytes
1024
Megabyte
MB or M
1024 kilobytes
1,048,576
Gigabyte
GB
1024 megabytes
1,073,741,824
Terabyte
TB
1024 gigabytes
1,099,511,627,776
Petabyte
PB
1024 terabytes
1,125,899,906,843,624
Exabyte
EB
1024 petabytes
1,152,921,504,607,870,976
Zettabyte
ZB
1024 exabytes
1,180,591,620,718,458,879,424
Yottabyte
YB
1024 zettabytes
1,208,925,819,615,701,892,530,176
Defining Memory
Access
Memory works like an elaborate set of pigeonholes used by post office workers to sort
local mail. A memory location called an address is assigned to each piece of information to
be stored. Each address corresponds to one pigeonhole, unambiguously identifying the
location of each unit of storage. The address is a label, not the storage location itself
(which is actually one of those tiny electronic capacitors, latches, or fuses).
Direct Access
Because the address is most often in binary code, the number of bits available in the code
determines how many such unambiguous addresses can be directly accessed in a memory
system. As noted before, an eight-bit address code permits 256 distinct memory locations
8
16
(2 = 256). A 16-bit address code can unambiguously define 65,536 locations (2 = 65,536).
The available address codes generally correspond to the number of address lines of the
microprocessor in the computer, although strictly speaking they need not.
The amount of data stored at each memory location depends on the basic storage unit,
which varies with the design of the computer system. Generally, each location contains
the same number of bits that the computer processes at one time. Although todays
Pentium-class microprocessors have 32-bit registers and 64-bit data buses, the smallest
unit of memory they can individually address is actually four double-words (16 bytes).
Smaller memory units cannot be individually retrieved because the four least-significant
address lines are absent from these microprocessors. Because the chips prefer to deal with
data one line at a time, greater precision in addressing is unnecessary.
In writing to memory, where a microprocessor might need to change an individual byte
but can only address a full line, the chip uses a technology termed masking. The mask preserves all the memory locations in the line that are not to be changed. Although they
address the byte by the chunk it lies within, the mask prevents overwriting the bytes of
memory that do not need to change.
Memory chips do not connect directly to the microprocessors address lines. Instead, special circuits that comprise the memory controller translate the binary data sent to the
memory address register into the form necessary to identify the memory location
requested and retrieve the data there. The memory controller can be as simple as addressdecoding logic circuitry or an elaborate application-specific integrated circuit that combines several memory-enhancing functions.
To read memory, the microprocessor activates the address lines corresponding to the
address code of the wanted memory unit during one clock cycle. This action acts as a
request to the memory controller to find the needed data. During the next clock cycle, the
memory controller puts the bits of code contained in the desired storage unit on the
microprocessors data bus. This operation takes two cycles because the memory controller
cant be sure that the address code is valid until the end of a clock cycle. Likewise, the
517
518
microprocessor cannot be sure the data is valid until the end of the next clock cycle.
Consequently, all memory operations take at least two clock cycles.
Writing to memory works similarly: The microprocessor first sends off the address to
write to, the memory controller finds the proper pigeonhole, and then the microprocessor
sends out the data to be written. Again, the minimum time required is two cycles of the
microprocessor clock.
Reading or writing can take substantially longer than two cycles, however, because microprocessor technology has pushed into performance territory far beyond the capabilities of
todays affordable DRAM chips. Slower system memory can make the system microprocessorand the rest of the computerstop while it catches up, extending the memory
read/write time by one or more clock cycles.
Bank-Switched Memory
In some applications, a computer or piece of electronic gear has more memory than
addresses available for storing it. (Such address shortfalls were admittedly more common
in the past than today, when microprocessors can address terabytes of memory.) For
example, the telephone company long agoback when there was but one telephone companyfaced its own addressing shortage when the number of phones in use exceeded the
ten million distinct seven-digit telephone numbers. The telephone companys solution was
to break the nation (and the world) into separate ranges, what we know now as area codes.
Each area code has the full ten million phone numbers available to it, expanding the range
of available telephone numbers by a factor equal to the number of area codes.
When computers were limited to a few dozen kilobytes by the addressing range of their
microprocessors, clever engineers developed their own version of the area code. They
divided memory into banks, each of which individually fit into the address range of the
microprocessor. Using the computer equivalent of a giant channel selector knob, engineers enabled the computer to switch any one of the banks into the addressing range of
the microprocessor while removing the rest from the chips control. The maximum
addressable memory of the system becomes the product of the addressing range of the
chip and the number of banks available. Because the various banks get switched in and out
of the systems limited addressing range, this address-extension technique is usually called
bank-switching.
The bank-switching technique was once used to extend the addressing range of personal
computers, creating what was called expanded memory. Bank-switching is still sometimes
used in video display systems, especially in older display modes.
The memory banks on the motherboards of most modern computers have nothing to do
with bank switching. In this context, a bank of memory is any size block of memory that is
arranged with its bits matching the number of data connections to your microprocessor.
That is, a bank of memory for a Pentium 4 is a block of memory arranged 64 bits wide.
Solid-State Memory
True bank-switching requires a special memory board that incorporates data registers that
serve as the actual bank switches. In addition, programmers must specifically write their
applications to use bank-switched memory. The program (or operating system that it runs
under) must know how to track every switch and turn off memory to be sure that the correct banks are in use at any given moment.
Solid-State Memory
To remember a single bitwhether alone or as part of a nibble, byte, word, or doublewordcomputer memory needs only to preserve a single state (that is, whether something
is true or false, positive or negative, or a binary one or zero). Almost anything can suffice
to remember a single statewhether a marble is in one pile or another, whether a dab of
marzipan is eaten or molding on the shelf, whether an electrical charge is present or
absent. The only need is that the memory unit has two possible states and that it will
maintain itself in one of them once it is put there. Should a memory element change on
its own, randomly, it would be useless because it does not preserve the information that
its supposed to keep.
Although the possibilities of what can be used for remembering a single state are nearly
endless, how the bits are to be used makes some forms of memory more practical than
others. The two states must be both readily changeable and readily recognizable by whatever mechanism is to use them. For example, a string tied around your finger will help
you remember a bit state but would be inconvenient to store information for a machine.
Whatever the machine, it would need a mechanical hand to tie the knot and some means
of detecting its presence on your fingera video camera, precision radar set, or even a gas
chromatography system.
True old-timers who ratcheted themselves down into computers from the mainframe
computers that ran rampant in business and industry in the 1950s, 60s, and 70s sometimes
use the term core when speaking of a computers memory system. The term doesnt derive
from the centrality of memory to the operation of the computer but rather from one of
the first memory technologies used by ancient computersa fabric of wires with a ferrite
doughnut, called a core, woven into (literally) each intersection of the warp and woof.
Although today the term core is but a memory, all current memory technologies share with
it one important characteristic: Electricity is able to alter its contents. After all, todays
computers think with electricity.
In digital computers, it is helpful to store a state electrically so the machine doesnt need
eyes or hands to check for the string, marble, or marzipan. Possible candidates for electrical state-saving systems include those that depend on whether an electrical charge is present or whether a current will flow. Both of these techniques are used in computer
memories for primary storage systems.
519
520
The analog of electricity, magnetism, can also be readily manipulated by electrical circuits
and computers. Core memory, in fact, used the magnetic fields of the ferrite cores to store
bits. Today, however, magnetic storage is mostly reserved for secondary storage because
magnetism is one step removed from electricity. Storage devices have to convert electricity
to magnetism to store bits and convert magnetic fields to electrical pulses to read them.
The conversion process takes time, energy, and effortall of which pay off for long-term
storage, at which magnetism excels, but are unnecessary for the many uses inside the computer.
The vast majority of memory used in computers is based on storing electrical charges
rather than magnetic fields. Because all the other signals inside a computer are normally
electronic, the use of electronic memory is only natural. It can operate at electronic speed
without the need to convert technologies. Chip-makers can fabricate electronic memory
components exactly as they do other circuits, even on the same assembly lines. Best of all,
electronic memory is cheap. In fact, its the most affordable of all direct-access technologies.
Modern memory circuits are made from the same silicon semiconductors as other electronic circuits such as microprocessors. In other words, most primary storage in todays
computers uses solid-state semiconductor technology.
Read-Write Memory
In this post-modern age, we wantwe expecteverything to be recyclable. We want whatever we use to be used again rather than used once and thrown away. Its one of the most
ecologically sound policies (although using less would be even better).
Computers prefer to treat their memory the same way and for a similar reason. They
dont worry about physical waste. They dont worry at all. But they can run out of
resources. If a computer could only use memory once, it would quickly use up all its
addresses. As it has new things to store, it would slide each bit into a memory address
until it ran out of places to put them. At that point, it would no longer be able to do anything new, at least until you added more memory.
In computer terms, people dont talk about using memory, nor do they use the mechanical
terms of recording and playing back. Rather, they use terms of authorship. When a computer remembers something, it writes it to memory. To recall something is to read it from
memory. Reusable memory that can be written to and read from whenever the computer
wants or needs to is called read-write memory.
As straightforward as it sounds, the term read-write memory is almost never used. Instead,
engineers have called the read-write memory of most computers RAM, the familiar random-access term. They use the fast random-access capability of primary storage to distinguish it from slower secondary storage, which confusingly is also mostly random access.
Solid-State Memory
They also use RAM to distinguish read-write memory from read-only memory (which we
will talk about next), even though ROM is also random access. About the best that can be
said is that when someone says RAM, what he usually means is read-write memory.
The electronic circuits that make random-access read-write memory possible take two
forms: dynamic and static.
Dynamic Memory
The most common design that brings memory to life inside computers uses minute electrical charges to remember memory states. This form of memory stores electrical charges
in small capacitors.
The archetypical capacitor comprises two metal plates separated by a small distance thats
filled with an electrical insulator. A positive charge can be applied to one plate and,
because opposite charges attract, it draws a negative charge to the other nearby plate. The
insulator separating the plates prevents the charges from mingling and neutralizing each
other. Its called a capacitor because it has the capacity to store a given amount of electricity, measured as its capacitance.
The capacitor can function as memory because a computer can control whether the
charge is applied to or removed from one of the capacitor plates. The charge on the plates
can thus store a single state and a single bit of digital information.
In a perfect world, the charges on the two plates of a capacitor would forever hold themselves in place. One of the imperfections in the real world results in no insulator being
perfect. Theres always some possibility that a charge will sneak through any material
(although better insulators lower the likelihood, they cannot eliminate it entirely). Think
of a perfect capacitor as being like a glass, holding whatever you put inside it (for example,
water). A real-world capacitor inevitably has a tiny leak through which the water (or electrical charge) drains out. The leaky nature of capacitors themselves is made worse by the
circuitry that charges and discharges the capacitor because it, too, allows some of the
charge to leak off.
This system seems to violate the primary principle of memoryit wont reliably retain
information for very long. Fortunately, this capacitor-based system can remember long
enough to be usefula few or a few dozen millisecondsbefore the disappearing charges
make the memory unreliable. Those few milliseconds are sufficient that practical circuits
can be designed to periodically recharge the capacitor and refresh the memory.
Refreshing memory is akin to pouring extra water into a glass from which it is leaking out.
Of course, you have to be quick to pour the water while theres a little left so you know
which glass needs to be refilled and which is supposed to be empty.
521
522
To ensure the integrity of their memory, computers periodically refresh memory automatically. During the refresh period, the memory is not available for normal operation.
Accessing memory also refreshes the memory cell. Depending on how a chip-maker has
designed its products, accessing a single cell also may refresh the entire row or column
containing the accessed memory cell.
Because of the changing nature of this form of capacitor-based memory and its need to be
actively maintained by refreshing, it is termed dynamic memory. Integrated circuits that
provide this kind of memory are termed dynamic RAM (DRAM) chips.
In personal computer memories, special semiconductor circuits that act like capacitors are
used instead of actual capacitors with metal plates. A large number of these circuits are
combined together to make a dynamic memory-integrated circuit chip. As with true
capacitors, however, dynamic memory of this type must be periodically refreshed.
Static Memory
Whereas dynamic memory tries to trap evanescent electricity and hold it in place, static
memory allows the current flow to continue on its way. It alters the path taken by the
power, using one of two possible courses of travel to mark the state being remembered.
Static memory operates as a switch that potentially allows or halts the flow of electricity.
A simple mechanical switch will, in fact, suffice as a form of static memory. It, alas, has the
handicap that it must be manually toggled from one position to another by a human or
robotic hand.
A switch that can itself be controlled by electricity is called a relay, and this technology
was one of the first used for computer memory. The typical relay circuit provided a latch.
Applying a voltage to the relay energizes it, causing it to snap from not permitting electricity to flow to it. Part of the electrical flow could be used to keep the relay itself energized, which would, in turn, maintain the electrical flow. Like a door latch, this kind of
relay circuit stays locked until some force or signal causes it to change, thus opening the
door or the circuit.
Transistors, which can behave as switches, can also be wired to act as latches. In electronics, a circuit that acts as a latch is sometimes called a flip-flop because its state (which stores
a bit of data) switches like a political candidate who flip-flops between the supporting and
opposing views on sensitive topics. A large number of these transistor flip-flop circuits,
when miniaturized and properly arranged, together make a static memory chip. Static
RAM is often shortened to SRAM by computer professionals. Note that the principal
operational difference between static and dynamic memory is that static RAM does not
need to be periodically refreshed.
Errors
Read-Only Memory
Not all memory must be endowed with the ability to be changed. Just as there are many
memories you would like to retainyour first love, the names of all the constellations in
the zodiac, the answers to the chemistry exama computer is better off when it can
remember some particularly important things without regard to the vagaries of the power
line. Perhaps the most important of these more permanent rememberings is the program
code that tells a microprocessor that its actually part of a computer and how it should
carry out its duties.
You can render a simple memory system, such as a light switch, unchangeable by carefully
applying a hammer. With enough assurance and impact, you could guarantee that the system would never forget. In the world of solid-state, the principle is the same, but the programming instrument is somewhat different. All that you need is switches that dont
switchor, more accurately, that switch once and jam. This permanent kind of memory is
so valuable in computers that a whole family of devices called read-only memory (ROM)
chips has been developed to implement it. These devices are called read-only because the
computer that they are installed in cannot store new code in them. Only what is already
there can be read from the memory.
Just as RAM isnt really RAM, most ROM isnt really ROM. The only true ROM is called
mask ROM because its storage contents get permanently set in place when the chip is fabricated. When the circuits of the chip are laid out using a mask, they already reflect the
information stored in the chip.
Most kinds of ROM chips can have their contents altered. They differ from RAM chips in
that they cannot be written to in the normal operation of the computer. That is, nothing
the computer normally does can change the contents of these almost-ROM chips. Their
contents can be changed only using special equipmentor special programming routines
inside the computer.
Errors
No memory system is perfect. Certainly your computers memory wont mislay its gloves
or forget your birthday, but it can suffer little slips, an errant bit here and there. Although
one bit in a few million might not seem like a big deal, its enough to send your system
into a tailspin, or worse, alter the answer to some important calculation, jogging the decimal point a few places to the left.
Causes
The memory errors that your computer is likely to suffer fall into two broad classes: soft
errors and hard errors. Either can leave you staring at an unflinching screen, sometimes
523
524
but not always emblazoned with a cryptic message that does nothing to help you regain
the hours work irrevocably lost. The difference between them is transience. Soft errors are
little more than disabling glitches that disappear as fast as they come. Hard errors linger
until you take a trip to the repair shop.
Soft Errors
For your computer, a soft memory error is an unexpected and unwanted change.
Something in memory might turn up different from what its supposed to be. One bit in a
memory chip may suddenly, randomly change state. Or a glitch of noise inside your system may get stored as if it were valid data. In either case, one bit becomes something
other than what its supposed to be, thus possibly changing an instruction in a program or
a data value.
With a soft error, the change appears in your data rather than hardware. Replace or
restore the erroneous data or program code, and your system will operate exactly as it
always has. In general, your system needs nothing more than a reboota cold boot being
best to gain the assurance of your computers self-test of its circuits (including memory).
The only damage is the time you waste retracing your steps to get back to the place in
your processing at which the error occurred. Soft errors are the best justification for the
sage advice, Save often.
Most soft errors result from problems either within memory chips themselves or in the
overall circuitry of your computer. The underlying mechanism behind these two types of
soft errors is entirely different.
Chip-Level Errors
The errors inside memory chips are almost always a result of radioactive decay. The problem is not nuclear waste (although nuclear wastes is a problem) but something even more
devious. The culprit is the epoxy of the plastic chip package, which like most materials
may contain a few radioactive atoms. Typically one of these minutely radioactive atoms
will spontaneously decay and shoot out an alpha particle into the chip. (There are a number of radioactive atoms in just about everythingthey dont amount to very much, but
they are there. And by definition, a radioactive particle will spontaneously decay sometime.) An alpha particle is a helium nucleus, two protons and two neutrons, having a small
positive charge and a lot of kinetic energy. If such a charged particle hits a memory cell in
the chip, the charge and energy of the particle can cause the cell to change state, blasting
the memory bit it contains to a new and different value. This miniature atomic blast is not
enough to damage the silicon structure of the chip itself, however.
Whether a given memory cell will suffer this kind of soft error is unpredictable, just as
predicting whether a given radioactive atom will decay is unpredictable. When you deal
with enough atoms, however, this unpredictability becomes a probability, and engineers
Errors
can predict how often one of the memory cells in a chip will suffer such an error. They
just cant predict which one.
In the early days of computers, radioactive decay inside memory chips was the most likely
cause of soft errors in computers. Thanks to improved designs and technology, each generation of memory chip has become more reliable, no matter whether you measure per bit
or per chip. For example, any given bit in a 16KB chip might suffer a decay-caused soft
error every billion or so hours. The likelihood of any given bit in a modern 16MB chip
will suffer an error is on the order of once in two trillion hours. In other words, modern
memory chips are about 5000 times more reliable than those of first generation computers, and the contents of each cell is about five million times more reliable once you take
into account that chip capacities have increased a thousand-fold. Although conditions of
use influence the occurrence of soft errors, the error rate of modern memory is such that
a typical computer with 8MB of RAM would suffer a decay-caused soft error once in 10 to
30 years. The probability is so small that many computer-makers now ignore it.
System-Level Errors
Sometimes the data traveling though your computer gets hit by a noise glitchon the
scale of memory cells, a little glitch can be like you getting struck by lightning. Just as you
might have trouble remembering after such a glitch, so does the memory cell. If a pulse of
noise is strong enough and occurs at an especially inopportune instant, it can be misinterpreted by your computer as a data bit. Such a system-level error will have the same effect
on your computer as a soft error in memory. In fact, some system-level errors may be
reported as memory errors (for example, when the glitch appears in the circuitry between
your computers memory chips and the memory controller).
The most likely place for system-level soft errors to occur is on your computers buses. A
glitch on a data line can cause your computer to try to use or execute a bad bit of data or
program code, thus causing an error. Or your computer could load the bad value into
memory, saving it to relish (and crash from) at some later time. A glitch on the address
bus will make your computer similarly find the wrong bit or byte, and the unexpected
value may have exactly the same effects as a data bus error.
The probability of a system-level error occurring depends on the design of your computer. A careless designer can leave your system not only susceptible to system-level errors
but even prone to generating the glitches that cause them. Pushing a computer design to
run too fast is particularly prone to cause problems. You can do nothing to prevent system-level soft errors other than to choose your computer wisely.
Hard Errors
When some part of a memory chip actually fails, the result is a hard error. For instance, a
jolt of static electricity can wipe out one or more memory cells. As a result, the initial
525
526
symptom is the same as a soft errora memory error may cause an error in the results
you get or a total crash of your system. The operative difference is that the hard error
doesnt go away when you reboot your system. In fact, your machine may not pass its
memory test when you try to start it up again. Alternately, you may encounter repeated,
random errors when a memory cell hovers between life and death.
Hard errors require attention. The chip or module in which the error originates needs to
be replaced.
Note, however, that operating memory beyond its speed capability often causes the same
problem as hard errors. In fact, operating memory beyond its ratings causes hard errors.
You can sometimes clear up such problems by adding wait states to your systems memory
cycles, a setting many computers allow you to control as part of their advanced setup procedure. This will, of course, slow down the operation of your computer so that it can
accommodate the failing memory. The better cure is to replace the too-slow memory with
some that can handle the speed.
Detection and Prevention

Most computers check every bit of their memory for hard errors every time you switch
your system on or perform a cold boot, although some computers give you the option of
bypassing this initial memory check to save time. Soft errors are another matter entirely.
They rarely show up at boot time. Rather, they are likely to occur at the worst possible
momentwhich means just about any time youre running your computer.
Computer-makers use two strategies to combat memory errors: parity and detection/correction. Either one will ensure the integrity of your systems memory. Which is bestor
whether you need any error compensation at allis a personal choice. Manufacturers
consider the memory systems in most modern computers so reliable that they opt to save
a few bucks and omit any kind of error detection. Where accuracy really counts, in
servers, network managers usually insist their computers have the best error prevention
possible by using memory that supports error-correction code.
Parity
When memory chips were of dubious reliability, computer manufacturers followed the
lead of the first computer and added an extra bit of storage to every byte of memory. The
extra bit was called a parity check bit, and it allows the computer to verify the integrity of
the data stored in memory. Using a simple algorithm, the parity check bit permits a computer to determine that a given byte of memory has the right number of binary zeros and
ones in it. If the count changes, your computer knows an error occurred.
When a microprocessor writes a byte to memory, the value stored in the parity check bit
is set either to a logical one or zero in such a way that the total of all nine bits storing the
Errors
byte is always odd. Every time your computer reads a given portion of memory, the memory controller totals up the nine bits storing each byte, verifying that the overall total
(including the parity check bit) remains odd. Should the system detect a total thats even,
it immediately knows that something has happened to cause one bit of the byte to change,
making the stored data invalid.
The philosophy behind parity memory is that having bad data is worse than losing information through a system crash. One bad bit results in a halt to all computing. You wont
make a mistakeand you wont like it when your computers memory does. Although parity errors happened rarely, when they did, most people questioned the value of parity
checking. Consequently, few people minded when manufacturers started omitting it from
their designs.
Fake Parity
Fake parity memory is a means of cutting the cost of memory modules for computers with
built-in parity checking. Instead of actually performing a parity check of the memory on
the module, the fake parity system always sends out a signal indicating that memory parity
is good. No extra bits of memory means the vendors of fake parity chips dont have to
pay the extra bucks for extra memory.
Fake parity has two downsides. The cost savings often are not passed down to you, at least
explicitly. The fake parity modules are often sold as ordinary parity modules with no indication of the underlying shortchange on technology. Moreover, a fake parity module does
not offer protection against parity errors causing erroneous data in your computer. One
fake parity module defeats the operation and purpose of the parity memory system in your
computer. Fortunately, with the decline in the use of actual parity memory, the incentive
to sell fake parity memory has disappeared.
Detection/Correction
Parity checking can only locate an error of one bit in a byte. More elaborate error-detection schemes can detect larger errors. Better still, when properly implemented, these
schemes can fix single-bit errors without crashing your system. Called error-correction code
(ECC), this scheme in its most efficient form requires three extra bits per byte of storage.
The additional bits allow your system not only to determine the occurrence of a memory
error but also to locate any single bit that changed so that the error can be reversed. Some
people call this technology Error Detection and Correction (EDAC).
Most server computers use ECC memory. In fact, many desktop machines also allow you
to use ECC. Support for ECC is built into many chipsets. If your BIOS allows you to
turn this facility on, you can use ECC, but youll also need ECC memory modules (which
means tossing out the ones that you have).
527
528
Server-makers have another reason for shifting from parity to ECC memory. As the width
of the data bus increases, error correction memory become less expensive to implement.
In fact, with todays Pentium computerswhich have 64-bit data busesthe cost of extra
memory for parity checking and full error correction are the same. Table 15.4 summarizes
the penalty required by parity and ECC technology for various bus widths.
TA B L E 1 5 . 4
Comparison of Parity and ECC Memory
Extra Bits Required
Cost Increase
Bus Width
Parity
ECC
Parity
ECC
12.5%
62%
16
12.5%
38%
32
12.5%
25%
64
12.5%
12.5%
Parity and ECC memory are technologies for people who want or need to take every possible precaution. They are the kind of people who check the weather report, then put on
their galoshes and carry an umbrella even when the prediction is 100 percent in favor of
sun. They are the folks who get the last laugh when they are warm and dry after a oncein-a-century freak thunderstorm rumbles through. These people know that the best solution to computer and memory problems is prevention.
Mass Storage Systems

Secondary storage is essentially an electronic closet, a place where you put information
that you dont want to constantly hold in your hands but that you dont want to throw
away, either. As with the straw hats, squash rackets, wallpaper tailings, and all the rest of
your dimly remembered possessions that pile up out of sight behind the closet door,
retrieving a particular item from mass storage can take longer than when you have what
you want at hand.
Access
Mass storage takes two forms: online storage, which is instantly accessible by your microprocessors commands, and offline storage, which requires some extra, possibly human,
intervention (such as you sliding a cartridge into a drive) for your system to get the bytes
that it needs. Sometimes, the term near-line storage is used to refer to systems in which
information isnt instantly available but can be put into instant reach by a microprocessor
command. The jukeboxan automatic mechanism that selects CD-ROM cartridges, and
sometimes tape cartridgesis the most common example.
Moving bytes from mass storage to memory determines how quickly stored information
can be accessed. In practical online systems, the time required for this access ranges from
less than 0.01 seconds in the fastest hard disks to 1000 seconds in some tape systems,
spanning a range of 100,000 or five orders of magnitude.
By definition, the best offline storage systems have substantially longer access times than
the quickest online systems. Even with fast-access disk cartridges, the minimum access
time for offline data is measured in seconds because of the need to find and load a particular cartridge. The slowest online and the fastest offline storage system speeds, however,
may overlap because the time to ready an offline cartridge can be substantially shorter
than the period required to locate needed information written on a long online tape.
Data Organization
Computer mass storage systems differ in the way they organize and allow access to the
information that they store. Engineers class the storage systems in modern computers as
either sequential or random access. From a practical standpoint, the difference is functionalhow the computer system finds the information it needs. But it also has a historic
dimensionsequential computer mass storage predates random access. You can also look
at the difference topographically. Sequential storage is one dimensional, whereas random
access has two (or possibly more) dimensions.
Neither technology is inherently better. Both are used in modern computers. For a given
application, of course, one of the two is typically more suitable. But because each has its
own strengths, both will likely persist as long as computers populate desktops, if not
longer.
Sequential Media
A fundamental characteristic of tape recording is that information is stored on tape onedimensionallyin a straight line across the length of the tape. This form of storage is
called sequential because all the bits of data are organized one after another in a strict
sequence, like those paper-based dots. In digital systems, one bit follows after the other
for the full length of the tape. Although in multiple-track and helical recording systems,
bits may also spread across the width of the tape, conceptually these, too, store information in one dimension only.
CDs and DVDs designed for audio and video applications (that is, CD-DA, CD-Video,
DVD-Video, and DVD-Audio) are designed to work best as one-dimensional media as
well. They record data as a continuous stream meant to be displayed (or listened to) in the
same sequential order it was recorded. However, the two-dimensional nature of the disc (a
disc is a two-dimensional plane, after all) allows these media to transcend some of the limitations of sequential recording.
529
530
In the Newtonian universe (the only one that appears to make sense to the normal human
mind), the shortest distance between two points is always a straight line. Alas, in magnetic
tape systems, the shortest distance between two bits of data on a tape may also be a long
time. To read two widely separated bits on a tape, all the tape between them must be
passed over. Although all the bits in between are not to be used, they must be scanned in
the journey from the first to second bits. If you want to retrieve information not stored in
order on a tape, the tape must shuttle back and forth to find the data in the order that you
want it. All that tape movement to find data means wasted time.
In theory, theres nothing wrong with sequential storage schemesdepending on the storage medium thats used, they can be very fast. For example, one form of solid-state computer memory, the all-electronic shift register, moves data sequentially at nearly the speed
of light.
The sequential mass storage systems of todays computers are not so blessed with speed,
however. Because of their mechanical foundations, most tape systems operate somewhat
slower than the speed of light. For example, although light can zip across the vacuum of
the universe at 186,000 miles per second (or so), cassette tape crawls along at one and
seven-eighths inches per second. Although light can get from here to the moon and back
in a few seconds, moving a cassette tape that distance would take about ten billion times
longer, several thousand years.
Although no tape stretches as long as the 238,000 mile distance to the moon, sequential
data access can be irritatingly slow. Instead of delivering the near-instant response most of
todays impatient power users demand, picking a file from a tape can take as long as ten
minutes. Even the best of todays tape systems require 30 seconds or more to find a file. If
you had to load all your programs and data files from tape, you might as well take up crocheting to tide you through the times youre forced to wait.
Most sequential systems store data in blocks, sometimes called exactly that, and sometimes
called records. The storage system defines the structure and contents of each block.
Typically, each block includes identifying information (such as a block number) and errorcontrol information in addition to the actual data. Blocks are stored in order on tape. In
some systems, they lie end to end while others separate them with blank areas called interrecord gaps.
Most tape systems use multiple tracks to increase their storage (some systems spread as
many as 144 tracks across tape just one-quarter inch wide). The otherwise stationary readwrite head in the tape machine moves up and down to select the correct track. This multitrack design also improves the access speed of the medium. With multiple tracks, the drive
doesnt have to scan the whole tape to get at a particular block. With 144 tracks, a drive
needs only to scan 1/144th the length of the tape to find a particular block. Sequential
CDs and DVDs provide a greater speed-up because they can select among thousands of
radial positions along the disc to read.
To take advantage of the speed-up afforded by multiple tracks, the drive needs to know on
which track to find the block youre seeking. Old tape systems didnt keep track of location information, so they would scan back and forth along the tape, one track after
another, to find the blocks you wanted. New tape standards put a directory on the tape
that holds the location of information on the tape. By consulting the directory, the drive
can determine which track holds the information you want and zero in on the correct
track to trim the response time of the tape system.
Random-Access Media
On floppy and most hard computer disks, as well as some CD and DVD formats, the
recorded data is organized to take advantage of the two-dimensional aspect of the flat,
wide disk surface to give even faster access than is possible with the directory system on
tape. Instead of being arranged in a single straight line, disk-based data is spread across
several concentric circles like lanes in a circular racetrack or the pattern of waves rolling
away from a splash. Some optical drives follow this system, but many other optical systems
modify this arrangement, changing the concentric circles into one tightly packed spiral
that continuously winds from the edge to the center of the disk. But even these continuous-data systems behave much as if they had concentric circles of information.
The mechanism for making this arrangement is quite elementary. The disk moves in one
dimension under the read/write head, which scans the tape in a circle as it spins and
defines a track, which runs across the surface of the disk much like one of the lanes of a
racetrack. In most disk systems, the head can move as well; otherwise, the read/write head
would be stuck forever hovering over the same track and the same stored data, making it a
sequential storage system that wastes most of the usable storage surface of the disk.
In most of todays disks systems, the read/write moves across a radius of the disk, perpendicular to a tangent of the tracks. The read/write head can quickly move between the different tracks on the disk. Although the shortest distance between two points (or two bytes)
remains a straight line, to get from one byte to another, the read/write head can take
shortcuts across the lanes of the racetrack. After the head reaches the correct track, it still
must wait for the desired bit of information to cycle around under it. However, disks spin
relatively quickly300 revolutions per minute for most floppy disks and up to 7200 rpm
for some hard disksso you only need to wait a fraction of a second for the right byte to
reach your system.
Because the head can jump from byte to byte at widely separated locations on the disk
surface and because data can be read and retrieved in any order or at random in the twodimensional disk system, disk storage systems are often called random-access devices, even
531
532
though they fall a bit short of the mark with their need to wait while hovering over a
track.
The random-access capability of magnetic disk systems makes the combination much
faster than sequential tape media for the mass storage of data. Disks are so superior and so
much more convenient than tapes that tape is almost never used as a primary mass storage
system. Usually, tape plays only a secondary role as a backup system. Disks are used to
store programs and files that need to be loaded on a moments notice.
Magnetic Storage
The most popular mass storage systems in modern computers are based on magnetism.
Hard disks use magnetic storage, as do floppy disks and tape drives. Magnetic storage is
one of the oldest storage mediums used by computers (only punch cards predate its application), and it ranks as one of the most reliable. It also yields systems with the lowest cost
per byte of storage.
Magnetic systems are popular both because of the unique properties of magnetism and
because of its heritage. It has been used so long that its operating principles are well
understood. It gives engineers a head start when they look for ways to improve the technologyand improve it they have. In 20 years, the typical capacity of a hard disk system
has grown by a factor of 10,000, while the disk mechanism itself has actually shrunk.
Every time some new breakthrough technology threatens to beat the storage capabilities
of magnetic media, engineers simply roll out a refinement or two that lets the old medium
easily eclipse the new.
The original electronic mass storage system was magnetic tapethat thin strip of paper
(in the United States) upon which a thin layer of refined rust had been glued. Later, the
paper gave way to plastic, and the iron oxide coating gave way to a number of improved
magnetic particle based on iron, chrome dioxide, and various mixtures of similar compounds.
The machine that recorded upon these rust-covered ribbons was the Magnetophon, the
first practical tape recorder, created by the German division of the General Electric
Company, Allgemeine Elektricitaets Gesellschaft (AEG) in 1934. Continually improved
but essentially secret through the years of World War II, despite its use at German radio
stations, the Magnetophon was the first device to record and play back sound indistinguishable from live performances. After its introduction to the United States (in a demonstration by John T. Mullin to the Institute of Radio Engineers in San Francisco on May
16, 1946), tape recording quickly became the premiere recording medium and within a
decade gained the ability to record video and digital data. Today, both data cassettes and
streaming tape systems are based on the direct offspring of the first Magnetophon.
Magnetic Storage
The principle is simple. Some materials become magnetized under the influence of a magnetic field. Once the material becomes magnetized, it retains its magnetic field. The magnetic field turns a suitable mixture or compound based on one of the magnetic materials
into a permanent magnet with its own magnetic field. A galvanometer or similar device
can later detect the resulting magnetic field and determine that the material has been
magnetized. The magnet material remembers.
Magnetism
Key to the memory of magnetism is permanence. Magnetic fields have the wonderful property of being static and semi-permanent. On their own, they dont move or change. The
electricity used by electronic circuits is just the opposite. It is constantly on the go and
seeks to dissipate itself as quickly as possible. The difference is fundamental. Magnetic
fields are set up by the spins of atoms physically locked in place. Electric charges are carried by mobile particlesmostly electronsthat not only refuse to stay in place but also
are individually resistant to predictions of where they are or where they are going.
Given the right force in the right amount, however, magnetic spins can be upset, twisted
from one orientation to another. Because magnetic fields are amenable to change rather
than being entirely permanent, magnetism is useful for data storage. After all, if a magnetic field were permanent and unchangeable, it would present no means of recording
information. If it couldnt be changed, nothing about it could be altered to reflect the
addition of information.
At the elemental particle level, magnetic spins are eternal, but taken collectively, they can
be made to come and go. A single spin can be oriented in only one direction, but in virtually any direction. If two adjacent particles spin in opposite directions, they cancel one
another out when viewed from a larger, macroscopic perspective.
Altering those spin orientations takes a force of some kind, and thats the key to making
magnetic storage work. That force can make an alteration to a magnetic field, and after
the field has changed, it will keep its new state until some other force acts upon it.
The force that most readily changes one magnetic field is another magnetic field. (Yes,
some permanent magnets can be demagnetized just by heating them sufficiently, but the
demagnetization is actually an effect of the interaction of the many minute magnetic fields
of the magnetic material.)
Despite their different behavior in electronics and storage systems, magnetism and electricity are manifestations of the same underlying elemental force. Both are electromagnetic phenomena. One result of that commonalty makes magnetic storage particularly
desirable to electronics designersmagnetic fields can be created by the flow of electrical
energy. Consequently, evanescent electricity can be used to create and alter semi-permanent magnetic fields.
533
534
When set up, magnetic fields are essentially self-sustaining. They require no energy to
maintain, because they are fundamentally a characteristic displayed by the minute particles
that make up the entire universe (at least according to current physical theories). On the
submicroscopic scale of elemental particles, the spins that form magnetic fields are, for the
most part, unchangeable and unchanging. Nothing is normally subtracted from them
they dont give up energy, even when they are put to work. They can affect other electromagnetic phenomena (for example, they can be used in mass to divert the flow of
electricity). In such a case, however, all the energy in the system comes from the electrical
flowthe magnetism is a gate, but the cattle that escape from the corral are solely electrons.
The magnetic fields that are useful in storage systems are those large enough to measure
and effect changes on things that we can see. This magnetism is the macroscopic result of
the sum of many microscopic magnetic fields, many elemental spins. Magnetism is a characteristic of submicroscopic particles. (Strictly speaking, in modern science magnetism is
made from particles itself, but we dont have to be quite so particular for the purpose of
understanding magnetic computer storage.)
Magnetic Materials
Three chemical elements are magneticiron, nickel, and cobalt. The macroscopic
strength as well as other properties of these magnetic materials can be improved by alloying them, together and with nonmagnetic materials, particularly rare earths such as samarium.
Many particles at the molecular level have their own intrinsic magnetic fields. At the
observable (macroscopic) level, they do not behave like magnets because their constituent
particles are organizedor disorganizedrandomly so that in bulk, the cumulative effects
of all their magnetic fields tend to cancel out. In contrast, the majority of the minute magnetic particles of a permanent magnet are oriented in the same direction. The majority
prevails, and the material has a net magnetic field.
Some materials can be magnetized. That is, their constituent microscopic magnetic fields
can be realigned so that they reveal a net macroscopic magnetic field. For instance, by
subjecting a piece of soft iron to a strong magnetic field, the iron will become magnetized.
Magnetic Storage
If that strong magnetic field is produced by an electromagnet, all the constituents of a
magnetic storage system become available. Electrical energy can be use to alter a magnetic
field, which can be later detected. Put a lump of soft iron within the confines of an electromagnet that has not been energized. Any time you return, you can determine whether
Magnetic Storage
the electromagnet has been energized in your absence by checking for the presence of a
magnetic field in the iron. In effect, you have stored exactly one bit of information.
To store more, you need to be able to organize the information. You need to know the
order of the bits. In magnetic storage systems, information is arranged physically by the
way data travels serially in time. Instead of being electronic blips that flicker on and off as
the milliseconds tick off, magnetic pulses are stored like a row of dots on a piece of
papera long chain with a beginning and end. This physical arrangement can be directly
translated to the temporal arrangement of data used in a serial transmission system, just
by scanning the dots across the paper. The first dot becomes the first pulse in the serial
stream, and each subsequent dot follows neatly in the data stream as the paper is scanned.
Instead of paper, magnetic storage systems use one or another form of mediagenerally a
disk or long ribbon of plastic tapecovered with a magnetically reactive mixture. The
form of medium directly influences the speed at which information can be retrieved from
the system.
In operation, the tape moves from left to right past a stationary read/write head. When a
current is passed through an electromagnetic coil in this head, it creates the magnetic field
needed to write data onto the tape. When the tape is later passed in front of this head, the
moving magnetic field generated by the magnetized particles on the tape induces a minuscule current in the head. This current is then amplified and converted into digital data.
The write current used in putting data on the tape overpowers whatever fields already
exist on the tape, both erasing them and imposing a new magnetic orientation to the particles representing the information to be recorded.
No matter whether its tape or disk, when a magnetic storage medium is blank from the
factory, it contains no information. The various magnetic domains on it are randomly oriented. Recording on the medium reorients the magnetic domains into a pattern that represents the stored information, as shown in Figure 15.1.
After you record on a magnetic medium, you can erase it by overwriting it with a strong
magnetic field. In practice, you cannot reproduce the true random orientation of magnetic
domains of the unused medium. However, by recording a patter with a frequency out of
the range of the reading or playback systema very high or low frequencyyou can
obscure previously recorded data and make the medium act as if it were blank.
Digital Magnetic Systems

Computer mass storage systems differ in principle and operation from tape systems used
for audio and video recording. Whereas audio and video cassettes record analog signals on
tape, computers use digital signals.
535
536
FIGURE 15.1
Orientation of magnetic domains in
blank and recorded
media.
In the next few years, this situation will likely change as digital audio and video tape
recorders become increasingly available. Eventually, the analog audio and video tape will
become historical footnotes, much as the analog vinyl phonograph record was replaced by
the all-digital compact disc.
In analog systems, the strength of the magnetic field written on a tape varies in correspondence with the signal being recorded. The intensity of the recorded field can span a range
of more than six orders of magnitude. Digital systems generally use a code that relies on
patterns of pulses, and all the pulses have exactly the same intensity.
The technological shift from analog to digital is rooted in some of the characteristics of
digital storage that make it the top choice where accuracy is concerned. Digital storage
resists the intrusion of noise that inevitably pollutes and degrades analog storage. Every
time a copy is made of an analog recording, the noise that accompanies the desired signal
essentially doubles because the background noise of the original source is added to the
background noise of the new recording medium; however, the desired signal does not
change. This addition of noise is necessary to preserve the nuances of the analog recordingevery twitch in the analog signal adds information to the whole. The analog system
cannot distinguish between noise and nuance.
In digital recording, however, theres a sharp line between noise and signal. Noise below
the digital threshold can be ignored without losing the nuances of the signal.
Consequently, a digital recording system can eliminate the noise built up in making
copies. Moreover, noise can creep into analog recordings as the storage medium deteriorates, whereas the digital system can ignore most of the noise added by age. In fact, properly designed digital systems can even correct minor errors that get added to their signals.
Saturation
Digital recordings avoid noise because they ignore all strength variations of the magnetic
field except the most dramatic. They just look for the unambiguous its either there or
not style of digital pulses of information. Analog systems achieve their varying strengths
Magnetic Storage
of field by aligning the tiny molecular magnets in the medium. A stronger electromagnetic field causes a greater percentage of the fields of these molecules to line up with the
field, almost in direct proportion to the field strength, to produce an analog recording.
Because digital systems need not worry about intermediate levels of signal, they can lay
down the strongest possible field that the tape can hold. This level of signal is called saturation because much as a saturated sponge can suck up no more water, the particles on the
tape cannot produce a stronger magnetic field.
Although going from no magnetic field to a saturated field would seem to be the widest
discrepancy possible in magnetic recordingand therefore the least ambiguous and most
suitable for digital informationthis contrast is not the greatest possible or easy to
achieve. Magnetic systems attempt to store information as densely as possible, trying to
cram the information in so that every magnetic particle holds one data bit. Magnetic particles are extremely difficult to demagnetize, but the polarity of their magnetic orientation
is relatively easy to change. Digital magnetic systems exploit this capability to change
polarity and record data as shifts between the orientations of the magnetic fields of the
particles on the tape. The difference between the tape being saturated with a field in one
direction and the tape being saturated with a field in the opposite direction is the greatest
contrast possible in a magnetic system and is exploited by nearly all of todays digital magnetic storage systems.
Coercivity
One word that you may encounter in the description of a magnetic medium is coercivity, a
term that describes how strongly a magnetic field resists change, which translates into
how strong of a magnetic field a particular medium can store. Stronger stored fields are
better because the more intense field stands out better against the random background
noise that is present in any storage medium. Because a higher coercivity medium resists
change better than a low coercivity material, it also is less likely to change or degrade
because of the effects of external influences. Of course, a higher coercivity and its greater
resistance to change means that a recording system requires a more powerful magnetic
field to maximally magnetize the medium. Equipment must be particularly designed to
take advantage of high-coercivity materials.
With hard disks, which characteristically mate the medium with the mechanisms for life,
matching the coercivity of a medium with the recording equipment is permanently handled by the manufacturer. The two are matched permanently when a drive is made.
Removable media devicesfloppy disks, tape cartridges, cassettes, and so onpose more
of a problem. If media are interchangeable and have different coercivities, you face the
possibility of using the wrong media in a particular drive. Such problems often occur with
floppy disks, particularly when you want to skimp and use cheaper double-density media
in high-density or extra-density drives.
537
538
Moreover, the need for matching drive and medium makes upgrading a less-than-simple
matter. Obtaining optimum performance requires that changes in media be matched by
hardware upgrades. Even when better media are developed, they may not deliver better
results with existing equipment.
The unit of measurement for coercivity is the Oersted. As storage media have been miniaturized, the coercivity of the magnetic materials as measured in Oersteds has generally
increased. The greater intrinsic field strength makes up for the smaller area upon which
data is recorded. With higher coercivities, more information can be squeezed into the
tighter confines of the newer storage formats. For example, old 5.25-inch floppy disks had
a coercivity of 300 Oersteds. Todays high-density 3.5-inch floppies have coercivities of
750 Oersteds. Similarly, the coercivities of the tapes used in todays high-capacity quarterinch cartridges are greater than those of the last generation. Older standards used 550
Oersted media; data cartridges with capacities in excess of 1.5GB and minicartridges with
capacities beyond 128MB require 900-Oersted tape. Although invisible to you, the coercivities of tiny modern hard disk drives are much higher than big old drives.
Coercivity is a temperature-dependent property. As the temperature of a medium
increases, its resistance to magnetic change declines. Thats one reason you can demagnetize an otherwise permanent magnet by heating it red hot. Magnetic media dramatically
shift from being unchangeable to changeablemeaning a drop in coercivityat a material-dependent temperature called the Curie temperature. Magneto-optical recording systems take advantage of this coercivity shift by using a laser beam to heat a small area of
magnetic medium that is under the influence of a magnetic field otherwise not strong
enough to affect the medium.
At room temperature, the media used by magneto-optical systems have coercivities on the
order of 6000 Oersteds; when heated by a laser, that coercivity falls to a few hundred
Oersteds. Because of this dramatic change in coercivity, the magnetic field applied to the
magneto-optical medium changes only the area heated by the laser above its Curie temperature (rather than the whole area under the magnetic influence). Because a laser can be
tightly focused to a much smaller spot than is possible with traditional disk read/write
heads, using such a laser-boosted system allows data to be defined by tinier areas of a
recording medium. A disk of a given size thus can store more data when its magnetic storage is optically assisted. Such media are resistant to the effects of stray magnetic fields
(which may change low coercivity fields) as long as they are kept at room temperature.
Retentivity
Another term that appears in the descriptions of magnetic media is retentivity, which measures how well a particular medium retains or remembers the field that it is subjected to.
Although magnetic media are sometimes depended upon to last foreverthink of the
Optical Storage
master tapes of phonograph recordsthe stored magnetic fields begin to degrade as soon
as they have been recorded. A higher retentivity ensures a longer life for the signals
recorded on the medium.
No practical magnetic material has perfect retentivity, however; the random element of
modern physical theories ensures that. Even the best hard disks slowly deteriorate with
age, showing an increasing number of errors as time passes after data has been written. To
avoid such deterioration of so-called permanent records, many computer professionals
believe that magnetically stored recordings should be periodically refreshed. For example,
they exercise tapes stored in mainframe computer libraries periodically (in intervals from
several months to several years, depending on the personal philosophy and paranoia of the
person managing the storage). Although noticeable degradation may require several years
(perhaps a decade or more), these tape caretakers do not want to stake their dataand
their jobson media written long ago.
Disk-makers view such precautions as verging on paranoia. The magnetic medium used in
modern disk drives is nothing like that in the old, self-erasing tapes. Their plated media
(see Chapter 17, Magnetic Storage) have much higher retentivities. Moreover, new
technologies such as S.M.A.R.T. let drives detect any degradation in the magnetic medium
and warn you before you stand a chance of losing your data.
Old floppy disks and backup tapes are another matter. They are likely to deteriorate with
time, enough so that the data on decade-old disks and tapes may be at risk. If you have old
records on old media, for safetys sake you should consider copying the records to fresher
media or a newer technology, such as an optical disc.
Optical Storage
Today the chief alternative to magnetic media in mass storage system is optical technology. Cheap and easy duplication has made computer optical disc storage an ideal distribution medium. Its versatility (the same basic technology stores data, sound, and video) came
about because all engineers eagerly search for big, cheap storage. Both the Compact Disc
(CD) and Digital Versatile Disc (DVD) have both of those virtues by design. In addition,
because the discs both systems use for storage are removable and interchangeable, CD and
DVD systems can serve as both online and offline storage.
Engineers made optical media into true mass storage systems by adding writing capabilities to both the CD and DVD designs. By altering the materials used in the discs, they
were able to allow the same equipment that read data to write it as well.
One of the great virtues of any kind of optical storage is data densitylittle discs mean a
lot of megabytes. By using light beams provided by lasers, optical storage earns several
advantages. Lenses can focus a beam of lightparticularly the coherent beam of a laser
539
540
to a tiny spot smaller than the most diminutive magnetic domain writable on a hard disc
drive. Unlike the restricted magnetic fields of hard discs that have to be used within a
range of a few millionth of an inch, light travels distance with ease. Leaping along some
5.9 trillion miles in a year, some beams have been traveling since almost the beginning of
the universe, 10 to 15 billion years ago. A small gap between the source and storage
medium is consequently no problem. The equipment that generates the beam of light that
writes or reads optical storage need not be anywhere near the medium itself, which gives
equipment designers more freedom than they possibly deserve.
The basic idea behind optical disc storage is that you can encode binary data as a pattern
of black and white splotches just as on and off electrical signals can. You can make your
mark in a variety of ways. The old reliable method is plain, ordinary ink on paper. The
bar codes found universally on supermarket products do exactly that.
Reading the patterns of light and dark takes only a photo-detector, an electrical component
that reacts to different brightness levels by changing its resistance. Light simply allows
electricity to flow through the photo-detector more easily. Aim the photo-detector at the
bar code, and it can judge the difference in reflected light between the bars and background as you move it (or move the product along in front of it). The lasers that read barcodes in the supermarket checkout line quicken the scan. The photo-detector watches the
reflections of the red laser beam and patiently waits until a recognizable patternthe bar
code as the laser scans across itemerges from the noise.
You could store the data of a computer file in one gigantic bar code and bring back paper
tape as a storage mediumone long strip of bars that stretches past the horizon. Even if
you were willing to risk your important data to a medium that turns yellow and flakes
apart under the unblinking eye of the sun like a beach bum with a bad complexion, youd
still have all the joy of dealing with a sequential storage medium. That means renew your
subscription to your favorite magazines because youll have a lot of waiting to do.
The disc, with its random-access capabilities, is better suited as a storage system. Its twodimensional storage makes randomly accessing a particular block of data a matter of milliseconds rather than miles. The choice of a fast-spinning disc was obvious even to the
audio-oriented engineers who put the first compact disc systems together. They had a successful pattern to follow: the old black vinyl phonograph record. The ability to drop a
needle on any track of a record had become ingrained in the hearts and minds of music
lovers for over 100 years. Any new music storage system needed equally fast and easy
access to any selection. The same fast and easy access suits computer storage equally well.
Optical storage brings another benefit. The optical patterns that encode data in both the
CD and DVD systems can be formed mechanically. Both systems use physical pits that
show up as small black holes against a shiny silver background as the dark and bright spots
Virtual Memory
for optically coding data. The pits are mechanical features that can be duplicated mechanically. The process is inexpensive and fast. Machines can flawlessly mold multiple copies in
a few seconds, and with multiple master molds, a factory can crank out a million copies in
a few days. In other words, optical technology allows for duplicating megabytes, even
gigabytes, en masse. Magnetic media, on the other hand, can duplicate data only byte by
byte.
Virtual Memory
Even before anyone conceived the idea of the first personal computer, mainframe computer designers faced the same tradeoff between memory and mass storage. Mass storage
was plentiful and cheap; memory was expensive, so much so that not even large corporations could afford as much as they wanted. In the early years of computers, engineers tried
to sidestep the high cost of memory by faking it, making computers think they had more
memory than they actually did. With some fancy footwork and a lot of shuffling around of
bytes, they substituted mass storage for memory. The engineers called the memory the
computer thought it had but in reality didnt exist virtual memory. Although the cost of
memory has fallen by a millionfold, memory use has risen even faster. Programs need for
memory still outpaces what any reasonable person or organization can afford (or wants to
pay for), so virtual memory not only exists but is flourishing.
Microprocessors cannot ordinarily use disk storage to hold the data they work on. Even if
they could, it would severely degrade the performance of the system because the access
time for disk storage is thousands of times longer than for solid-state memory. To prevent
performance problems and keep active programs in solid-state memory where your computers microprocessor can use them, virtual memory systems swap blocks of code and
data between solid-state and disk storage.
Demand Paging
Most modern computers take advantage of a feature called demand paging that has been
part of all Intel microprocessors since the 386. These chips are able to track memory content as it is moved between disk and solid state memory in 4K blocks. The microprocessor
assigns an address to the data in the block, and the address stays constant regardless of
where the data actually gets stored.
Once solid-state memory reaches its capacity, the virtual memory system copies the contents of one or more pages of memory to disk as the memory space is required. When the
system needs data that has been copied to disk, it copies the least recently used pages to
disk and refills their space with the disk-based pages it needs. When your system attempts
to read from a page with an address thats not available in solid-state memory, it creates a
541
542
page fault. The fault causes a virtual memory manager routine to handle the details of
exchanging data between solid-state and disk storage.
This reactive kind of control is called demand paging because it swaps data only when the
microprocessor demands unavailable addresses. It makes no attempt to anticipate the
needs of the microprocessor.
Virtual memory technology allows your computer to run more programs than would be
otherwise possible given the amount of solid-state memory in your system. The effective
memory of your system approaches the spare capacity of your disk. The downside is that
it takes away from the capacity of your hard disk (although disk storage is substantially less
expensive than solid-state memory by about two orders of magnitude). In addition, performance slows substantially when your system must swap memory.
Virtual memory is an old technology, harking back to the days of mainframe computers
when, as now, disk storage was cheaper than physical memory. Many DOS applications
took advantage of the technology, and it has been part of every version of Windows since
Windows 386which means not only before Windows 95 but also Windows 3.1.
Windows uses a demand-paging system thats based on a least-recently used algorithm.
That is, the Windows virtual memory manager (VMM) decides which data to swap to disk
based on when it was last used by your system. The Windows VMM also maintains the
virtual memory page table, which serves as a key to which pages are used by each application, which are kept in solid-state storage, and which are on disk.
Windows decides which pages to swap to disk using two flags for each page. The accessed
flag indicates that the page has been read from or written to since the time it was loaded
into memory. The dirty flag indicates the page has been written to.
When Windows needs more pages in solid-state memory, it scans through its page table
looking for pages showing neither the accessed flag nor the dirty flag. As it makes its scan,
it resets the accessed flag but not the dirty flag. If it does not find sufficient unflagged
pages, it scans through the page table again. This time, more pages should be unflagged
because of the previous resetting of the accessed flags. If it still cannot find enough available pages, the Windows virtual memory manager then swaps pages regardless of the
flags.
Swap Files
The disk space used by a virtual memory system usually takes the form of an ordinary file,
though one reserved for its special purpose. Engineers call the virtual memory file a swap
file because the memory system swaps data to and from the file as the operating system
requires it. Although earlier versions of Windows had several options you could choose
Data Compression
for swap filesprimarily the choice between a temporary file that swelled as you used it
and a permanent file that forever stole disk spaceWindows 95 erased such complications.
Under Windows 95 as well as Windows 98, the swap file mixes together features of temporary and permanent swap files. Like a temporary swap file, the current Windows swap
file is dynamic, expanding as your system demands virtual memory and contracting when
it does not. In addition, it can shuffle itself into the scatter clusters of a fragmented hard
disk. It can even run on a compressed hard disk.
Windows gives you full menu control of its swap file. You can start the virtual memory
control system by clicking the Virtual Memory button in the Performance tab of Device
Manager. (You start the Device Manager by clicking the System icon in Control Panel.)
Once you click the Virtual Memory button, you should see a dialog box like the one
FIGURE 15.2
The Virtual Memory
dialog box for controlling Windows 95
(and Windows 98)
swap files.
You can opt to let Windows choose the size and place of your swap file or take direct control. By default, Windows puts your swap file in the Windows directory of your C: drive.
By selecting the appropriate box, you can tell Windows the disk and directory in which to
put your swap file and set minimum and maximum limits for its size.
Data Compression
Just as virtual memory can stretch the capacity of your primary storage system, data compression can boost the capacity of your secondary storage system. Compression eliminates
the waste in a storage system. It seeks to put every bit of information in the smallest possible space. In effect, the compression system squeezes the air out of the data stream. Data
compression can reduce fat files into their slimmest possible representation, which can
later, through a decompression process, be reconstituted into their original form.
Most compression systems work by reducing recurrent patterns in the data stream into
short tokens. For example, the two-byte pattern at could be coded as a single byte, such
543
544
as @, thus cutting the storage requirement in half. Most compression systems dont permanently assign tokens to bit patterns but instead make the assignments on the fly. They
work on individual blocks of data, one at a time, starting afresh with each block.
Consequently, the patterns stored by the tokens of one block may be entirely different
from those used in the next block. The key to decoding the patterns from the tokens is
included as part of the data stream.
Disk compression systems put data compression technology to work by increasing the
apparent capacity of your disk drives. Generally, they work by creating a virtual drive with
expanded capacity, which you can use as though it were a normal (but larger) disk drive.
The compression system automatically takes care of compressing and decompressing your
data as you work with it. The information is stored in compressed form on your physical
disk drive, which is hidden from you.
The compression ratio compares the resultant storage requirements to those required by
the uncompressed data. For example, a compression ratio of 90 percent would reduce
storage requirements by 90 percent. The compressed data could be stored in 10 percent
of the space required by its original form. Most data-compression systems achieve about a
50-percent compression ratio on the mix of data that most people use.
Because the compression ratio varies with the kind of data you store, the ultimate capacity
of a disk that uses compression is impossible to predict. The available capacity reported by
your operating system on a compressed drive is only an estimate based on the assumed
compression ratio of the system. You can change this assumption to increase the reported
remaining capacity of your disk drive, but the actual remaining capacity (which depends
on the data you store, not the assumption) will not change.
Most compression systems assume that you want to get back every byte and every bit you
store. You dont want numbers disappearing from your spreadsheets or commands from
your programs. You assume that decompressing the compressed data will yield everything
you started withwithout losing a bit. The processes that deliver that result are called
lossless compression systems.
Sometimes, however, your data may contain more detail than you need. For example, you
might scan a photo with a true-color 24-bit scanner and display it on an ordinary VGA
system with a color range of only 256 hues. All the precise color information in your scan
is wasted on your display, and the substantial disk space you use for storing it could be put
to better use.
Analog images converted to digital form and analog audio recordings digitized often contain subtle nuances beyond the perception of most people. Some data-reduction schemes
called lossy compression systems ignore these fine nuances. The reconstituted data does
not exactly replicate the original. For viewing or listening, the restored data is often good
Data Compression
enough. Because lossy compression systems work faster than lossless schemes and because
their resultant compression ratios are higher, they are often used in time- and space-sensitive applications, such as digital image and sound storage.
Compression has proved to be such a valuable technology that it is used in other ways
besides increasing disk storage. For example, advanced modem protocols often include
data compression to increase throughput levels. In addition, file-archiving software, such
as the popular program PKZip, also takes advantage of compression to more effectively
use your disks space.
File-compression and file-archiving software differs from ordinary disk compression in
several ways. It works on a file-by-file basis rather than across a complete disk. It is not
automatic; you manually select the files you want to archive and compress. The archiving
software does not work on the fly but instead executes upon your command. It compresses
files individually but can package several files together into a single archive file. Archive
files are stored as ordinary Windows files but can be read or executed only after they have
been uncompressed.
Because these archiving systems do not compress on the fly, they can spend extra time to
optimize their compression (for example, trying several compression algorithms to find
the most successful one). They can often achieve higher compression ratios than standard
disk-compression software. (Because they are time insensitive, these programs can try
more complex compression methods and avoid the rule that a compressed file can be
compressed no further.) Your disk compression software, however, cant squeeze their contents any tighter.
Perhaps the most popular application for file-compression software is preparing files for
transmission by modem. It allows you to package together a group of related files, shrink
them to the minimal possible size, and conveniently ship them off with a single send command. Of course, the resultant files will not be further compressible, so your modem will
apparently operate at a slower speed, passing along the compressed data, byte for byte.
All compression systems use essentially the same compression methods, even the same
algorithms. As a result, using more than one of these methods is counterproductive. After
you have compressed data, you cannot squeeze it again (at least using the same algorithm).
Layering multiple levels of compression wont yield more space and may in fact waste
spaceand it definitely impacts performance.
545
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
16
Memory
About the only time most people think about the memory in their computers is when they dont have enough. And all they care to learn about
it is how to add moreor, better still, get someone else to do it for
them.
Theyve got the right idea. With modern computers, memory isnt that
big of a deal. In the past, you had to worry about all sorts of different
kinds of memory with odd names and obscure functions. Now you can
be safe in dealing with memory. Just go to the store, hand over your
plastic, and take home all the memory you need.
Well, its not quite that easy. What you dont know about memory can be
a big handicap and can lead to your buying a handful of unusable chips
and even a computer that no longer works. Moreover, if you have too little memory installed in your computer, there may be no overt symptomsexcept your computer runs a lot more slowly than it could. Youll
never see a warning that says, Slowdown ahead. Add more megabytes.
Windows and your programs leave you on your own to guess whether
you have enough or too little or too much.
Modern computers make adding memory as easy as sliding a small card
called a memory module into a slot. Usually you have nothing more to
adjust. You system comes to life with more memory and greater capabilitiesif it comes to life again at all. Adding the wrong kind of memory
can, in fact, stop it from booting up. Even the relatively benign mistake
of adding memory with too low of a rated speed can compromise your
systems performance. A fast computer can lose a big fraction of its performance when you slide in a module with the wrong rating.
548
Dig a bit deeper into your computer, and youll discover that theres a lot more to memory than just quantity, however. Memory is storage, certainly, but it is also a portal into
your display. Your computers operating system slices and dices it to suit its needs, storing
vital data here and there and stealing blocks of memory addresses for the interfaces it uses.
Memory is more than modules, too, because the same stuff (or at least a close cousin) of
the stuff that makes your programs run also has become a convenient secondary storage
medium for applications on the go. Solid-state memory makes digital cameras and MP3
workable and convenient. Without it, youd need a handful of floppy disks for every song
you want to play or picture youve taken.
Well start our tour of electronic storage technology with a practical look at how much
memory you need, why you come up short even when you have enough, and how much is
enough. From there, well look at the technologies involved in different memory types so
you can understand the differences in the kinds of memory availableand the kind you
might want to install to increase your systems memory capacity. After an examination of
memory moduleswhat your computer uses for its basic memory needswell look at
other memory packages: the cards, chips, and sticks that serve MP3 players and digital
cameras as a mass storage and exchange medium.
Requirements
How much memory you actually need in a computer depends on what you want to do. If
all you want to do is boot up your computer, you need amazingly little memory. The very
first IBM personal computer could boot up with 16 kilobytes of memorythats about
1/100th of what your computer uses for a typical full-screen display on your monitor.
Loading an operating system and the programs you want to run takes more. With the latest of Microsofts Windows operating systems, you need about 4000 times more.
Minimums
The big numbers reflect big changes in computer software. Most people would never
want to tangle with the black screen with a few characters in green pasted on it that the
first computers beckoned with. Adding the friendly and familiar face of Windows (its at
least familiar by now) required about 8MB in its first modern incarnation, Windows 95.
Since then, every major new version of Windows has required about double the previous
version. Table 16.1 summarizes the minimal memory requirements of the various versions
of Windows.
CHAPTER 16 MEMORY
Requirements
TA B L E 1 6 . 1
Memory Requirement of Various Windows Versions
Windows Version
Minimum Memory
Recommended Memory
Windows 95
8MB
16MB
Windows 98
16MB
32MB
Windows Me
32MB
64MB
Windows NT
32MB
64MB
Windows 2000
64MB
128MB
Windows XP
128MB
256MB
Once Windows loads, it can cope with almost any programeven one that uses more
memory than you have in your system. When Windows runs out of physical memory in
your computer, it dips into virtual memory and uses part of your hard disk to simulate
extra solid-state memory. Because virtual memory relies on your hard disk with its physical speed limits, its use slows your system dramatically. You can avoid such slowdowns by
adding memory above the listed minimum to your system. Doubling the memory minimum usually provides enough memory if you mostly rely on running a single application.
If youre seriously into multitasking and regularly switch between applications, or if you
regularly edit large images or videos, youll want even more memoryup to the maximum allowed by your system (if you can afford it).
System Resource Memory

If you run an early version of Windows on your computer, you can often run into out-ofmemory errors, even when you have sufficient memory in your computer and even after
you install more memory. This problem arises because Windows internally allocates
blocks of memory to specific purposes. Windows calls these blocks heaps, and together
they make up system resource memory. For reasons reaching back into the history of the
Windows environment, these blocks are fixed in size and can impose severe limits on the
operation of your computer. You may have dozens of megabytes available in your computer, but Windows will issue an out-of-memory error when one of these heaps
approaches its capacity. It may also tell you to unload some of your applications to free up
memory, although this rarely helps. Windows just doesnt deal with depleted resources
well, and the best cure is usually a reboot.
For example, if you try to put too many icons on the screen at once, you may run out of
memory in a heap. Windows will report too little memory and fail to rasterize the icons,
leaving them as black boxes, even though you have multiple megabytes of free memory in
549
550
your system. Free up some memory, and the icons will come back, but whatever problem
caused the shortfalltypically a poorly designed programwill still nibble away at your
resources, often even after you unload the errant application.
This problem is particularly acute with Windows 95. With Windows 98, Microsoft
increased the storage allocated to system resources, although the problem can still occur
(even with Windows Me). The Windows NT family (including Windows 2000 and
Windows XP) uses a different structure that makes such errors unlikely.
You can check the available system resources in a Windows 95 or 98 system by choosing
the System icon from the Control Panel window. Select the Performance tab, and
Windows will list its memory and system resource usage, as shown in Figure 16.1.
FIGURE 16.1
The Windows
Performance tab.
Maximum Usable Memory

You face four limits to the maximum amount of memory you can install in a computer.
Two are imposed by hardware, one by Windows, and one by your budget.
The physical design of your computer limits the number of memory modules you can
install. You cannot fit more memory into your system than will fit into its memory sockets. The specification of most computers will state this memory maximum. Usually it is
listed as the number of slots for memory modules and the maximum size of module supported. For example, a computer with three slots each allowing a 512MB memory module
has a maximum capacity of 1.5GB.
The chipset in your computer also limits the amount of memory you can install. Although
todays microprocessors readily handle 64GB or more, most chipsets are more severely
constrained. Again, the chipset limits the number of slots for adding modules and maximum module size. These numbers typically match or exceed the physical spaces available
in your computer.
CHAPTER 16 MEMORY
Requirements
Each version of Windows has its own addressing limit. Add more memoryeven when
space is available in your computerand the excess will be unusable. In some cases, too
much memory can even stop your programs and computer from working.
In DOS-based versions of Windowsthat is, Windows 95, 98 and Meyou will run into
a roadblock if you install more than 512MB of memory. Your system may erroneously
report that it does not have enough memory to run a specific application or even
Windows itself, refusing to run the operating system. The problem arises from the 32-bit
cache driver used by these versions of Windows. This driver automatically sets its size
based on the amount of memory it detects in your system. If you have more than 512MB,
it can steal memory from the Windows system area, leaving none for such functions as
loading a DOS prompt.
To use more than 512MB memory with these versions of Windows, you must adjust the
size of vcache in the system.ini file. This is a text file that you can view and modify with
any text editor, such as Notepad (which is included with Windows). But be careful when
editing any system filea mistake can cause programs, and Windows itself might not to
work properly. Find the entry [VCache] and adjust the MaxFileCache setting to 512MB or
less. The entry is in kilobytes rather than megabytes, so you need to specify 524,288 kilobytes or less, as shown here:
[VCache]
MinFileCache=51200
MaxFileCache=524288
In general, the addressing limits of the Windows NT family exceed the physical capacity
of most of todays computers. Earlier Windows versions may, however, constrain the
amount of useful memory you can install in your computer. Table 16.2 summarizes the
maximum memory usable by popular versions of Windows.
TA B L E 1 6 . 2
Windows Addressing Limits
Windows Version
Maximum Addressable Memory
Windows 95
768MB*
Windows 98
768MB*
Windows Millennium Edition
768MB*
Windows NT
4GB
Windows 2000 Professional
4GB
Windows 2000 Server
4GB
Windows 2000 Advanced Server
8GB
continues
551
552
TA B L E 1 6 . 2
Continued
Windows Version
Maximum Addressable Memory
Windows XP Home Edition
4GB
Windows XP Professional
4GB
Windows XP 64-Bit Edition
16GB
Logical Organization
Some people fall into the egregious habit of thinking all memory is alike. Nefarious computer vendors are particular culprits. They are likely to lump together anything inside a
computer involving memory chips as memory, implying that you can use it for running
your programs. To them memory is memory, and thats that. But all memory is not the
same. The random access memory used for running software gets divvied up among multiple functions inside your computer, putting it off-limits to your applications. You often
end up with less usable memory than you might think you have inside your computer.
Some of your computers RAM is compartmentalized for special functions. Computers
with Intel microprocessors reserve a megabyte simply for getting started. Windows assigns
other chunks for its own functions. Engineers describe this compartmentalization of RAM
with a memory map, a graphic representation of the addresses of memory reserved for particular purposes.
In addition to RAM, most computers have other kinds of memory thats physically distinct
and dedicated to specific purposes. This memory often is not within the contiguous
addressing range of your computers microprocessor. For example, the primary and secondary caches in your computers microprocessor are chunks of memory, and each has its
own dedicated purpose. High-performance interface cards may also include a megabyte or
more of cache. The video systems of most computers also include a block of memory
reserved as a frame buffer.
The result is that computers are stuck with a hierarchy of memory types, each with different capabilities and compatibilitiessome useful to some applications, some useless to all
but a few. Rather than improving with age, every advance adds more to the memory mix
up. The classification of memory depends, in part, on the operating system that you run.
Part of modern operating systems is memory management software that smoothes over
the differences in memory type.
Real Mode Memory

The basic memory in any Intel-architecture computer is that which can be addressed
by your computers microprocessor while it is running in real modea special limited
CHAPTER 16 MEMORY
Logical Organization
operating mode thats chiefly used today as your computer boots up. All Intel-architecture
microprocessors start up in real mode.
For todays microprocessors based on the Intel design to be backward compatible and able
to run other software, they must mimic the memory design of the original 8086 family.
The hallmark of this design is the real operating mode in which they must begin their
operation. Because of the original Intel 8086 microprocessor design, real mode only
allows for 1MB of memory. Because it serves the host microprocessor operating in its real
mode, this starting memory is termed real mode memory.
The address range used by real mode memory starts at the very beginning of the address
range of Intel microprocessorszero. The last address is one shy of a megabyte because
counting starts at zero instead of one (that is 1,048,575 in decimal, expressed in hexadecimal as 0FFFFF). Because this memory occurs at the base or bottom of the microprocessor
address range, it is also called base memory.
The first 640KB of real-mode memory is sometimes called lower memory or DOS memory,
because only this small chunk was usable by programs running under DOS, the early personal computer operating system. The rest of the real mode addressing range above lower
memory is called, logically enough, upper memory. In most computers, the top 32KB of
upper memory addresses are occupied by the ROM holding the BIOS code of the system.
The BIOS uses the first kilobyte of space for remembering information about the system
and the location of certain sections of code that are executed when specific software interrupts are made. Among other functions, this memory range holds data used by BIOS
functions, and it is consequently called the BIOS data area.
A design quirk in Intel microprocessors allows them to address extra memory, a total of
64KB minus 16 bytes, above the real mode memory area. Called the high memory area,
this memory cannot be used by ordinary DOS programs but was exploited by the operating system before todays versions of Windows (which run in protected mode) became
popular.
Protected Mode Memory

The rest of the memory that can be addressed by modern microprocessors is termed protected mode memory. As the name implies, this memory can be addressed only when the
microprocessor is running in its protected mode. The address range of protected mode
memory stretches from the top of real mode memory to the addressing limit of your
microprocessor. In other words, it starts at 1MB1,048,576 or 100000(hex)and extends
to 16MB for 286 microprocessors, to 4GB for 386 through Pentium Pro and Celeron
microprocessors, and to 64GB with the Pentium II, III, and 4 microprocessors (including
the Xeon offshoots). Contrasted with base memory, protected mode memory is sometimes
called extended memory.
553
554
Frame Buffer Memory

Computer video systems are memory mapped, which means the color of every pixel on your
monitor gets stored in a memory location that your computers microprocessor can
directly alter, the same way it writes data into memory. Your computer holds a complete
image frame in memory. Your video system scans the memory address by address to draw
an image frame on your monitor screen. The memory that holds a complete image frame
is termed a frame buffer.
Although the frame buffer in most video adapters is physically distinct from ordinary system RAM, it is usually mapped into a block of addresses in the range used by RAM. This
mapping affords the microprocessor the fastest possible access to video memory.
A memory aperture is the address range used by computer peripherals for memory-mapped
input/output operation and control, with one of the most common uses being to reach the
frame buffer. That is, your computer sends data to the frame buffer by writing to the
range of addresses in the aperture. The video controller picks up the data there, converts
it to video, and passes it along to your monitor.
Unified Memory Architecture

Some computers do not use physically distinct frame buffers. Instead, they use a block of
system RAM as the frame buffer, a design called Unified Memory Architecture, sometimes
abbreviated as UMA. This design unifies the memory in a computer by moving the video
frame buffer from its privileged position, connected through a bus to the video controller,
to part of the main memory of the computer. In other words, part of the memory of the
system gets allocated for storing video images. Because both the microprocessor and video
controller have access to this memory, it is sometimes called shared memory architecture.
The chief attraction of this design is that it cuts the cost of a computer. Manufacturers no
longer need to pay for video board memory. But because some of the memory in a UMA
system must be given over to the video buffer, it is not available to your applications.
Hence, a UMA system acts as if its has less memory than the specification sheet implies.
To the digital logic of your computers microprocessor, the UMA design appears no different from a conventional video system. The frame buffer appears in the same logical
location as always. If it did not, your applications would not be able to access it.
Shadow Memory
Although microprocessors used 64-bit data buses, most ROM chips allowed only an eightbit connection. Consequently, microprocessors would slow down to about one-eighth
speed to access code in ROM, which includes the system BIOS. Before Windows was
CHAPTER 16 MEMORY
Technologies
widely adopted, the BIOS code was some of the most used by computer operations, so the
slow access to ROM severely slowed system operations.
To break through this speed barrier, computer designers created shadow memory, a block of
system RAM containing a copy of the contents of the eight-bit BIOS. The software controlling the shadow memory rerouted calls to the BIOS to the fast-memory copy, speeding
up system operation. Although the BIOS is essentially ignored by all recent versions of
Windows, most systems retain shadow memory capabilities. Most BIOS setup procedures
allow you to switch on or off the shadowing of various memory ranges. Switching on
shadowing brings no benefit when running Windows and can interfere with the operation
of some programs. Consequently, shadow memory is best left off.
Technologies
Just as some kinds of memory play different roles in the operation of your computer, the
various kinds of memory work differently and are made differently. Read-write memory is
optimized for speed. It has to run as fast as it can to try to keep up with the high speeds of
modern microprocessorsand it inevitably loses the race. Read-only memory, on the
other hand, takes a more leisurely look at what it remembers, opting for intransigence
instead of incandescent speed.
Both major divisions of the memory world have also endured a long and twisted evolution. Read-only memory has quested ever-higher speed by dashing from one technology
to the next. The old withers and dies, soon forgottenand when not forgotten, expensive.
Older technology memory quickly becomes rare, and with that its price becomes dear.
Unfortunately, you can rarely substitute faster, new technology memory for the slower,
older kind. (Heres a buying tip: When a new technology appears ready to ascend to dominance, youre well advised to stock up on the oldbuy all you think youll ever need.)
With read-only memory, the improvements come mostly in convenience and versatility.
You can do more with newer kinds of read-only memory, but the old types remain useful.
Nearly all kinds of read-only memory remain in production. Although thats good if you
ever need to replace some, you almost never need to replace some. But you do stand to
reap the benefits of newer technologies. For example, without Flash memory, youd have
to stick a battery-hungry disk drive or some other primitive storage device in your digital
camera or MP3 player.
Read-Only Memory
Although read-only memory is usually made from the same silicon stuff as other memory
circuits, silicon is merely a matter of convenience, not a necessity. After all, you can make
read-only memory with a handful of wires and a soldering iron. (Okay, maybe only
555
556
someone who knows what he is doing could, but you get the idea.) Any electrical circuit
that can be permanently changed can be read-only memory.
The challenge for engineers has been to design read-only memory thats easier to change.
That seeming contradiction has a practical basis. If someone had to solder a handful of
wires every time a circuit needed ROM, a computer would be a complicated rats nest of
circuits just waiting for an unpropitious moment to fail. If that doesnt sound too different
from the system on your desk, a machine with wires instead of silicon ROM would likely
be as expensive as a battleship and about as easy to maneuver around your desk. If you
wanted to change a bit of ROMsay, to make an upgrade so your system doesnt crash
whenever AOL delivers your mailyoud have to call a service technician to lug over his
soldering iron and schematic diagrams.
Changeable ROM is upgradeable ROM. And engineers have made steady progress to
make your upgrades easier. The first ROM was as unchangeable as your grandfathers
mind. Then engineers developed chips you could change in special machines. After that
they created chips that could be changed by special circuits inside your computer. Today,
you can update read-only memory in a flashand thats what they call it.
Mask ROM
If ROM chips cannot be written by the computer, the information inside must come from
somewhere. In one kind of chip, the mask ROM, the information is built into the memory
chip at the time it is fabricated. The mask is a master pattern thats used to draw the various circuit elements on the chip during fabrication. When the circuit elements of the chip
are grown on the silicon substrate, the pattern includes the information that will be read
in the final device. Nothing, other than a hammer blow or its equivalent in destruction,
can alter what is contained in this sort of memory.
Mask ROMs are not common in personal computers because they require their programming be carried out when the chips are manufactured; changes are not easy to make, and
the quantities that must be created to make things affordable are daunting.
PROM
One alternative is the programmable read-only memory chip, or PROM. This style of circuit
consists of an array of elements that work like fuses. Too much current flowing through a
fuse causes the fuse element to overheat, melt, and interrupt the current flow, thus protecting equipment and wiring from overloads. The PROM uses fuses as memory elements. Normally, the fuses in a PROM conduct electricity just like the fuses that protect
your home from electrical disaster. Like ordinary fuses, the fuses in a PROM can be
blown to stop the electrical flow. All it takes is a strong enough electrical current, supplied
by a special machine called a PROM programmer or PROM burner.
CHAPTER 16 MEMORY
Technologies
PROM chips are manufactured and delivered with all their fuses intact. The PROM is
then customized for its given application using a PROM programmer to blow the fuses,
one by one, according to the needs of the software to be coded inside the chip. This
process is usually termed burning the PROM.
As with most conflagrations, the effects of burning a PROM are permanent. The chip
cannot be changed to update or revise the program inside. PROMs are definitely not
something for people who cant make up their mindsor for a fast changing industry.
EPROM
Happily, technology has brought an alternativethe erasable programmable read-only memory chip, or EPROM. Sort of like self-healing semiconductors, the data inside an EPROM
can be erased and the chip reused for other data or programs.
EPROM chips are easy to spot because they have a clear window in the center of the top
of their packages. Invariably this window is covered with a label of some kind, and with
good reason. The chip is erased by shining high-intensity ultraviolet light through the
window. If stray light should leak through the window, the chip could inadvertently be
erased. (Normal room light wont erase the chip because it contains very little ultraviolet.
Bright sunshine does, however, and can erase EPROMs.) Because of their versatility, permanent memory, and easy reprogrammability, EPROMs are ubiquitous inside personal
computers.
EEPROM
A related chip is called electrically erasable programmable read-only memory, or EEPROM
(usually pronounced double-E PROM). Instead of requiring a strong source of ultraviolet
light, EEPROMs need only a higher than normal voltage (and current) to erase their contents. This electrical erasability brings an important benefitEEPROMs can be erased
and reprogrammed without popping them out of their sockets. EEPROM gives electrical
devices such as computers and their peripherals a means of storing data without the need
for a constant supply of electricity. Note that whereas EPROM must be erased all at once,
each byte in EEPROM is independently erasable and writable. You can change an individual byte if you want. Consequently, EEPROM has won favor for storing setup parameters
for printers and other peripherals. You can easily change individual settings yet still be
assured the values you set will survive switching the power off.
EEPROM has one chief shortcomingit can be erased only a finite number of times.
Although most EEPROM chips will withstand tens or hundreds of thousands of eraseand-reprogram cycles, thats not good enough for general storage in a computer that
might be changed thousands of times each second you use your machine. This problem is
exacerbated by the manner in which EEPROM chips are erasedunlike ordinary RAM
chips in which you can alter any bit whenever you like, erasing an EEPROM means
557
558
eliminating its entire contents and reprogramming every bit all over again. Change any
one bit in an EEPROM, and the life of every bit of storage is shortened.
Flash Memory
Todays most popular form of nonvolatile memorythe stuff that makes the memory
cards in your camera and MP3 player workis Flash memory. Although strictly speaking,
Flash is a kind of EEPROM, it earns its distinction (and value) by eliminating special circuits for reprogramming. Instead of requiring special, higher voltages to erase it contents,
Flash memory can be erased and reprogrammed using the normal voltages inside a computer. Normal read and write operations use the standard power (from 3.3 to 5 volts) that
is used by the rest of the computers logic circuits. Only erasing requires a special voltage,
one higher than usual called a super-voltage, which is typically 12 volts. Because the supervoltage is out of the range of normal memory operations, the contents of Flash memory
remain safe whether the power is off or your computer is vigorously exercising its
memory.
For system designers, the electrical reprogrammability of Flash memory makes it easy to
use. Unfortunately, Flash memory is handicapped by the same limitation as EEPROM
its life is finite (although longer than ordinary EEPROM), and it must be erased and
reprogrammed as one or more blocks instead of individual bytes.
The first generation of Flash memory made the entire memory chip a single block, so the
entire chip had to be erased to reprogram it. Newer Flash memory chips have multiple,
independently erasable blocks that may range in size from 4KB to 128KB. The old,
all-at-once style of Flash ROM is now termed bulk erase Flash memory because of the
need to erase it entirely at once.
New multiple-block Flash memory is manufactured in two styles. Sectored-erase Flash
memory is simply divided up into multiple sectors that your computer can individually
erase and reprogram. Boot block Flash memory specially protects one or more blocks from
normal erase operations so that special data in itsuch as the firmware that defines the
operation of the memorywill survive ordinary erase procedures. Altering the boot block
typically requires applying the super-voltage to the reset pin of the chip at the same time
as performing an ordinary write to the boot block.
Although modern Flash memory chips can be erased only in blocks, most support random
reading and writing. Once a block is erased, it will contain no information. Each cell will
contain a value of zero. Your system can read these blank cells, though without learning
much. Standard write operations can change the cell values from zero to one but cannot
change them back. Once a given cell has been changed to a logical one with a write operation, it will maintain that value until the Flash memory gets erased once again, even if the
power to your system or the Flash memory chip fails.
CHAPTER 16 MEMORY
Technologies
Flash memory is an evolving technology. The first generation of chips required that your
computer or other device using the chips handle all the minutiae of the erase and write
operations. The current generation of chips have their own onboard logic to automate
these operations, making Flash memory act more like ordinary memory. The logic controls the timing of all the pulses used to erase and write to the chip, ensures that the
proper voltages reach the memory cells, and even verifies that each write operation was
carried out successfully.
On the other hand, the convenience of using Flash memory has led many developers to
create disk emulators from it. For the most effective operation and longest life, however,
these require special operating systems (or, more typically, special drivers for your existing
operating system) that minimize the number of erase-and-reprogram cycles to prolong the
life of the Flash memory.
Read-Write Memory
In the single-minded quest for speed, engineers have developed and cast away a variety of
technologies. Certainly faster chips almost automatically result when they design smaller
chips to fit more megabytes on a silicon sliver, but making chips faster is a lot easier said
than done. By carefully designing chips to trim internal delays and taking advantage of the
latest fabrications technologies, chip-makers can squeeze out some degree of speed
improvement, but the small gains are hard won and expensive.
By altering the underlying design of the chips, however, engineers can wring out much
greater performance increases, often with little increase in fabrication cost. In a quest for
quicker response, designers have developed a number of new memory chip technologies.
To understand how they work and gain their edge, you first need to know a bit about the
design of standard memory chips.
The best place to begin a discussion of the speed limits and improvements in Dynamic
Random Access Memory (DRAM) is with the chips themselves. The critical issue is how
they arrange their bits of storage and allow it to be accessed.
The traditional metaphor for memory as the electronic equivalent of pigeonholes is apt.
As with the mail sorters pigeonholes, memory chips arrange their storage in a rectangular
matrix of cells. A newer, better metaphor is the spreadsheet, because each memory cell is
like a spreadsheet cell, uniquely identified by its position, expressed as the horizontal row
and vertical column of the matrix in which it appears. To read or write a specific memory
cell, you send the chip the row and column address, and the chip sends out the data.
In actual operation, chips are somewhat more complex. To keep the number of connections (and thus, cost) low, the addressing lines of most memory chips are multiplexed (that
is, the same set of lines serve for sending both the row and column addresses to the chip).
559
560
To distinguish whether the signals on the address lines mean a row or column, chips use
two signals. The Row Address Strobe signal indicates that the address is a row, and the
Column Address Strobe signal indicates the address is a column. These signals are most
often abbreviated as RAS and CAS, respectively, with each acronym crowned with a horizontal line indicating the signals are inverses (logical complements), meaning that they
indicate on when they are off. Just to give engineering purists nightmares (and to
make things typographically easier and more understandable, this book uses a slightly different convention, putting a minus sign in front of the acronyms for the same effect).
Multiplexing allows 12 address lines plus the -CAS and -RAS signals to encode every possible memory cell address in a 4Mb chip (or 4MB memory module).
In operation, the memory controller in your computer first tells the memory chip the row
in which to look for a memory cell and then the column the cell is in. In other words, the
address lines accompanied by the -RAS signal select a memory bank; then a new set of
signals on the address lines accompanied by the -CAS signal selects the desired storage
cell.
Even though electricity travels close to the speed of light, signals cannot change instantly.
Changing all the circuits in a chip from row to column addressing takes a substantial time,
at least in the nanosecond context of computer operations. This delay, together with the
need for refreshing, is the chief limit on the performance of conventional memory chips.
To speed up memory performance, chip designers have developed a number of clever
schemes to sneak around these limits.
These memory technologies have steadily evolved. When ordinary DRAM chips proved
too slow to accommodate microprocessor needs, chip-makers first tried static column
memory before hitting on fast page mode DRAM chips (which proved the industry stalwart through 1995). Next, they tinkered with addressing cycles to create Extended Data
Out (EDO). They even tried to improve EDO with a burst-mode version. But these conventional technologies proved too slow to keep up with gigahertz and faster microprocessors.
The current kinds of memory take different strategies. Lower-cost systems use SDRAM
chips (discussed later in this chapter) to coax more speed from the conventional memory
design. The quickest memory systems of the fastest computers break with all the past
technologies to use a new form of memory addressing, commonly called Rambus.
Fast Page-Mode RAM

The starting point for discussions of modern memory is fast page-mode RAM, which was
still an alternative for computers about five years ago. (Other kinds of memory preceded
FPM chips, but the machines using it are obsolete today.)
CHAPTER 16 MEMORY
Technologies
FPM earns its name by allowing repeated access to memory cells within a given page to
occur quickly. The memory controller first sends out a row address and then activates the
-RAS signal. While holding the -RAS signal active, it then sends out a new address and
the -CAS signal to indicate a specific cell. If the -RAS signal is kept active, the controller
can then send out one or more additional new addresses, each followed by a pulse of the
-CAS, to indicate additional cells within the same row. In memory parlance, the row is
termed a page, hence, the name of the chips.
The chief benefit of this design is that your computer can rapidly access multiple cells in a
single memory page. With typical chips, the access time within a page can be trimmed to
25 to 30 nanoseconds, fast enough to eliminate wait states in many computers. Of course,
when your computer needs to shift pages, it must change both the row and column
addresses with the consequent speed penalty.
Extended Data Out Memory

Rather than a radical new development, Extended Data Out (EDO) is a variation on fast
page-mode memory (which allows waitless repeated access to bits within a single page of
memory). The trick behind EDO is elegant. Whereas conventional memory discharges
after each read operation and requires recharging time before it can be read again, EDO
keeps its data valid until it receives an additional signal. EDO memory modifies the
allowed timing for the -CAS signal. The data lines remain valid for a short period after
the -CAS line switches off (by going high). As a result, your system need not wait for a
separate read cycle but can read (or write) data as fast as the chip will allow address access.
It doesnt have to wait for the data to appear before starting the next access but instead
can read it immediately. In most chips, a 10-nanosecond wait period is normally required
between issuing the column addresses. The EDO design eliminates this wait, allowing the
memory to deliver data to your system faster. Standard page-mode chips turn off the data
when the -CAS line switches off. For this system to work, however, your computer has to
indicate when it has finished reading the data. In the EDO design, the memory controller
signals with the Output Enable signal.
In effect, EDO can remove additional wait states, thereby boosting memory performance.
In theory, EDO could give a performance boost as high as 50 to 60 percent over FPM. In
reality and the latest computers, the best EDO implementations boosted performance by
10 to 20 percent.
Physically, EDO chips and SIMMs appear identical to conventional memory. Both use the
same packaging. You cant tell a difference just by lookingunless youre intimately familiar with the part numbers. Telling the difference is important, however. You cant just plug
EDO into any computer and expect it to work. It requires a completely different management system, which means the system (or at least its BIOS) must match the memory
561
562
technology. Although you can install EDO SIMMs in most computers, they will work, if
at all, as ordinary memory and deliver no performance advantage.
Note that the speed ratings of EDO chips are given in nanoseconds, much like page-mode
chips. For a given nanosecond rating, however, EDO memory will act as if it is about 30
percent faster. For example, whereas a 70 ns page-mode chip can deliver zero wait state
operation to a 25MHz memory bus, a 70 ns EDO chip can operate at zero wait states on a
33MHz bus.
Burst EDO DRAM

To gain more speed from EDO memory, Micron Technology added circuitry to the chip
to make it match the burst mode used by Intel microprocessors since the 486. The new
chips, called Burst EDO DRAM (BEDO), perform all read and write operations in fourcycle bursts. The same technology also goes by the more generic name pipeline nibble mode
DRAM, because it uses a data pipeline to retrieve and send out the data in a burst.
The chips work like ordinary EDO or page-mode DRAM in that they send out data when
the -CAS line goes active. However, instead of sending a single nibble or byte of data
(depending on the width of the chip), a two-bit counter pulses the chip internally four
times, each pulse dealing out one byte or nibble of data.
Although BEDO was relatively easy and inexpensive to fabricate, because it requires a
minimum of changes from ordinary EDO or page-mode DRAM, it never caught on with
computer-makers. They opted to leapfrog the design and move to far faster technologies.
Synchronous DRAM
Because of their multiplex operation, ordinary memory chips cannot operate in lock-step
with their host microprocessors. Normal addressing requires alternating cycles. By
redesigning the basic chip interface, however, memory chips can make data available every
clock cycle. Because these resulting chips can (and should) operate in sync with their computer hosts, they are termed synchronous DRAM (SDRAM).
Although altering the interface of the chip may remove system bottlenecks, it does nothing to make the chip perform faster. To help SDRAM chips keep up with their quicker
interface, a pipelined design is used. As with pipelined microprocessors, SDRAM chips are
built with multiple, independently operating stages so that the chip can start to access a
second address before it finishes processing the first. This pipelining extends only across
column addresses within a given page.
Some SDRAM memory is registered, which means that the memory controller on the
motherboard relies on the memory modules themselves to drive and synchronize their
own memory control signals. Nonregistered modules get the required synchronizing
CHAPTER 16 MEMORY
Technologies
signals from the memory controller. Through some strange quirk in engineering language, nonregistered memory is generally called unbuffered, whereas buffered memory is
termed registered.
All SDRAM chips suffer from a delay between when your computer makes its request to
read memory and the time valid memory becomes available. Engineers call this delay CAS
latency, because it is the measure of time between when your computer applies the
Column Address Strobe signal and when data becomes available. They measure the delay
in clock cycles. With todays memory products, the CAS latency is typically two or three,
and it is a function of the memory chips themselves. Chips and memory modules are rated
as to their CAS latency (although you may need to be an engineer to dig the information
out of a products data sheet). Knowing the CAS latency of memory is useful because it is
one of the parameters you can adjust with some BIOS setup procedures to speed up your
computer.
Single Data-Rate Memory
Ordinary SDRAM memory is often called single data-rate (SDR) memory in the modern
world to contrast it with, well, double data-rate (DDR) memory. It earns its new name
because SDR memory modules transfer data at the same rate as the system bus clock rate.
On each clock pulse, SDR transfers a single bit down each line of its bus.
SDR memory is rated by the speed of the system bus to which it attaches, although only
two speeds are commonly recognized and sold: PC100 for computers with 100MHz system buses, and PC133 for computers with 133MHz system buses. You can assume that
unrated SDR memory is meant for 66MHz system buses, although the designation PC66
appears sometimes.
Double Data-Rate Memory
Memory chips that use DDR technology are rated by their effective data speed (that is,
twice their actual clock speed). Three speed ratings of chips are available. They are designated DDR 200, DDR 266, and DDR 333.
The speed ratings on DDR memory modules are based on peak bandwidth rather than bus
speed. The result is a much more impressive number. For example, a DDR module on a
100MHz bus transfers data at 200MHz, but its peak bandwidth is 1.6GBps. A module at
this speed becomes a PC1600 memory module. These large figures serve to distinguish
DDR from SDR modules, so you are less likely to confuse the two. Table 16.3 lists the
rated speeds and bus speeds of DDR modules.
563
564
TA B L E 1 6 . 3
DDR Memory Speed Ratings
DDR Speed Rating
Module Peak Bandwidth
Data Rate
System Bus Speed
PC1600
1.6GBps
200MHz
100MHz
PC2100
2.1GBps
266MHz
133MHz
PC2700
2.7GBps
333MHz
166MHz
Note that both SDR and DDR memory have two busesone for addressing and one for
transferring data. In DDR, only the data bus operates at double speed. The address bus
still works at the standard clock speed. Consequently, DDR memory only speeds up part
of the memory cycle, the part when data actually moves. Because most memory transfers
now occur in bursts, this handicap is not as substantial as it might seemfor some
requests, DDR memory may need only one address cycle for a burst of a full page of
memory, so the overhead is only one slow cycle for 4096 DDR memory transfers.
Quad Data-Rate Memory
The next generation of SDRAM is exactly what you might expectquad data-rate (QDR)
memory. This new technology has already been developed through the joint efforts of
Cypress, Hitachi, IDT, Micron, NEC, and Samsung. Chips using QDR have already been
developed, but they have not been applied to memory modules. Currently there are not
even specifications for QDR modules.
Rather than doubling up the data speed once again, QDR uses a double-ported design.
Each chip has an input and an output port that can operate simultaneously, and each port
uses double-clocking to transfer data. The result is that information can move through a
QDR chip four times faster than ordinary SDRAM, hence the quad in the name.
According to the QDR promoters, the double-ported design eliminates even the possibility of contention between the memory chip and its controller. In addition, QDR boosts
the speed of the address bus to the same double rate as used by the data bus, giving QDR
an automatic edge on DDR.
The developers of QDR maintain an informational Web site in support of the technology
at www.qdrsram.com.
Rambus DRAM
The next step up in memory speed comes from revising the interface between memory
chips and the rest of the system. The leading choice is the Rambus design, developed by
the company with the same name. Intel chose Rambus technology for its fastest systems
designs. An earlier incarnation of the technology was used in the Nintendo 64 gaming
system.
CHAPTER 16 MEMORY
Technologies
The Rambus design has evolved since its beginnings. Rambus memory chips use an internal 2048-byte static RAM cache that links to the dynamic memory on the chip through a
very wide bus that allows the transfer of an entire page of memory into the cache in a single cycle. The cache is fast enough that it can supply data at a 15-ns rate during hits.
When the cache misses, the chip retrieves the request data from its main memory and at
the same time transfers the page containing it into the cache so that it is ready for the next
memory operation. Because subsequent memory operations will likely come from the
cache, the dynamic portion of the chip is free to be refreshed without stealing system time
or adding wait states.
The Rambus operates like a small network, sending data in packets that can be up to 256
bytes long. The system uses its own control language to control memory and steer bytes
around. The overhead from this system saps about 10 percent of the bandwidth from the
peak transfer rate of the system.
You wont ordinarily deal with individual Rambus chips but rather complete memory
modules. Because of the different technology used by Rambus chips, the modules work
differently with your system, too. Whereas with conventional memory it is important to
match the width of the memory bus to that of the rest of your computer, such matches are
unnecessary with Rambus. The memory controller reorganizes the data it pulls from
Rambus memory to make it fit the bus width of the host computer. Consequently, a
Rambus module with a 16-bit bus works in a computer with a 64-bit bus. Usually, however, Rambus systems put two or more modules in parallel. The goal of this design is to
increase memory speed or bandwidth.
Rambus modules dont link to a computer system like standard memory. Instead, the
design uses a special high-speed bus (hence, the origin of the Rambus name). In current
Rambus modules, memory connects to the chipset in your computer through two highspeed buses. One of them, called the request bus, is used to send control and address information to memory. This bus uses an eight-bit connection. In addition, the data bus is a
16-bit-wide channel that transfers the data for reading from and writing to memory. The
two buses are synchronized, although the data bus transfers bits on both edges of the
clock, much as is done with DDR memory. The speed rating of Rambus modules is usually given by the data rate on the data bustwice the actual clock frequency of the overall
memory system. Rambus modules also have a third, low-speed bus thats used only for initialization and power management.
The Rambus itselfthe lines that carry the request bus, data bus, and timing signalsis
designed as a transmission line like those carrying radio signals. Transmission lines are
sensitive to changes in tuning, much like the rabbit-ear antennae on portable televisions
that, if not perfectly adjusted, make a scene look like a blizzard across the prairie. The
Rambus transmission line loops through each module in a Rambus circuit before stopping
565
566
at the line termination at its end (which terminates the signal by absorbing it all and not
letting any reflect back down the transmission line to interfere with other signals).
This design has important practical implications. Because the Rambus signals loop
through each module, in current computers all Rambus sockets must be filled with memory modules. Most computer-makers design their systems to accommodate two Rambus
modules per Rambus circuit. Typically, computer-makers install one memory module at
the factory to allow you a socket for future memory expansion. Because even the Rambus
signal must loop through unused sockets, the second socket in each Rambus circuit in
most new computers gets filled with a dummy module. The memory system will not
work unless dummy modules (or memory upgrades) are properly installed.
To increase memory speed and capacity, most of todays computers use two Rambus circuits. This doubles both the total bandwidth of the memory system and the potential total
memory in the system. As a consequence, computers that use todays 16-bit Rambus modules must expand their memories two modules at a time.
In the future, Rambus modules will use both 32-bit and 64-bit bus widths in addition to
the 16-bit modules currently in use. Modules with 32-bit buses will be essentially the
same as two 16-bit modules on one card with two request buses and two data buses. The
32-bit modules will use a different design with a single shared request bus and four data
buses. These wider bus modules are incompatible with todays computers and their memory systems. Consequently, Rambus designed the module connectors to be physically
incompatible, too, so you cannot inadvertently slide the wrong kind of Rambus module
into your computer.
Current Rambus modules for personal computers are designed to operate at 800MHz or
1066MHz. Because of the double clocking of the data buses in the modules, 800MHz
modules perfectly match 400MHz system buses, and 1066MHz modules match 533MHz
system buses. In older computers using Rambus, 100MHz system buses are best matched
by 800MHz modules, and 133MHz system buses best match 1066MHz modules.
Video Memory
Memory access problems are particularly prone to appear in video systems. Memory is
used in display systems as a frame buffer, where the onscreen image is stored in digital
form with one unit of memory (be it a bit, byte, or several bytes) assigned to each element
of the picture. The entire contents of the frame buffer are read from 44 to 75 times a second as the stored image is displayed on the monitor screen. All the while, your computer
may be attempting to write new picture information into the buffer to appear on the
screen.
With normal DRAM chips, these read and write operations cannot occur simultaneously.
CHAPTER 16 MEMORY
Packaging
One has to wait for another. The waiting negatively affects video performance, your systems speed, and your patience.
The wait can be avoided with special memory chips that have a novel design twisttwo
paths for accessing each storage location. With two access paths, this memory acts like a
warehouse with two doorsyour processor can push bytes into the warehouse through
one door while the video system pulls them out through another. Strictly speaking, this
memory can take two forms: True dual-ported memory allows simultaneous reading and
writing; video memory chips (often called VRAM for video random access memory) give
one access port full read and write random access, while the other port only allows
sequential reading (which corresponds to the needs of scanning a video image).
The chief disadvantage of VRAM technology is that it is more expensive because it
requires more silicon (about 20 percent more area on the chip die). It more than makes up
for its higher cost with its speed advantage. Using VRAM can speed up video systems by
as much as 40 percent.
Packaging
Most memory comes in the form of chips, individual integrated circuits meant for permanent installation on a printed circuit board. The capacity of each chip is measured in
bitsor, as is more likely in the modern world, megabits. The first chips had narrow, onebit-wide data buses, so they moved information one bit at a time. To achieve a byte-wide
data bus, eight chips had to be used together (under the coordination of the memory controller). Most modern chips use wider data busesfour or eight or more bitsbut none
comes close to matching the 64-bit data buses of modern computers.
To make memory more convenient to install and upgrade in practical computers, memory-makers package several memory chips on a small circuit board to make a memory module. Despite the added circuit board, modules provide a more compact memory package
because the chips are soldered to the modules, thus eliminating space-wasting sockets.
Soldering the chips down also allows them to be installed closer together (because no
individual access is required). Moreover, because the discrete chips that are installed in the
memory module never need to be individually manipulated except by a machine, they can
use more compact surface-mount packages. A single memory module just a few inches
long can thus accommodate a full bank of hundreds of megabytes of memory.
Memory modules are the large economy sizeRAM in a bigger package to better suit the
dietary needs of todays computers. Besides the more convenient package that allows you
to deftly install a number of chips in one operation, the memory module also better
matches the way your computer uses memory. Unlike most chips, which are addressed at
the bit level, memory modules usually operate in bytes. Whereas chip capacities are mea-
567
568
sured in kilobits and megabits, memory modules are measured in megabytes.

The construction of a memory module is straightforward; its simply a second level of
integration above the basic memory chip. Several chips are brought together on a small
glass-epoxy circuit board with their leads soldered down to the circuit traces and the
entire assembly terminated in an external connector suitable for plugging into a socket or
soldering to another circuit board.
Memory modules come in a variety of types, many of which are no longer common. Single
Inline Memory Modules, which are commonly called SIMMs, connect the contacts on the
opposite sides of the board together so that each pair provides a single signal contact.
Single Inline Pin Package modules, or SIPPs, have a row of pins projecting from the bottom
of the module instead of edge connectors. Both of these packages are no longer used in
new computers.
Several designs are in current use. Dual Inline Memory Modules, called DIMMs, put separate signals on each contact so that the contacts on each side of the board serve different
purposes. Small Outline DIMMs, termed SoDIMMs, shrink the package to fit more compact computers. Rambus Inline Memory Modules, or RIMMs, basically follow the DIMM
design but use Rambus instead of SDRAM memory.
Dual Inline Memory Modules

Engineers created the DIMMs to accommodate the needs of Pentium computers. Earlier
module designsin particular SIMMshad 8- or 32-bit data buses. As a consequence, a
single bank of 64-bit Pentium memory required several modules. Not only were multiple
modules a pain when you wanted to upgrade, they took up valuable motherboard space.
Moreover, the many sockets they required were incompatible with the high-speed needs
of modern memory systems. With more sockets and longer circuit traces, the more difficult it is to achieve high-frequency operation. Figure 16.2 shows a typical DIMM.
FIGURE 16.2
A 168-pin DIMM module showing components.
All DIMMs provide a 64-bit data bus that exactly matches the needs of Pentium-class
computers. Practical module capacity now ranges from 64MB to 512MB. Larger capacity
modules often put memory chips on both sides of the circuit board. To fit even more
memory on each module, one manufacturer (Kingston Technology) has developed a
CHAPTER 16 MEMORY
Packaging
stacking scheme that piggybacks a physically larger chip over a smaller package, thus doubling the holding capacity of a single module.
DIMMs come in three types: Those using old-technology (FPM or EDO memory) chips,
those for SDR DRAM memory, and those for DDR memory. In size, the three types of
modules are identical. They are 5.25 inches wide and (usually) 1 or 1.5 inches tall,
although modules with unusually large memory endowments often are taller. Figure 16.3
shows the dimensions of a typical DIMM. To prevent you from sliding a DDR module
into a slot meant for a SDR module, the two designs use different connector designs that
are keyed to be incompatible.
FIGURE 16.3
Dimensions of a typical 168-pin DIMM.
DIMMs can be equipped with non-parity, parity, or error-correction code (ECC) memory. The packages look identical and slide into the same sockets. Old-technology memory
modules use signals on several connector contacts called presence detect pins to indicate
these parameters. SDR and DDR modules use a more sophisticated design that is sometimes called serial presence detect. Each module is equipped with 2048 bytes of onboard
nonvolatile memory that your computer can read and use to identify the module memory
type. After reading this information, your computer can then configure itself appropriately.
Old Technology DIMMs

Memory modules that use old memory technology have edge connectors with 168 separate contacts arrayed on both sides of their small circuit boards. Eight of these contacts
are reserved for presence-detect indications. In particular, these presence-detect signals
indicate the speed rating of the module. Four of these contacts indicate the organization
and addressing characteristics of the memory on the module. One pin, PD5, indicates
whether the module contains FPM or EDO memory. Two, PD6 and PD7, indicate the
speed rating of the chip, as indicated by the code listed in Table 16.4. (Note the rating in
nanoseconds does not correspond to todays PC66/100/133 nomenclature because these
modules are not used in modern computers.) The eighth contact indicates whether the
module uses parity or ECC technology. In addition to the eight presence-detect indicators, two other contacts carry module ID signals that describe the module type and
refresh mode used by the chip.
569
570
TA B L E 1 6 . 4
Old-Technology (FPM and EDO) 168-Pin DIMM Speed Code
Signal
Pin Number
60 ns
70 ns
80 ns
PD6
165
PD7
82
SDR DIMMs
SDR modules have 168 pins on their bottom edge. They are keyed with two notches,
dividing the contacts into three groups with short gaps between. The first group runs
from pin 1 to pin 10; the second group from pin 11 to pin 40; and the third group from
pin 41 to pin 84. Pin 85 is opposite pin 1. The notches are designed to prevent you from
sliding a smaller SIMM with fewer connections into a DIMM socketthe longest space
between two notches is shorter than the shortest SIMM. The asymmetrical arrangement
of the notches on the DIMM prevent its inadvertent improper insertion into its socket
turn a DIMM end-for-end and it wont fit into a socket. Some DIMMs have holes to
allow you to latch the modules into their sockets, although some DIMMs lack these holes.
DDR DIMMs
The standard DIMM for DDR memory has a connector with 184 pins. It bears a single
notch near the center of the connector. Otherwise, DDR modules look much the same as
SDR modules and can be handled similarly.
Rambus Inline Memory Modules

Because of the radical interface required by Rambus memory, memory-makers redesigned
individual modules to create a special Rambus Inline Memory Module (RIMM) to accommodate the technology. The basic module appears similar to a standard DIMM, with individual memory chips soldered to a printed circuit board substrate that links to a socket
using a conventional edge connector. Despite the relatively narrow bus interface used by
the Rambus systems (16 bits), the RIMM package pushes its pin count to 184, partly
because alternate pins are held to ground potential as a form of shielding to improve stability and decrease interference at the high clock frequency for which the modules are
designed (800MHz).
RIMMs also differ in their operating voltage. The Rambus standard calls for 2.5-volt
operation, as opposed to the 3.3 volts standard with DIMMs. Future RIMMs may operate
at 1.8 volts.
The standard 184-pin design supports both non-parity and error-correction RIMMs, with
bus widths of 16 and 18 bits, respectively. Figure 16.4 shows a RIMM package.
CHAPTER 16 MEMORY
Packaging
FIGURE 16.4
A RIMM package.
Small Outline DIMMs

Although convenient for desktop computers, full-size memory modules are unnecessarily
large for notebook machines. Consequently, the makers of miniaturized computers
trimmed the size of ordinary memory modules about in half to create what they called the
Small Outline DIMM (SoDIMM).
The first SoDIMMs were offshoots of the 72-pin SIMMs used for FPM and EDO memory. They had the same 72 contacts as full-size SIMMs but arrayed them on both sides of
the module (making it dual inline). In a SIMM, the contacts on opposite sides of the
modules circuit board are electrically connected together. On a SoDIMM, each has a different function, putting the connection space to more efficient use and allowing the
smaller size. Figure 16.5 illustrates a Small Outline Dual Inline Memory Module of this
design.
FIGURE 16.5
A 72-pin Small
Outline Dual Inline
Memory Module
package.
As you would expect, a 72-pin SoDIMM is about half the length of a 72-pin SIMM, measuring about 2.35 inches long. As with other module styles, a notch at one end of a 72-pin
SoDIMM prevents you from latching it into its socket with the wrong orientation. The
notch is on your left when you look at the chip side of the SoDIMM. Figure 16.6 shows
the dimensions of a typical SoDIMM.
FIGURE 16.6
Dimensions of a typical 72-pin Small
Outline Dual Inline
Memory Module.
571
572
The SoDIMM package proved so compelling it has been adapted for SDRAM modules.
The SDRAM SoDIMMs are slightly longer than those used by older technologies, measuring 2.67 inches (67.6 millimeters) long.
As with full-size DIMMs, SDRAM SoDIMMs come in two physically incompatible styles
for SDR and DDR.
SDR SoDIMMs have 144 contacts on their edge connectors. DDR modules have 200.
Each style of SoDIMM has a single notch, but the position of the notch is different for
SDR and DDR modules. With SDR modules, the notch is only slightly offset from the
center; with DDR modules, the notch is near one edge (the pin 1 side) of the module.
Figure 16.7 shows the difference in the notches used by the two varieties of SDRAM
modules.
FIGURE 16.7
The difference
between the SDR and
DDR SoDIMM notches.
Installation
In general, the memory of your computer will work throughout the life of the machine
without requiring any attention. Memory has no moving parts that require service or
lubrication. The only time youre likely to tangle with memory is when you decide to add
to your computers native endowment, either by adding to its main memory on the motherboard or by adding an expansion board to enhance the size of a disk cache or video
frame buffer.
Memory installation is a process that requires care but does not demand expertise or
extreme precision. You can usually add a few megabytes in a few minutes without worrying about damaging the memory, your computer, or yourself.
The care required involves taking simple precautions. Avoid static electricity, which can
damage any solid state circuitry, including memory chips. The most cautious sources usually recommend that you ground yourself to a cold water pipe before you dig into the circuitry of your computer. Most people just plug in chips or modules without a thought to
static. Unless you draw sparks when you touch things on your workbench, youre probably safe to install memory modules.
In that one of the primary design goals of memory modules is simplified installation,
modules make expanding the memory of your computer easyor at least easier than it
CHAPTER 16 MEMORY
Installation
would be with a box full of discrete chips. Modules do not, however, make memory installation or expansion trouble free. You still must match modules with your computer and
insert them in their sockets properly.
Memory Speed
Memory speed is one of the most critical factors in computer performance. Faster memory can move data to the microprocessor faster. Todays microprocessors are so fast that
they outrace any kind of memory, so anything you do to make your systems memory
faster can help performance.
Most of what you can do to speed up memory requires an advanced degree in tinkering.
You can experiment with the advanced setup adjustments in your computers BIOS (if it
affords you such adjustments) to see how many corners you can cut before your computer
freezes up and dies.
Mostly what you do with memory makes your computer slower. The design of your computer limits the top speed at which its memory can operate (even though faster memory
might benefit your computer). Adding faster memory wont break through this speed
limit. On the other hand, add memory thats too slow when you expand your system, and
your whole system might shift into low gear. Modern memory systems can detect the
speed rating of memory modules, and they will slow down their operation to match the
slowest memory in your computer.
The best strategy when adding memory to your computer is to duplicate what you already
have. If your system has PC100 memory installed, always install more PC100 memory.
Mixing Modules
Modern memory module design precludes you from using the wrong kind of memory in
your computer. Because memory modules are made from discrete chips, they use the same
underlying technologies. SIMMs made using static, dynamic, video, and page-mode RAM
technologies are available. As with discrete chips, different module technologies are not
completely interchangeable, so you need to specify the correct type when ordering memory for your computer.
At one time, however, the different memory typesfast page-mode, EDO, and Burst
EDOall used exactly the same style of module: the standard SIMM. In general, a computer designed for fast page-mode SIMMs could use other SIMM technologies but
derives no benefit from doing so. Similarly, the higher speed of the more advanced memories could be wasted. Sliding a fast page-mode SIMM into an EDO socket would slow
your entire system down to the fast page-mode speed.
573
574
With DIMMs, you ordinarily do not need to match modules. Each one constitutes a full
bank of memory. You should be able to mix different sizes and install modules in any
order. Mixing module speeds will only result in your system operating at the speed of the
slower module (if it doesnt reject the laggardly memory as too slow). In reality, you may
find slight differences between modules and motherboards. (A motherboard can sometimes be persnickety about where you put the modules.) That said, upgrading memory is
much easier than it used to be, and youre much less likely to encounter problems if you
just slide modules into the sockets that they fit into.
Contact Material
At one time, the contacts used by memory modules sometimes came plated with tin and
sometimes with gold. The purpose of this plating is to prevent oxidation or corrosion that
would increase the resistance of the contact and lower the reliability of the connection. In
the abstract, gold is the preferred material because it oxidizes less readily. Gold also costs
substantially more, encouraging manufacturers to opt for tin.
One manufacturer of electronic component sockets, AMP, points out that the material is
not so important as the match. You should choose memory modules with contacts plated
with the same material as used in the sockets in your computer. The explanation goes
deeper than the plating.
According to AMP, putting tin in contact with gold and exerting pressure on the contact
results in a problem called fretting. Part of the tin rubs off, attaches to the gold, and oxidizes there, thus building up a thin layer of tin oxide. The longer the two materials contact one another, the thicker the oxide build up.
Gold is not the only material susceptible to this oxide buildup. It can develop when tin
contacts nearly any dissimilar metal. The problem does not arise when tin-plated contacts
press against other tin-plated contacts. Of course, the tin surface will naturally oxidize on
the tin-plated contacts, but when you insert a tin-plated edge connector into a tin-plated
socket, the soft tin gives way. Pushing the contacts together squeezes or wipes the tin
oxide away from the contact area so that tin presses against tin.
Today, nearly all DIMMs and their sockets use gold plating, so youre unlikely to
encounter problems with modern computers. With older SIMMs, however, youll find
both contact materials are prevalent. Matching material is the best bet. If you have an
older computer with a metals mismatch in its sockets and memory errors, you may be able
to correct the problem by cleaning the contacts on the memory modules. Although special
contact cleaners are available, you will get good results wiping the contacts clean with an
ink eraser (the kind that is slightly abrasive). Dont clean too hard, though, because you
may wipe the contact material entirely off.
CHAPTER 16 MEMORY
Memory Cards
Orientation
When installing memory modules, remember that their orientation is important. In general, all modules in a computer will face in the same direction. Of course, with two-sided
modules with components on both sides, you cant easily tell which direction anything is
facing. The key is to look at the notches in the edge connector at the bottom of the module. Youll see the notches line up with tabs in the socket when the module is oriented in
only one directionthe right way. Should you try to insert a module and discover that it
refuses to fit, you probably have it in the wrong orientation or improperly seated.
Memory Cards
Computers dont have a monopoly on memory. Todays high-tech toys such as MP3 players and digital cameras also need memory for storing their softwarethe music you play
on your MP3 player and the pictures you take with your digital camera.
The requirements for this memory differs from that you install in your computer.
Although you want computer memory to be as fast as possible, speed is not as important
as permanence for other electronics. You want your music and pictures to always be there
when you want them, not subject to the vagaries of your batteries or utility electricity supply. In a word, cameras and MP3 players want nonvolatile memory.
In addition, you want to be able to quickly add to or exchange the contents of the memory in your camera or MP3 player or to interchange blocks of memory. To allow you to
make such exchanges quickly and easily, memory-makers have developed a wide array of
memory cards. They work like a PC Card designed to the PCMCIA standard (we considered a PC Card as an expansion bus, which it really is, in Chapter 9, Expansion Buses).
You simply slide them in, and they work. But, thanks to the benefit of more refined technologies and less need to accommodate non-memory devices such as network adapters
and modems, memory cards can be smaller. Some, in fact, are hardly bigger than memory
chips.
The first two designs, CompactFlash and SmartMedia, came from two different but wellestablished directions. CompactFlash simply put the PC Card interface on a smaller card.
SmartMedia did little more than repackage Flash memory chips. The latest additions to
the memory card zoo all add copy management provisions so that music publishers can
control what you put on cards and whether (or even how many times) you duplicate copyrighted material.
The question as to which is better often is meaningless. The best card is the one that your
camera or music player uses. However, the medium used may influence your choice of
camera or player. A device using CompactFlash gives you more storage and versatility,
although other cards yield more compact storage.
575
576
CompactFlash
Sometimes, even a PC Card is too big, so in a quest for miniaturization, the electronics
industry, led by Intel, developed a yet-smaller standard. Appropriately called Miniature
Card, it was designed as a reduced-size version of a PC Card aimed particularly at memory devices. Although it never caught on, its heir, CompactFlash, has found wide application in digital cameras. Although engineers designed the Miniature Card to be a
general-purpose expansion format, one application dominated the minds of designers eyeing the small cards: memory. A variety of electronic devices have a need for compact and
removable memory, perhaps the most important of which initially was the digital camera.
Intel developed Miniature Card and released the initial version (logically enough, version
1.0) on February 29, 1996. Envisioned as the perfect memory solution for digital cameras
and miniaturized digital audio recorders, Miniature Cards measured less than an inch and
a half square but could store up to 64MB using any standard memory technology. The
limit was imposed by the bus design, pure PC Card.
In 1994, PC Card maker SanDisk developed a new small card design as CompactFlash. As
the name implies, the company aimed the product specifically at accommodating Flash
memory in a format small enough to fit into a digital camera that wouldnt require a mule
team to transport.
The CompactFlash specification is now governed by the CompactFlash Association
(CFA), founded in 1995 by Apple Computer, Canon, Eastman Kodak, Hewlett-Packard,
LG Semicon, Matsushita, Motorola, NEC, Polaroid, SanDisk, Seagate, and Seiko Epson.
The organization maintains a Web site at www.compactflash.org.
An early promoter of the various PCMCIA specifications, SanDisk patterned the signals
used by CompactFlash cards on the PC Card bus design. Logically, the interface conforms
to the AT Attachment specifications, and software accesses the memory on the card as if it
were a disk drive.
Electrically, CompactFlash cards act much like full-size PC Cards. As a result, each card
must include its own control logic and conform to the logical format standards promulgated by PCMCIA. This makes the cards more complex than the competing tiny-card
format SmartMedia, discussed next. Although the larger amount of circuitry also adds to
the cost of producing the cards, large-scale integration and high volumes hold the potential of minimizing any such handicap. In fact, the prices of the two media are comparable,
although the edge usually goes to SmartMedia.
The added complexity of PC Card compatibility has one big benefit. It allows
CompactFlash cards to be readily adapted to full-size PC Card slots. The required adapter
card need not contain any electronics, thus making it inexpensive. The only need is to
CHAPTER 16 MEMORY
Memory Cards
make the small card physically fit into a larger socket, rerouting its circuits to fit the larger
connector used by the PC Card standard.
Standard CompactFlash cards measure 43 mm (1.7 in) by 36 mm (1.4 in) by 3.3 mm (0.13
in), the same thickness as a PC Card and about one-quarter the size. To increase physical
capacity, the CFA has designed a slightly thicker format, designated CompactFlash Type
II. At 5 mm thick, the new format conforms to the thickness of Type II CardBus and PC
Cards.
Unlike physical disk drives, however, CompactFlash cards can withstand more physical
abuse (about 2000 Gs) and wider temperature ranges (-25 C to +75 C for CompactFlash
compared to +5 C to +55 C for typical disk drives). The expected life of the medium is
in excess of 100 years, with data integrity protected by both built-in dynamic defect management and error correction.
CompactFlash capacities start at 2MB. Currently the largest capacity per card is 512MB.
Up to 1GB is available in the same package using a micro hard disk drive. Systems can
double the data storage of any card using compression software, although digital cameras
use JPEG compression, which ordinarily cannot be further compressed by standard software-compression algorithms.
The CompactFlash design requires dual-voltage operation. All cards are completely compatible with both 5.0- and 3.3-volt systems. Any card will work in any system, regardless
of its operating voltage.
Most operating systems support the PCMCIA logical interface used by CompactFlash
cards through software drivers. CompactFlash cards can be used with DOS, Windows 95
and 98, OS/2, Apple Macintosh, and most flavors of Unix.
SmartMedia
SmartMedia is the pioneering minimal memory format, at one time (but no longer) the
thinnest and smallest type of removable Flash memory card. About one-third the size of a
PC Card, a SmartMedia chip (theres no better description) measures 45 millimeters (mm)
by 37 mm by 0.76 mm. It appears like a monolithic slice of plastic, and thats almost what
it is. Its overmolded thin package (OMTP) simply encases one or more memory chips
with a solid protective outer shell. Two rows of a total of 22 surface contacts provide the
necessary connections for an eight-bit interface.
Originally developed under the name Solid-State Floppy Disk Card (SSFDC), a name that
still rules over its supporting bodys Web site (www.ssfdc.com), its original creator
(Toshiba) has trademarked the name SmartMedia for the format. The SmartMedia format
is currently used by the makers of many digital cameras, including nearly all Japanese
brands as well as Agfa and Apple.
577
578
The small size of the SmartMedia cards does not stand in the way of large capacities.
Although originally introduced only in sizes from 2MB to 8MB, Toshiba has developed a
high-density connection and packaging system to increase the potential storage of the
SmartMedia format. Current specifications allow cards storing up to 128MB.
The total capacity of each card is arrayed in blocks of about 16KB, each block comprising
32 pages of 528 bytes each. Because the memory is solid-state, access time is brief, rated at
only 7 microseconds. Transferring a bit through the serial interface requires only 50
nanoseconds, an effective clock speed of 20Mbps. In actual operation, transfer speeds of
0.5 to 1.0Mbps can be achieved. Writing is slowerabout two milliseconds to erase a
16KB block of memory and 200 microseconds to write it again.
The heart of each SmartMedia card is one or more NAND Flash EEPROM (electrically
erasable programmable read-only memory) chips embedded in the plastic shell. The
NAND Flash memory technology used by SmartMedia provides fast write and erase functions plus high-speed read with minimal power consumption.
This simple electrical design is key to the low-cost potential of SmartMedia. Despite the
name, the cards are not very intelligent. They contain only memory. The controller is in
the host. To work in a PC Card slot, an adapter for a SmartMedia card must itself contain
a controller. Adapters are consequently more expensive than slot adapters for PC Cards.
The SmartMedia logical data format is based on the AT Attachment and DOS file standards and is compatible with computers running Windows 95, Windows 98, Windows
NT, or OS/2 Warp. It also works on Apple Macintosh computers.
Both 5.0-volt and 3.3-volt cards are permitted under the SmartMedia specification. A
notch codes the voltage used by each card to prevent you from putting a board in a host
supplying the wrong voltage. A right notch indicates 3.3-volt cards; a left notch, 5.0-volts
cards.
The SmartMedia standards are administered through the SSFDC Forum, which you can
access on the Web at www.ssfdc.or.jp/english/.
Memory Stick
Sony Corporation developed the Memory Stick package as a proprietary technology. Sony
uses Memory Stick as a cross-platform package. The same stick fits digital cameras, digital
video recorders, and MP3 players.
Memory Sticks resemble an old-fashioned stick of chewing gum, about 2 inches long (50
millimeters), 0.85 inches wide (21.45 millimeters), and a bit more than 1/10th inch (2.8
millimeters) thick. For applications demanding even smaller media, Sony offers the
Memory Stick Duo, which measures 1.25 inches (31 millimeters) by 0.75 inches (20 millimeters) and 1/16th inch (1.6 millimeters) thick.
CHAPTER 16 MEMORY
Memory Cards
The plastic package of each Memory Stick comes in one of two colors. The basic Memory
Stick is blue. It is meant for general-purpose storage and holds nothing but memory and
interfacing circuitry. A white Memory Stick has Sonys Magic Gate copy-management
protocol.
Sony calls Memory Stick the de facto standard for digital networks. What that means is
that Memory Stick does not bear the imprimatur of a recognized standards organization
but lots of companies use the same design (under license from Sony). Currently five manufacturers offer Memory Stick products: Apacer Technology, Hagiwara Sys-Com Co.
Ltd., I-O Data Device, Lexar Media, SanDisk, and Sony.
At one time, the Memory Stick format was more expensive than competing card formats.
Now, Sony has licensed other manufacturers to make and sell Memory Sticks, and the
price is competitive with other packages. Two Web sites are meant to provide information
about Memory Sticks. The general public can access www.memorystick.com. Memory Stick
licensees can access www.memorystick.org.
Multimedia Cards
Almost as a response to SmartMedias claims to be the smallest, SanDisk Corporation and
Siemens AG/Infineon Technologies AG developed the MultiMedia Card, which is less
than half the size of the SmartMedia card (although twice as thick). According to its promoters, a MultiMedia Card is about the size of a postage stamp. Thats true in a way
(some commemorative postage stamps are nearly the size of envelopes). The MultiMedia
Card (MMC) actually measures a little more than an inch square, at 24 by 37 mm and 1.4
mm thick (big for a stamp but small for a memory and exchange medium).
Designed specifically for memory, MultiMedia Cards were originally designed with a
64MB maximum capacity. Cards come in two types: ROM-based for prerecorded music
(you cannot record on ROM), and Flash memory for recording your own music and photos.
The one-piece molded plastic package has seven gold pins on one edge to provide a serial
interface.
To promote MultiMedia Card technology, 14 companies founded the MultiMedia Card
Association in 1998. It has since grown to host over 100 members. The MultiMedia Card
Association maintains a Web site at www.mmca.org.
Secure Digital
Derived from the MultiMedia Card, you can consider the Secure Digital card to be MMC
with copy protection. The Secure Digital design is an obvious rethinking of the
MultiMedia Card concept.
579
580
The Secure Digital card uses a nine-pin serial interface, similar to that of the MultiMedia
Card. In fact, the two extra pins are grafted on to the layout of the older card design, with
a small pin 8 crammed between pin 7 and the edge of the card. Pin 9 resides at the opposite side, adjacent to pin 1 but in the notch area.
The initial design of the Secure Digital card was thicker than the MultiMedia Card to
allow the most complex circuitry required by its serial copy management electronics. The
Secure Digital specifications also describe a thin card, one with the same 1.4 mm thickness of the MMC.
The Secure Digital card also incorporates a write-protect tab. By sliding the tab up the
side of the card, you can prevent compliant electronic gear from disturbing the contents
of the card. The specification also allows for a read-only card without a sliding tab.
The current designs of the Secure Digital card accommodates capacities from 32MB to
256MB.
Secure Digital Association
719 San Benito Street, Suite C
Hollister, CA 95023
Phone: 831-636-7322 (SDCA)
Fax: 831-636-2522
Email: [email protected]
xD Picture Card
The latest player in the memory card game is the xD Picture Card, announced July 30,
2002. Jointly developed by Olympus and Fuji Photo Film Company, this card is about the
size of a postage stamp, as described by news releases from each company. In fact, the xD
Picture Card is actually the smallest of the memory card formats. The standard card
dimensions are 20 by 25 mm and 1.7 mm thick.
According to the developers, the name is taken from the phrase Extreme Digital.
Initially, Toshiba will manufacture the cards.
The card design allows for capacities up to 8GB per card, but initial real-world implementations store only 16 to 128MB.
Although only Fuji and Olympus cameras are initially slated to use the xD Picture Card,
with an adapter it can also be used in place of PC Cards. A card reader with a USB interface was also introduced at the same time as the first cards, and a Compact Flash adapter
is under development.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
17
Magnetic Storage
If youve ever seen a punch card, much less had to work with them, you
know why computer workers joyously sang the praises of magnetic storage. When all the world and computers were young, the long-term
memory of the best computer systems was entrusted to 3.58.5-inch
sheets of cardboard, punched with rectangular holes, each one storing a
single code symbol, a whole card holding about a sentence. A program
took a file drawer filled with them. Cardiac arrest required only dropping the drawer, because the order in which the cards were arranged was
crucial.
Magnetic storage is faster, more compact, and more permanent.
Although magnetic media appear vulnerable (especially if youve had or
heard of a disk crash), magnetic records are among the longest-lived of
any storage method. In fact, magnetic records store the itinerary of the
drifting continents, preserving the details of their shifts in data back at
least 250 million years.
In your computer, three different storage systems use magnetism as their
working medium, none of them quite as permanent as hot magma but
with the potential to outlast you. The most important is your hard disk,
without which you simply could not run modern software. The other
two magnetic systems are incidental, at least these days. You could live
without them, and an increasing number of people do. The teetering
twosome includes the floppy disk, as much a vestige of our past as is continental drift (even though most computers still come with a floppy disk
drive), and tape, which has declined in importance as people put more
faith in the integrity of their hard disk systems.
584
Hard Disks
People who think they know such things often liken a computers hard disk to a filing cabinet. Such folk really dont know much about hard disksor the way real people store
information. A hard disk is more like the desk and office of a writer, who in his few lucid
moments between delirium tremens and alcoholic stupor crafts work of unparalleled
insight and beauty from the scribbled notes, torn-out newspaper stories, yellowing magazines, and broken-back books stacked everywhere in piles that defy both organizational
sense and the law of gravity. (Think Poe with the benefit of modern chemistry.)
Your computers hard disk only looks organized, an illusion as compelling as your own
feelings of prowess and courage that arise after youve drunk a meal with your friend, the
alcoholic writer. In reality, its much more like the pack rats stash the writer calls an office,
with uncountable snippets of thoughts, beat-up old photographs, and little bright shiny
things all heaped together, some worth keeping, some that should have been tossed long
ago, and a trash bag full of things that were thrown away but havent yet been carried out.
Your hard disk holds everything you work with on your computer as well as a surprising
amount of stuff that youll never use again. It is your computers long-term memory, but on
the disk, the information is confounded into pieces much like a messy desk or your own
mind. Unlike your mind or mess, however, the disk has an organizing aid, the operating
system. All the neatly arranged files you store on your disk owe their existence to the operating system. Dig deep into your hard disk and youll see its nothing but a receptacleor a
giant collection of receptaclesfor storing blocks of information. The disks job is simply to
put the information into the receptacles, preserve it, and retrieve it in the off-chance that
you need it again. As with everything else, the hard work of the hard disk is in the details.
The hard disk is actually a combination device, a chimera thats part electronic and part
mechanical. Electrically, the hard disk performs the noble function of turning evanescent
pulses of electronic digital data into more permanent magnetic fields. As with other magnetic recording devicesfrom cassette recorders to floppy disksthe hard disk accomplishes its end using an electromagnet, its read/write head, to align the polarities of
magnetic particles on the hard disk itself. Other electronics in the hard disk system control the mechanical half of the drive and help it properly arrange the magnetic storage and
locate the information that is stored on the disk.
History
Because of their ability to give nearly random access to data, magnetic disk drives have
been part of computing since long before there were personal computers. The first drives
suffered from the demands of data processing, however, and quickly wore out. Their
heads ground against their disks, leaving dust where data had been. For fast access, some
lined dozens of heads along the radius of the disk, each sweeping its own dedicated range
CHAPTER 17 MAGNETIC STORAGE
Hard Disks
of disk and data. Such designs had fast access speeds, dependent only on the speed of the
spin of the disk (which is still an issue, even today), and minimal maintenance worries
because they had a minimum of moving parts. But the size of the heads and the cost of
arraying a raft of them meant such drives were inevitably expensive. Though not a major
problem with mainframe computers priced in the millions, pricing a computer with such a
drive would put the computer within the budgets solely of those with personal Space
Shuttles in their garages.
The breakthrough came at IBMs Hursley Labs near Winchester in England. Researchers
there put a single head to work scanning across the disk to get at every square inch
(England had not yet gone metric) of its surface. Their breakthrough, however, totally
eliminated the wear of head against disk and was destined to set the standard for computer
storage for more than three decades. By floatingactually flyingthe read/write head on
a cushion of air, the head never touched the disk and never had a chance to wear it down.
Moreover, the essentially friction-free design allowed the head to move rapidly between
positions above the disk.
This original design had two sections: a fixed drive that kept its disk permanently inside
and a removable section that could be dismounted for file exchange or archiving. Each
held 30MB on a platter about 14 inches across. During development, designers called the
drive a 30/30 to reflect its two storage sections. In that Remington used the same designation for its most famous repeating riflethe gun that won the Westthis kind of drive
became known as a Winchester disk drive.
The name Winchester first referred to the specific drive model. Eventually it was generalized to any hard disk. In the computer industry, however, the term was reserved for drives
that used that same head design as the original Winchester. New disk drivesincluding
all those now in computersdo not use the Winchester head design.
Besides Winchester, you may also hear other outdated terms for what we today call a hard
disk. Many folks at IBM still refer to them as fixed disks. When computer people really
want to confound you, they sometimes use another IBM term from the dark ages of computing, DASD, which stands for Direct Access Storage Device. No matter the name, however, today all hard disks are essentially the same in principle, technology, and operation.
Mechanism
The mechanism of the typical hard disk is actually rather simple, comprising fewer moving
parts than such exotic devices as the electric razor and pencil sharpener. The basic elements
of the system include a stack of one or more plattersthe actual hard disks themselves.
Each of these platters serves as a substrate upon which is laid a magnetic medium in which
data can be recorded. Together the platters rotate as a unit on a shaft, called the spindle.
Typically the shaft connects directly to a spindle motor that spins the entire assembly.
585
586
Rotation
Hard disks almost invariably spin at a single, constant rate measured in revolutions per
minute (RPM). This speed does not change while the disk is in operation, although some
disks may stop to conserve power. Storing information using this constant spin is technically termed constant angular velocity recording. This technology sets the speed of the disks
spin at a constant rate so that, in any given period over any given track, the drives
read/write head hangs over the same-length arc (measured in degrees) of the disk. The
actual length of the arc, measured linearly (in inches or centimeters) varies depending on
the radial position of the head. Although the tiny arc made by each recorded bit has the
same length when measured angularly (that is, in degrees), when the head is farther from
the center of the disk, the bit-arcs are longer when measured linearly (that is, in inches or
millimeters). Despite, or because of, the greater length of each bit toward the outer edge
of the disk, each spin stores the same number of bits and the same amount of information.
Each spin at the outer edge of the disk stores exactly the same number of bits as those at
the inner edge.
Constant angular velocity equipment is easy to build because the disk spins at a constant
number of RPM. Old vinyl phonograph records are the best example of constant angular
velocity recordingthe black platters spun at an invariant 33 1/3, 45, or 78 RPM. Nearly
all hard disks and all ISO standard magneto-optical drives use constant angular velocity
recording.
A more efficient technology, called constant linear velocity recording, alters the spin speed of
the disk depending on how near the center tracks the read/write head lies, so that in any
given period, the same length of track passes below the head. When the head is near the
outer edge of the disk, where the circumference is greater, the slower spin allows more
bits and data to be packed into each spin. Using this technology, a given-size disk can hold
more information.
Constant linear velocity recording is ill-suited to hard disks. For the disk platter to be
properly read or written, it must be spinning at the proper rate. Hard disk heads regularly
bounce from the outer tracks to the inner tracks as your software request them to read or
write data. Slowing or speeding up the platter to the proper speed would require a lengthy
wait, perhaps seconds because of inertia, which would shoot the average access time of the
drive through the roof. For this reason, constant linear velocity recording is used for highcapacity media that dont depend so much on quick random access. The most familiar is
the Compact Disc, which sacrifices instant access for sufficient space to store your favorite
symphony.
Figure 17.1 illustrates the on-disk difference between the two methods of recording. The
sector length varies in constant angular velocity but remains constant using constant linear
velocity. The number of sectors is the same for each track in constant angular velocity
recording but varies with constant linear velocity recording.
Hard Disks
FIGURE 17.1
Comparison of constant angular and line
velocity recording
methods.
Constant Angular Velocity
Constant Linear Velocity
Modern hard disks compromise between constant angular velocity and constant linear
velocity recording. Although they maintain a constant rotation rate, they alter the timing
of individual bits depending on how far from the center of the disk they are written. By
shortening the duration of the bits (measured in microseconds) over longer tracks, the
drive can maintain a constant linear length (again, measured in inches or whatever) for
each bit. This compromise technique underlies multiple zone recording technology, which
we will more fully discuss later.
Speed
The first disk drives (back in the era of the original IBM Winchester) used synchronous
motors. That is, the motor was designed to lock its rotation rate to the frequency of the
AC power line supplying the disk drive. As a result, most motors of early hard disk drives
spun the disk at the same rate as the power line frequency, 3600 revolutions per minute,
which equals the 60 cycles per second of commercial power in the United States.
Synchronous motors are typically big, heavy, and expensive. They also run on normal line
voltage117 volts ACwhich is not desirable to have floating around inside computer
equipment where a couple of errant volts can cause a system crash. As hard disks were
miniaturized, disk-makers adopted a new technologythe servo-controlled DC motor
that eliminated these problems. A servo-controlled motor uses feedback to maintain a
constant and accurate rotation rate. That is, a sensor in the disk drive constantly monitors
how fast the drive spins and adjusts the spin rate should the disk vary from its design specifications.
Because servo motor technology does not depend on the power-line frequency, manufacturers are free to use any rotation rate they want for drives that use it. Early hard disks
with servo motors stuck with the standard 3600 RPM spin to match their signal interfaces
designed around that rotation rate. Once interface standards shifted from the device level
to the system level, however, matching rotation speed to data rate became irrelevant. With
system-level interfaces, the raw data is already separated, deserialized, and buffered on the
587
588
drive itself. The data speeds inside the drive are entirely independent from those outside.
With this design, engineers have a strong incentive for increasing the spin rate of the disk
platter: The faster the drive rotates, the shorter the time that passes between the scan of
any two points on the surface of the disk. A faster spinning platter makes a faster responding drive and one that can transfer information more quickly. With the design freedom
afforded by modern disk interfaces, disk designers can choose any spin speed without worrying about signal compatibility. As a result, the highest performing hard disks have spin
rates substantially higher than the old standardsome rotate as quickly as 10,000 or
15,000 RPM.
Note that disk rotation speed cannot be increased indefinitely. Centrifugal force tends to
tear apart anything that spins at high rates, and hard disks are no exception. Disk designers must balance achieving better performance with the self-destructive tendencies of
rapidly spinning mechanisms. Moreover, overhead in computer disk systems tends to
overwhelm the speed increases won by quickening disk spin. Raising speed results in
diminishing returns.
Today, most of the spins of hard disks fit a three-tier hierarchy. The slowest, turning 4200
RPM, are used only in notebook computers. Disks running at 5400 RPM are general-purpose consumer drives, although a few newer high-performance drives for notebook
machines now reach this speed. Faster drives are used in network servers and high-performance workstations. This high-performance category itself has three speed levels: 7500,
10,000, and 15,000 RPM.
Latency
Despite the quick and constant rotation rate of a hard disk, it cannot deliver information
instantly on request. Theres always a slight delay thats called latency. This term describes
how long after a command to read from or write to a hard disk the disk rotates to the
proper angular position to locate the specific data needed. For example, if a program
requests a byte from a hard disk and that byte has just passed under the read/write head,
the disk must spin one full turn before that byte can be read from the disk and sent to the
program. If read and write requests occur at essentially random times in regard to the spin
of the disk (as they do), on the average the disk has to make half a spin before the
read/write head is properly positioned to read or write the required data. Normal latency
at 3600 RPM means that the quickest you can expect your hard diskon the averageto
find the information you want is 8.33 milliseconds. For a computer that operates with
nanosecond timing, thats a long wait, indeed.
The newer hard disks with higher spin speeds cut latency. The relationship between rotation and latency is linear, so each percentage increase in spin pushes down latency by the
same factor. A modern drive with a 5400 RPM spin achieves a latency of 5.56 milliseconds. Table 17.1 lists the latency of disks based on rotation rate.
Hard Disks
TA B L E 1 7 . 1
Average Latency at Common Drive Rotation Rates
Rotation Rate
Average Latency
Typical Application
3600 RPM
8.33 ms
Obsolete
4500 RPM
6.67 ms
Notebook computers
5400 RPM
5.56 ms
Consumer computers
7200 RPM
4.17 ms
Servers and workstations
10,000 RPM
3.00 ms
Servers
15,000 RPM
2.00 ms
High-performance servers
Standby Mode
During operation, the platters in a hard disk are constantly spinning because starting and
stopping even the small mass of a two-inch drive causes an unacceptable delay in retrieving or archiving your data. This constant spin ensures that your data will be accessible
within the milliseconds of the latency period.
In some applications, particularly notebook computers, the constantly spinning hard disk
takes a toll. Keeping the disk rotating means constant consumption of power by the spindle motor, which means shorter battery life. Consequently, most hard disks are designed
to be able to cease spinning when they are not needed. Typically, the support electronics
in the host computer determine when the disk should stop spinning. Current versions of
Windows make this feature optional (youll find the controls in the Power Option section
of Control Panel). When the feature is activated (as it is by default in most notebook
computers), it means that if you dont access the hard disk for a while, the computer
assumes youve fallen asleep, died, or had your body occupied by aliens and wont be
needing to use the disk for some time. When you do send out a command to read or write
the disk, you then will have to wait while it spins back up to speedpossibly as long as
several seconds. Subsequent accesses then occur at high hard disk speeds until the drive
thinks youve died again and shuts itself down.
The powering down of the drive increases the latency from milliseconds to seconds. It can
be a big penalty. Consequently, most notebook computers allow you to adjust the standby
delay. The longer the delay, the more likely your drive will be spinning when you want to
access itand the quicker your computers battery will discharge. If you work within one
application, a short delay can keep your computer running longer on battery power. If you
shift between applications when using Windows or save your work often, you might as well
specify a long delay because your disk will be spinning most of the time, anyway. Note, too,
that programs with auto-saving defeat the purpose of your hard disks standby mode, particularly when you set the auto-save delay to a short period. For optimum battery life,
youll want to switch off auto-savingif you have sufficient faith in your computer.
589
590
Data-Transfer Rate
The speed of the spin of a hard disk also influences how quickly data can be continuously
read from a drive. At a given storage density (which disk designers try to make as high as
possible to pack as much information in as small a package as possible), the quicker a disk
spins, and the faster information can be read from it. As spin rates increase, more bits on
the surface of the disk pass beneath the read/write head in a given period. This increase
directly translates into a faster flow of datamore bits per second.
The speed at which information is moved from the disk to its control electronics (or its
computer host) is termed the data-transfer rate of the drive. Data-transfer rate is measured
in megabits per second, megahertz (typically these two take the same numeric value), or
megabytes per second (one-eighth the megabit per second rate). Higher is better.
The data-transfer rates quoted for most hard disks are computed values rather than the
speeds you should expect in using a hard disk drive in the real world. A number of factors
drive down the actual rate at which information can be transferred from a disk drive.
The measure of the actual amount of useful information that moves between a disk drive
and your computer is called the throughput. It is always lowersubstantially lowerthan
the disks data-transfer rate. The actual throughput achieved by a drive system varies with
where the measurement is made because each step along the way imposes overhead. The
throughput between your drive and controller is higher than between drive and memory.
And the actual throughput to your programswhich must be managed by your operating
systemis slower still. Throughput to your operating system on the order of a few hundred kilobytes per second is not unusual for hard disk drives that have quoted transfer
rates in excess of 10 or 20 megabytes per second.
Read/Write Heads
Besides the platters, the only other moving part in most hard disk drives is the head system. In nearly all drives, one read/write head is associated with each side of each platter
and flies just above or below its surface. Each of these read/write heads is flexibly connected to a more rigid arm, which supports the flying assembly. Usually several of these
arms are linked together to form a single moving (usually pivoting) unit.
Physical Design
The head is loosely connected to the actuator so that it can minutely rise or fall. When
the hard disk drive is turned off or in sleep mode so that its platters are not spinning, the
head rests lightly against them by a slight spring force. The physical design of the head
makes it into an airfoil much like an airplane wing. As the platters spin, they drag the air
in contact with them along for the ride. The moving air creates a slight breeze, which, like
the air whisking past the airplane wing, generates lift on the hard disk heads airfoil. The
head rises, flying a few millionths of an inch above the spinning surface of the platter.
Hard Disks
The height at which the read/write head of a hard disk flies is one factor in determining
the ultimate storage capacity of the drive. Magnetic fields spread out with distance, so the
farther the head is from the disk, the larger the apparent size of the field thats generated
by a flux transition on the disk. Moving the head closer shrinks the apparent size of the
flux transitions, allowing them to be packed closely together on the disk surface and
increasing the capacity of the disk. The typical first-generation hard disk head flew about
10 to 12 micro-inches (millionths of an inch) above the surface of the platter. Modern disk
drive heads fly closer, five or fewer micro-inches. These lower heights are possible thanks
to smoother platters and smooth thin-film media.
Electrical Design
Flying is a means to an end, not the ultimate purpose of the read/write head. The real job
of the head is to create or detect the magnetic pulses on the disk platter that correspond
to the data you store there. Modern hard disks use one of two basic designs to accomplish
this design purpose: inductive or magneto-resistive.
Inductive Read/Write Heads
An inductive read/write head is nothing more than a miniature electromagnet akin those of
childhood experimentation. Wrap a long length of wire around and around a nail, connect
the two ends of the wire to the positive and negative terminals of a battery, and the nail
becomes a magnet. The electricity flowing through the wire induces a magnetic field in the
nail. In the inductive read/write head, the wire is called the coil and the part that acts as the
nail is the core. The principle is the same. The disk drive electronics send a current through
the read/write head coil, which induces a magnetic field in the core. The magnetic field
alters the orientation of the magnetic particles on the nearby platter. The read process simply reverses the relationship. The magnetic fields of the particles on the platter slightly
magnetize the core, which, in turn, induces a small voltage in the coil. The disk drive electronics detect the small voltage fluctuations in the coil and interpret them as data.
The physical design of the core allows the focusing of the heads read and write capabilities into a small area. Instead of a long, thin nail, the core is folded so that its two poles
(ends) are not quite touching, separated by a thin gap. This design concentrates the magnetic field into the tiny gap. The first practical read/write heads, those in vintage tape
recorders, had nothing more than air in the gap. Basic disk drive read/write heads fill the
gap with a nonmagnetic metal. Such designs are termed metal-in-gap heads. Modern
read/write heads replace the coil of wire with a thin layer of copper deposited in coil form
as a film. Called thin-film heads, their technology allows for finer, lower-mass coils that are
easier and less expensive to fabricate.
Magneto-Resistive Heads
The latest trend in head design is magneto-resistive read/write heads. These heads work on
an entirely different physical principle from inductive heads. They measure the change in
591
592
electrical resistance that a magnetic field causes in some materials. The disk drive electronics send a small, constant current through the magneto-resistive material (usually an
alloy of iron and nickel) and measure the change in voltage across the headas the resistance of the head goes up, the voltage goes down. The change is minuscule but easily
detectable by modern precision electronics.
The magneto-resistive action is one-way. It can be used only to detect changes in magnetic fields. It cannot create the fields. In other words, the magneto-resistive principle
works only for read operations. Consequently, disk drives with magneto-resistive heads
actually have combination headsa magneto-resistive read head combined with an inductive write head. Dividing the functions of a read/write head into separate elements allows
each to be tailored to best operation. The magneto-resistive design allows for higher frequency operation, which equates to greater storage densities and operating speeds.
Write Precompensation
Constant angular velocity recording has another drawback: The shorter sectors closer to
the spindle require data to be packed into them more tightly, squeezing the magnetic flux
reversals in the recording medium ever closer together. The ability of many magnetic
media to hold flux transitions falls off as the transitions are packed more tightlypinched
together, they produce a feebler field and induce a lower current in the read/write head.
One way of dealing with this problem is to write on the disk with a stronger magnetic
field as the sectors get closer to the spindle. By increasing the current in the read/write
head when it writes nearer the center of the disk, the on-disk flux transitions can be made
stronger. They can then induce stronger currents in the read/write head when that area of
the disk is read.
This process is called write precompensation because the increased writing current compensates for the fall off in disk responses nearer its center at a place logically before the information is stored on the disk. The electronics of modern hard disk drives automatically
make the necessary compensation.
Partial Response Maximum Likelihood
Although group-coding techniques have served well through the evolution of the computer hard disk, another technique called Partial Response Maximum Likelihood (PRML)
technology works in the opposite direction. Instead of modifying the dataessentially
expanding it to make it work better with existing hardwarePRML modifies the read
electronics of the disk drive so they can better sort through densely recorded data. IBM
first introduced PRML electronics in 1990, and the technology found its way into computer hard disk drives a few years later.
PRML works only during reading the disk. Its specific goal is to compensate for intersymbol interference, a kind of distortion that appears when a drive packs data densely. As the
Hard Disks
read/write head scans the disk, it produces an analog signal. Conventional disk read electronics detect the peaks of the analog pulses and translate them into digital bits. At high
bit-rates, which occur when flux transitions are packed densely and the disk spins rapidly,
the peaks can blur together. The electronics can readily mistake two bits for one or make
similar errors. PRML read electronics can better sort through the analog signals and more
reliably translate the signals into data.
The first step in the PRML process is to filter the signal from the read/write head using
digital techniques, shaping it with a partial response characteristic by altering its frequency
response and timing characteristics. Using digital processing, the PRML system then
detects where signals from flux transitions are most likely to occur. The PRML system
uses a sequence-detection algorithm that accurately sorts through the data.
The chief advantage of PRML is that it ensures data integrity with high bit densities and
faster data rates between head and electronics. PRML does not require special coding
during recording. In fact, one of its advantages is that it sidesteps the increase in bit-count
that arises with group-coding techniques. Because fewer bits must be written to disk,
PRML allows more data to fit on a given disk. Because PRML allows higher disk densities, it can increase the read rate of a disk without altering its mechanical rate of rotation.
Head Actuators
Each read/write head scans the hard disk for information. Were the head nothing more
than that, fixed in position as is the head of a tape recorder, it would only be able to read a
narrow section of the disk. The head and the entire assembly to which it is attached must
be able to move in order to take advantage of all the recordable area on the hard disk.
The mechanism that moves the head assembly is called the head actuator. Usually the head
assembly is pivoted and is swung across the disk by a special head actuator solenoid or
motor.
The first head actuators were open loop. That is, the actuator moved the head to a preset
position and hoped it was right. Modern actuators are closed loop. They move the head,
check its position over the disk (by reading magnetically coded identification from the
disk), and readjust the position until they get it right. Closed-loop actuators are more
accurate and quicker.
The closed-loop system gets a constant stream of information regarding the head position
from the disk, so it always knows exactly where the head is. The system determines the
location of the head by constantly reading from a special, dedicated side of one platter
the servo surfacethat stores a special magnetic pattern that allows the drive mechanism
to identify each storage location on the disk. Some more recent magnetic hard disks put
the servo information on the same recording surface as the stored data. This combined
data-and-servo system is called embedded servo technology.
593
594
The most common of the closed-loop actuator systems uses a voice coil mechanism that
operates like the voice coil in a loudspeaker and is therefore called a servo-voice coil actuator. In
this design, a magnetic field is generated in a coil of wire wrapped around part of the actuator
(making it a solenoid) by the controlling electronics, and this field pulls the head mechanism
against the force of a spring. By varying the current in the coil, the head mechanism is drawn
farther from its anchoring spring, and the head moves across the disk. The voice coil mechanism connects directly to a pivoting arm, which also supports the read/write head above the
platter. The varying force of the voice coil swings the head in an arc across the platter surface.
Landing Zone
Hard disks are most vulnerable to head crash damage when they are turned off. As soon as
you flick the off switch on your computer, the platters of its hard disk must stop spinning,
and the airflow that keeps the heads flying stops. Generally, the airflow decreases gradually, and the head slowly progresses downward, eventually landing like an airplane on the
disk media.
In truth, however, any head landing is more of a controlled crash and holds the potential
for disk damage. Consequently, most hard diskseven those with thin-film mediahave a
dedicated landing zone reserved in their media in which no data can be recorded. This
landing zone is usually at the inner edge of the actual data storage area.
Park-and-Lock
Usually a software command is necessary to bring the head to the landing zone and hold
it there while the disk spins down. This process is called head parking. The first hard disks
had no special provisions for parking their heads and required a specific software command to move their heads to the landing zone. All modern hard disks are designed so that
whenever their power is switched off, the head automatically retracts to the landing zone
before the disk spins down. Such drives are said to have automatic head parking. In addition, the most modern drives latch their heads in the landing zone after power is removed.
The latch prevents an impact or other shock to the system from jarring the head out of
the landing zone and, in the process, bouncing it across the vulnerable medium. This feature is generally termed automatic park-and-lock. All drives now incorporate it.
Thermal Compensation
All materials expand and contract as temperatures change, and the metals used in constructing hard disk drives are no exception. As a drive operates, it generates heat from the
motors that spin its platters, the actuator that moves the heads, and the electronics that
control its operation. This heat causes the various components of the drive to expand
slightly, changing its dimensions slightly but measurably. Because of the miniaturization of
modern hard disks that packs thousands of tracks in an inch, even this slight thermal
expansion can alter the geometry of the drive sufficiently that heads and tracks can move
from their expected positions.
Hard Disks
To compensate for such changes, most hard disk drives periodically perform a thermal calibration, or T-cal. The disk moves its heads to read special calibration tracks to reestablish a
proper reference for head positioning. Drive manufacturers developed their own algorithms for determining when their drives would perform thermal calibration (for example,
at fixed intervals or upon the occurrence of seek errors). In general, the thermal compensation takes priority over normal read operations and sometimes imposes a delay when
you request data. The delay can amount to several dozen milliseconds because the drives
heads must move to the calibration tracks before fulfilling any data requests.
To avoid the delays imposed by thermal calibration, many high-performance drives have
the ability to delay the calibration until the completion of a read operation to avoid the
interrupt of the delivery of prolonged sequential data streams, such as those that might
occur in playing back a video clip. Most drives calibrate all heads simultaneously, which
results in the drive being unavailable for reading or writing data for the milliseconds
required by the recalibration. A few drives can now recalibrate heads individually, allowing
the other heads to retrieve data at the same time.
Medium
The disk spinning inside the hard disk drive is central to the drivein more ways than
one. The diameter of this platter determines how physically large a drive mechanism must
be. In fact, most hard disk drives are measured by the size of their platters. When the
computer first burst upon the scene, makers of hard disks were making valiant attempts at
hard disk platter miniaturization, moving from those eight inches in diameter (so-called
eight-inch disks) to 5.25-inch platters. Today the trend is toward ever-smaller platters. Most
large-capacity drives bound for desktop computer systems now use 3.5-inch platters.
Those meant for computers in which weight and size must be minimized (which means,
of course, notebook and smaller computers) have platters measuring 2.5, 1.8, or 1.3 inches
(currently the smallest) in diameter. (See Chapter 32, Cases, for form-factor details.)
To increase storage capacity in conventional magnetic hard disk storage systems, both
sides of a platter are used for storing information, each surface with its own read/write
head. (One head is on the bottom, where it must fly below the platter.) In addition, manufacturers often put several platters on a single spindle, making a taller package with the
same diameter as a single platter. The number of platters inside a hard disk also influences
the speed at which data stored on the hard disk can be found. The more platters a given
disk drive uses, the greater the probability that one of the heads associated with one of
those platters will be above the byte thats being searched for. Consequently, the time to
find information is reduced.
Substrates
The platters of a conventional magnetic hard disk are precisely machined to an extremely
fine tolerance, measured in micro-inches. They have to beremember, the read/write head
595
596
flies just a few micro-inches above each platter. If the disk juts up, the result is akin to a
DC-10 encountering Pikes Peaka crash thats good for neither airplane nor hard disk.
Consequently, disk-makers try to ensure that platters are as flat and smooth as possible.
The most common substrate material is aluminum, which has several virtues: Its easy to
machine to a relatively smooth surface. Its generally inert, so it wont react with the material covering it. Its nonmagnetic, so it wont affect the recording process. It has been used
for a long while (since the first disk drives) and is consequently a familiar material. And
above all, its cheap.
A newer alternative is commonly called the glass platter, although the actual material used
can range from ordinary window glass to advanced ceramic compounds akin to Space
Shuttle skin. Glass platters excel at exactly the same qualities as do aluminum platters,
only more so. They can be smooth and allow read/write heads to fly lower. They are also
less reactive than aluminum and, with the right choice of material, can be lighter.
Areal Density
The smoothness of the substrate affects how tightly information can be packed on the surface of a platter. The term used to describe this characteristic is areal densitythat is, the
amount of data that can be packed onto a given area of the platter surface. The most common unit for measuring areal density is megabits per square inch. The higher the areal
density, the more information that can be stored on a single platter. Smaller hard disks
require greater areal densities to achieve the same capacities as larger units.
Areal density is generally measured in megabytes per square inch of disk surface, and current products achieve values on the order of 500 to 1000 megabits per square inch.
A number of factors influence the areal density that can be achieved by a given hard disk
drive. The key factor is the size of the magnetic domain that encodes each bit of data,
which is controlled in turn by several factors. These include the height at which the
read/write head flies and the particle (grain) size of the medium.
Manufacturers make read/write heads smaller to generate smaller fields and fly them as
closely to the platter as possible without risking the head running into the jagged peaks of
surface roughness. The smoothness of the medium determines the lowest possible flying
heighta head can fly closer to a smoother surface.
The size of magnetic domains on a disk is also limited by the size of the magnetic particles
themselves. A domain cannot be smaller than the particle that stores it. At one time, ball
mills ground a magnetic oxide medium until the particle size was small enough for the
desired application. Platters were coated with a slurry of the resulting magnetic material.
Modern magnetic materials minimize grain size by electroplating the platters.
Hard Disks
Media
The first magnetic medium used in hard disks was made from the same materials used in
conventional audio recording tapesferric or ferrous oxide compounds (essentially fine
grains of rather exotic rust). As with recording tape, the oxide particles are milled in a
mixture of other compounds, including a glue-like binder and often a lubricant. The
binder also serves to isolate individual oxide particles from one another. This mud-like
mixture is then coated onto the platters. But this coating is rather rough, too rough for
todays high-capacity, small-size hard disks. Moreover, it is soft and prone to damage
should the read/write head touch it, resulting in a head crash that may render part of the
disk unusable. As a result, although once an important technology, oxide media have been
abandoned by drive-makers.
In all current hard disk drives, drive-makers have replaced oxide coatings with thin-film
magnetic media. As the name implies, a thin-film disk has a microscopically skinny layer
of a pure metal, or mixture of metals, mechanically bound to its surface. These thin-films
can be applied either by plating the platter, much the same way chrome is applied to automobile bumpers, or by sputtering, a form of vapor-plating in which metal is ejected off a
hot electrode in a vacuum and electrically attracted to the disk platter.
The very thinness of thin-film media allows higher areal densities because the magnetic
field has less thickness in which to spread out. Because the thin-film surface is smoother, it
allows heads to fly closer. Thin-film media also have higher coercivities, which allow
smaller areas to produce the strong magnetic pulses needed for error-free reading of the
data on the disk.
One reason that thin film can be so thin and support high areal densities is that, as with
chrome-plated automobile bumpers and faucets, plated and sputtered media require no
binders to hold their magnetic layers in place. Moreover, as with chrome plating, the thin
films on hard disk platters are genuinely hard, many times tougher than oxide coatings.
That makes them less susceptible to most forms of head crashingthe head merely
bounces off the thin-film platter just as it would your cars bumpers.
Contamination
Contaminants such as dust and air pollution particles stuck to the media surface can cause
problems. With older oxide media drives, contaminants could result in a head crash. With
plated media, contaminants arent as likely to cause damage, but they can interfere with
the proper read/write operation of the drive.
To help guard against contamination of the platter surface with dust, hair, and other floating gunk, most hard disks keep all their vulnerable parts in a protective chamber. In fact,
this need to avoid contamination is why nearly all computer hard disks use nonremovable
media, sealed out of harms way.
597
598
The disk chamber is not completely airtight. Usually a small vent is designed into the system to allow the air pressure inside the disk drive to adjust to changes in environmental
air pressure. Although this air exchange is minimal, a filter in this vent system traps particles before they can enter the drive. Microscopic pollutants, such as corrosive molecules in
the air, can seep through the filter, however, potentially damaging the disk surface.
Although the influx of such pollutants is smallthe hard disk vent does not foster airflow,
only pressure equalizationit is best not to operate a hard disk in a polluted environment.
You wouldnt want to be there to use it, anyhow.
Geometry
The geometry of a hard disk expresses the physical arrangement on the platters inside the
drive. Today it is an issue for engineers only because the logical block addressing used by
all new hard disk drives effectively hides the drive geometry from you and your operating
system. As long as a drive has all the blocks it advertises, they could be laid out like a
Jackson Pollock painting, and your operating system wouldnt know the difference.
If you accidentally enter the netherworld of your computers BIOS setup and jog your
hard disk away from automatic configuration, you likely will be confronted with a disk
parameter table that requests you enter the number of heads, cylinders, and sectors of your
drive. The best strategy is to notch the settings back to Auto. But the drive parameters
put you squarely in geometry territory.
Tracks
No matter the type of magnetic media or style of head actuator used by a disk, the
read/write head must stop its lateral motion across the disk whenever it reads or writes
data. While it is stationary, the platter spins underneath it. Each time the platter completes one spin, the head traces a full circle across its surface. This circle is called a track.
A disk drive stores the data-bits of a given track sequentially, as if it were a strip of tape
spliced end to end. With every spin, the same data passes by the head, as long as the drive
holds in the same place. The electronics of the drive select which portion of the track to
read (or write) to find a random block of data.
Cylinders
Each head traces out a separate track across its associated platter. The head actuator locks
all the heads together so that all are at the same position from the center of the disk along
a given radius. Because the combination of all the tracks traced out at a given head actuator position forms the skeleton of a solid cylinder, such a vertical stack of tracks is often
termed exactly thata cylinder.
The number of cylinders in a drive is the same as the number of tracks on a platter in that
drive. Both numbers are permanently determined when the manufacturer makes the drive.
Hard Disks
In most drives, the number of cylinders is set by a magnetic pattern called a servo pattern.
Older hard disks dedicated one surface of a platter to this servo information. Most modern disks put the servo information on the same surface as the stored data. The servo
information gets read along with the data, and the drive electronics sort everything out
using the servo information to find its place and sending the data to your applications.
This kind of hard disk is called an embedded servo drive.
The more cylinders in the drive, the more data the drive can store. The maximum number of cylinders is limited by physical factors inherent in the technology used by the drive.
More tracks on each platter means the tracks are squeezed closely together, forcing them
to be smaller. The minimum width of a track is set by the size of the head but is limited
by other factorssuch as how closely the head flies to the disk surfacethat also limit the
amount of information that the drive can fit into each track. Once hard disk drives had as
few as 312 cylinders. Modern drives have thousands.
Sectors
Most hard disk systems further divide each track into short arcs termed sectors, and the
sector is the basic storage unit of the drive. Some operating systems use the sector as their
basic storage unit, as does the NTFS system used by Windows NT and Windows 2000,
for example. Under the VFAT system of Windows 95 and Windows 98, however, the
operating system gathers together several sectors to make its basic unit of storage for disk
filesthe cluster.
Sectors can be soft, marked magnetically with bit-patterns embedded in the data on the
track itself, or hard, set by the drive mechanism itself. Soft sectors are demarcated using a
low-level format program, and their number can vary almost arbitrarily depending on the
formatting software and the interface used for connecting the disk. Disks with device-level
interfaces are essentially soft-sectored. For all practical purposes, disks with system-level
interfaces are hard-sectored because their sector size is set by the servo information
encoded on the drive platters, which cannot be changed once the drive leaves the factory.
Magneto-optical cartridges are hard-sectored by an embedded optical format prerecorded
on the medium.
In the computer hard disk industry, the size of a sector is, by convention, almost universally 512 bytes. The number of sectors per track depends on the design of the disk. The
sector count on any given track of older hard disks is the same as every other track
because of their use of constant angular velocity recording. Most modern hard disk drives
use a technique called multiple zone recording (MZR), which puts variable numbers of sectors on each track. MZR allows the drive-maker to use the storage capacity of the magnetic medium more efficiently.
A disk with a fixed number of sectors per track stores data at lower densities in its outer
tracks than it does in its inner tracks. Only the innermost tracks pack data at the highest
599
600
density allowed by the medium. All the other tracks must be recorded at a lower density,
an inevitable result of the constant angular velocity recording used by the hard disk and
the fixed frequency of the data signals.
Multiple zone recording allows the drive to maintain a nearly constant data density across
the disk by dividing it into zones. The drive alters the frequency of the data signals to
match each zone. Using higher frequencies in the zones near the outer tracks of a disk
increases their data density to about that of the inner tracks. This, in turn, can substantially increase overall disk capacity without compromising reliability or altering the constant spin needed for quick access.
Sometimes MZR technology is described as zoned constant angular velocity (ZCAV) recording, a term which confirms that the spin rate remains the same (constant angular velocity)
but the platter is divided into areas with different recording densities (zones). Seagate
Technologies uses a proprietary form of MZR called zone-bit recordingdifferent name,
same effect.
Physical Format
The geometry of a disk drive describes only the numbers of the various drive parameterscylinders, heads, and sectors. The drive format describes the arrangement and alignment of these parameters.
Disk geometry fixes the tracks as concentric circles, with the sectors as small arcs within
each track. The format defines the location of the sectors in regard to one anotherthat
is, the order in which they are read. Sectors need not be read one after another in a given
track. Moreover, their starting edges need to exactly align on the disk.
Neither tracks nor sectors are engraved on the surface of individual platters. They are
instead defined magnetically by coded bit-patterns recorded on the disk. Before data can
be written on such a disk, the sectors have to be marked to serve as guideposts markers so
that the information can later be found and retrieved. The process by which sectors are
defined on the hard disk is called low-level formatting because it occurs at a control level
below the reaches of normal Windows commands.
Three methods have found general application in defining tracks: simply by the count of
the stepper motor in the oldest band-stepper drives, by the permanently recorded track
servo data on the dedicated servo surface of old servo-voice coil drives, and by embedded
servo data in modern drives.
In classic hard disk drives, special bit-patterns on the disk serve as sector-identification
markings. The patterns indicate the start of the sector and encode an ID number that
gives the sector number within the track. The sector ID precedes each sector; error-correction data typically follows each sector. In normal operation, the disk servo system seeks
Hard Disks
a particular track, then the drive begins to read sector IDs until it finds the sector that
your computer has requested.
The sector ID can consume a significant portion of the available space on each disk track,
about 10 percent. Consequently, manufacturers have sought means to eliminate it. For
example, the No-ID Format developed by IBM eliminates sector IDs by putting a format
map in RAM. The map tells the drive where on each track each sector is located and
which sectors have been marked bad. The map, for example, tells the drive how many sectors are on a track in a zoned recording system and where each begins in reference to the
track servo information embedded on the disk. This format also improves access speed
because the drive can immediately locate a given sector without detours in chasing
replacements for defective sectors.
File System
To store a file on disk, the FAT file system breaks it down into a group of clusters, perhaps
hundreds of them. Each cluster can be drawn from anywhere on the disk. Sequential
pieces of a file do not necessarily have to be stored in clusters that are physically adjacent.
The earliestand now obsoleteversions of the FAT file system followed a simple rule in
picking which clusters are assigned to each file. The first available cluster, the one nearest
the beginning of the disk, is always the next one used. Therefore, on a new disk, clusters
are picked one after another, and all the clusters in a file are contiguous.
When a file is erased, its clusters are freed for reuse. These newly freed clusters, being
closer to the beginning of the disk, are the first ones chosen when the next file is written
to disk. In effect, a FAT-based file system first fills in the holes left by the erased file. As a
result, the clusters of new files may be scattered all over the disk.
The earliest versions of the FAT file system used this strange strategy because they were
written at a time when capacity was more important than speed. The goal was to pack
files on the disk as stingily as possible. For more than a decade, however, the FAT system
has used a different strategy. Instead of immediately trying to use the first available cluster
closest to the beginning of the disk, the file system attempts to write on never-before-used
clusters before filling in any erased clusters. This helps ensure that the clusters of a file are
closer to one another, a technique that improves the speed of reading a file from the disk.
File Allocation Table

To keep track of which cluster belongs in which file, the default file system of consumer
Windows (including 95, 98, Me, and XP) uses a file allocation table (FAT), essentially a map
of the clusters on the disk. When you read to a file, the FAT-based file system automatically and invisibly checks the FAT to find all the clusters of the file; when you write to
the disk, it checks the FAT for available clusters. No matter how scattered over your disk
the individual clusters of a file may be, youand your softwareonly see a single file.
601
602
FAT-based file systems simply number all the clusters in a manner similar to the way a
disk drive numbers logical blocks. The operating system keeps track of the cluster numbers and in what order clusters have been assigned to a given file. The operating system
stores most of the cluster data in the file allocation table.
The FAT file system works by chaining together clusters. The directory entry of a file or
subdirectory contains several bytes of data in addition to the files name. Along with the
date the file was last changed and the files attributes is the number of the first cluster used
to store the file or subdirectory.
When the operating system reads a file, it first checks the directory entry to find the first
cluster number. In addition to reading the data from the cluster from the disk, the operating system also checks the file allocation table for the entry with the number corresponding to the first cluster number. This FAT entry indicates the number of the next cluster in
the file. After reading that cluster, the operating system checks the entry corresponding to
that cluster number to find the next cluster. If the file has no additional clusters, the cluster entry has a value of 0FF(hex). The operating system assigns unused clustersthose
available for adding to files to store datathe value of zero.
When the standard FAT-based computer operating system erases a file, it merely changes
the first character of the filename in the directory entry to 0E5(hex) and changes all the
FAT entries of the file to zero. Because the rest of the directory information remains
intact (at least until the file system runs out of space for directory information and overwrites the entries of erased files), it can be recovered to help reconstruct accidentally
erased files. An unerase or undelete utility checks the directory for entries with the first
character of 0E5(hex) and displays what it finds as candidates for recovery. From the
remaining directory data, the unerasing program can locate the first cluster of the file.
Finding the remaining clusters from the FAT is a matter of making educated guesses.
The FAT of a disk is so important that Windows guards against losing its data by putting
two complete (and identical) copies of the FAT end to end on the disk.
Clusters
As clever as using clusters to allocate file data may be, the technique has its drawback. It
can be wasteful. Disk space is divvied up in units of a cluster. No matter how small a file
(or a subdirectory, which is simply a special kind of file) may be, it occupies at minimum
one cluster of disk space. Larger files take up entire clusters, but any fractional cluster of
space thats left over requires another cluster. On average, each file on the disk wastes half
a cluster of space. The more files, the more waste. The larger the clusters, the more
waste. Unless you work exclusively with massive files, increasing cluster size to increase
disk capacity is a technique to avoid whenever possible.
Hard Disks
The first versions of DOS used FATs with 12-bit entries for cluster numbers, which
allowed a total of 4096 uniquely named clusters. Later, Microsoft updated the FAT to use
16-bit entries, and this FAT structure, usually called FAT16, was used through Windows
95. By that time, however, disks had become larger than a 16-bit structure could reasonably handle. Microsoft introduced a new 32-bit FAT with the second edition of Windows
95. The new system, commonly called FAT32, reserves 4 of its 32 bits for future purposes,
so each cluster is actually identified with a 28-bit value. Using FAT32, recent Windows
versions can accommodate drives up to 2048GB and on smaller disks store files more efficiently with smaller clusters (see Table 17.2).
TA B L E 1 7 . 2
Drive Capacity Versus Cluster Size in 32-Bit FAT Systems
Drive Capacity
Cluster Size
0 up to 6GB
4K
6GB up to 16GB
8K
16GB up to 32GB
16K
32GB up to 2048GB
32K
FAT32 is the most recent version, and drivers for it ship with all current Windows versions. If you choose to use a FAT-based file system, Windows will ask if you want large
disk support and automatically install FAT32.
Compression
Microsoft includes a disk-compression system with many Windows versions that the company calls DriveSpace. The file system takes uncompressed data one cluster at a time and
maps it in compressed form into sectors in the compressed volume file. To locate which
sector belongs to each file, the file system uses a special FAT called the Microsoft
DoubleSpace FAT (or MDFAT, with DoubleSpace being the DriveSpace predecessor) that
encodes the first sector used for storing a given cluster, the number of sectors required for
coding the cluster, and the number of the cluster in the uncompressed volume thats
stored in those sectors. When the operating system needs the data from a file, the file system first searches for the clusters in the main disk FAT and then looks up the corresponding starting and length values in the MDFAT. With that information, the operating
system locates the data, uncompresses it, and passes it along to your applications.
To speed up operations when writing compressed data to disk, the operating system uses a
second kind of FAT in the compressed volume file. Called the BitFAT, this structure
reports which sectors reserved in the compressed volume file hold active data and which
are empty. The BitFAT uses only one bit for each sector as a flag to indicate whether a
sector is occupied.
603
604
New Technology File System

Windows NT, Windows 2000, and Windows XP give you two choices for your file system: the same old FAT-based system used since time began and the newer Windows NT
File System, usually termed NTFS.
The centerpiece of the NTFS is the master file table (MFT), which stores all the data
describing each directory and file on a given disk. The basic data about each file is contained in a file record in the master file table. These file records may be two, four, or eight
sectors long (that is, 1KB, 2KB, or 4KB). The first 16 records are reserved for system use
to hold data of special metadata files, the first of which stores the attributes of the master
file table itself.
To NTFS, a file is a collection of attributes, each of which describes some aspect of the
file. One of the attributes is the name of the file, another is the data contained in the file.
Others may include who worked on the file and when it was last modified. The master file
table tracks these attributes. To identify attributes, the file system assigns each file a
unique ID number, a 48-bit value (allowing for nearly 300 trillion entries).
Instead of the clusters used by the FAT system, NTFS uses the sector as its basic unit of
storage. Sectors on a disk or partition are identified by relative sector numbers, each of
which is 32 bits longsufficient to encode 4,294,967,296 sectors or a total disk space of
2048GB. Sectors are numbered sequentially, starting with the first one in the partition.
Files are allocated in multiples of single sectors; directories, however, are made from one
or more blocks of four sectors.
Each file or directory on the disk is identified by its File NODE, which stores descriptive
data about the file or directory. This information includes file attributes, creation date,
modification dates, access dates, sizes, and a pointer that indicates in which sector the data
in the file is stored. Each File NODE is one sector (512 bytes) long. Up to 254 bytes of
the File NODE of a disk file store an extended filename, which can include upper- and
lowercase characters, some punctuation (for example, periods), and spaces.
An NTFS disk organizes its storage from a root directory. In an NTFS system, however,
the root directory does not have a fixed location or size. Instead, the root directory is
identified by reference to the disk super block, which is a special sector that is always kept
as the 16th sector from the beginning of the HPFS partition. The 12th and 13th bytes
that is, at an offset of 0C(hex) from the start of the blockof the super block point to the
location of the root directory File NODE. Free space on the disk is identified by a
bitmapped table.
As with other FNODEs, a pointer in the root directory FNODE stores the location of the
first block of four sectors assigned to the root directory. The root directory is identical to
the other directories in the HPFS hierarchy, and like them it can expand or shrink as the
number of files it contains changes. If the root directory needs to expand beyond its initial
Hard Disks
four sectors, it splits into a tree-like structure. The File NODE of the root directory then
points to the base File NODE of the tree, and each pointer in the tree points to one directory entry and possibly a pointer to another directory node that may in turn point to
entries whose names are sorted before the pointer entry. This structure provides a quick
path for finding a particular entry, along with a simple method of scanning all entries.
NTFS can accommodate any length file (that will fit in the partition, of course) by assigning multiple sectors to it. These sectors need not be contiguous. NTFS, however, preallocates sectors to a file at the time it is opened, so a file may be assigned sectors that do not
contain active data. The File NODE of the file maintains an accurate total of the sectors
that are actually used for storing information. This preallocation scheme helps prevent
files from becoming fragmented. Normally, the block of sectors assigned to a file will be
contiguous, and the file will not become fragmented until all the contiguous sectors have
been used up.
Two types of sectors are used to track the sectors assigned a given file. For files that have
few fragments, the File NODE maintains a list of all the relative sector numbers of the first
sector in a block of sectors used by the file, as well as the total number of sectors in the
file before those of each block. To capture all the data in a file, the operating system finds
the relative sector number of the first block of sectors used by the file and the total number of sectors in the block. It then checks the next relative sector number and keeps
counting with a running total of sectors in the file.
If a file has many fragments, it uses a tree-style table of pointers to indicate the location of
each block of sectors. The entry in the files File NODE table then stores pointers to the
sectors, which themselves store pointers to the data. Each of these sectors identifies itself
with a special flag, indicating whether it points to data or to more pointers.
Besides its huge capacity, NTFS has significant advantages when dealing with large hierarchies of directories, directories containing large number of files, and large files.
Although both the NTFS and FAT use tree-structured directory systems, the directories
in the NTFS are not arranged like a tree. Each directory gets stored in a tree-like structure that, coupled with the presorting of entries automatically performed by the NTFS,
allows for faster searches of large directories. NTFS also arranges directories on the disk
to reduce the time required to access theminstead of starting at the edge of the disk,
they fan out from the center.
The master file table attempts to store all the attributes of a file in the record it associates
with that file. When the attributes of a file grow too large to be held in the MFT record,
the NTFS just spreads the attribute data across as many additional disk clusters records
to create as many nonresident attributes as are needed to hold the file. The master file table
keeps track of all the records containing the attributes associated with a given file by the
files ID number.
605
606
This system allows any file to grow as large as the complete storage space available while
preserving a small allocation unit size. No matter how large a disk or partition is, NTFS
never allocates space in increments larger than 4KB.
Performance Issues
When shopping for hard disks, many people become preoccupied with disk performance.
They believe that some drives find and transfer information faster than others. Theyre
right. But the differences between state-of-the-art hard disk drives are much smaller than
they used to be, and in a properly set-up system, the remaining differences can be almost
completely equalized.
The performance of a hard disk is directly related to design choices in making the mechanism. The head actuator has the greatest effect on the speed at which data can be
retrieved from the disk, with the number of platters exerting a smaller effect. Because the
head actuator designs used by hard diskmakers have converged, as have the number of
platters per drive because of height restrictions of modern form factors, the performance
of various products has also converged.
Clearly, however, all hard disks dont deliver the same performance. The differences are
particularly obvious when you compare a drive thats a few years old with a current product. Understanding the issues involved in hard disk performance will help you better
appreciate the strides made by the industry in the last few years and show you what
improvements may still lie ahead.
Average Access Time

Youve already encountered the term latency, which indicates the average delay in finding a
given bit of data imposed because of the spin of the disk. Another factor also influences
how long elapses between the moment the disk drive receives a request to reveal whats
stored at a given place on the disk and when the drive is actually ready to read or write at
that placethe speed at which the read/write head can move radially from one cylinder to
another. This speed is expressed in a number of ways, often as a seek time. Track-to-track
seek time indicates the period required to move the head from one track to the next. More
important, however, is the average access time (sometimes rendered as average seek time),
which specifies how long it takes the read/write head to move on the average to any cylinder (or radial position). Lower average access times, expressed in milliseconds, are better.
The type of head actuator technology, the mass of the actuator assembly, the physical
power of the actuator itself, and the width of the data area on the disk all influence average access time. Smaller drives have some inherent advantages in minimizing average
access time. Their smaller, lighter head and actuators have less inertia and can accelerate
and settle down faster. More closely spaced tracks mean the head needs to travel a shorter
distance in skipping between them when seeking data.
Hard Disks
Real-world access times vary by more than a factor of ten. The first computer-size hard
disks had access times hardly better than floppy disks, sometimes as long as 150 milliseconds. The newest drives are usually below ten milliseconds; some are closer to six
milliseconds.
How low an average access time you need depends mostly on your impatience. Quicker is
always better and typically more costly. You can feel the difference between a slow and fast
drive when you use your computer, particularly when you dont have sufficient memory to
hold all the applications you run simultaneously. Once access time is below about ten
microseconds, however, you may be hard pressed to pay the price of improvement.
Disk-makers have explored all sorts of exotic technologies to reduce access time. Some
primeval hard disks had a dozen or more fixed heads scanning huge disks. Because the
heads didnt move, the access time was close to zeromore correctly, half the latency of
the drive. About a decade ago, drive-makers experimented with dual-actuator drivestwo
heads mean less than half the waiting because with an intelligent controller the drive
could overlap read and write requests. None of these technologies made it into the computer mainstream because an even better ideaone much simpler and cheaperhas taken
the forefront: the disk array (discussed later). Instead of multiple actuators in a single
drive, arrays spread the multiple actuators among several drives.
Advanced disk controllers, particularly those used in disk arrays, are able to minimize the
delays caused by head seeks using a technique called elevator seeking. When confronted
with several read or write requests for different disk tracks, the controller organizes the
requests in the way that moves the head the least between seeks. Like an elevator, it
courses through the seek requests from the lower-numbered tracks to the higher-numbered tracks and then goes back on the next requests, first taking care of the higher-numbered tracks and working its way back to the lower-numbered tracks. The data gathered
for each individual request is stored in the controller and doled out at the proper time.
Data-Transfer Rate
Once a byte or record is found on the disk, it must be transferred to the host computer.
Another disk system specificationthe data-transfer ratereflects how fast bytes are batted back and forth, effecting how quickly information can shuttle between microprocessor
and hard disk. The transfer rate of a disk is controlled by a number of design factors completely separate from those of the average access time.
The transfer rate of a hard disk is expressed in megahertz (MHz) or megabytes per second
(or MBps, which is one-eighth the megahertz rate). The figure is invariably the peak transfer rate, the quickest that a single byte can possibly move, at least in theory, when all conditions are the most favorable and theres a strong tailwind. In truth, information never moves
at the peak transfer rateonly the signals containing the data switch at the peak rate. The
actual flow of data is burdened by overhead of various sorts, both in the hardware interface
607
608
and the software data exchange protocol. That said, although the actual number expressed
in the peak transfer rate is purely bogus if you want to count how fast data moves, such
numbers do allow you to compare interfaces. Transferring information in modern systems
requires about the same overhead, notwithstanding the interface standard (although there
are differences). Therefore, a disk that has a peak transfer rate of 320MBps is faster than
one with a rate of 133MBps, regardless of how fast each can actually move bytes.
In modern usage, the peak transfer rate is often reserved for discussions of disk interfaces.
Disks themselves are constrained by their physical nature. How fast they can produce
information is governed by how closely the information is linearly packed and how fast it
moves under the read/write head. The faster it moves, the more data the head sees in a
second.
The biggest factor in determining this speed is the spin rate of the disk. A faster spinning
disk reveals more flux transitions to the read/write head. As a result, today most people
focus on the spin rate of the disk to measure its data-transfer performance.
Disk Caching
The ultimate means of isolating your computer from the mechanical vagaries of hard disk
seeking is disk caching. Caching eliminates the delays involved in seeking when a read
request (or write request in a system that supports write caching) involves data already
stored in the cachethe information is retrieved at RAM speed. Similarly, the cache
pushes the transfer rate of data stored in the cache up to the ceiling imposed by the slowest interface between the cache and host microprocessor. With an on-disk cache, the drive
interface will likely be the primary constraint; with a hardware cache in the disk controller
or host adapter, the bus interface is the limit; with a software cache, microprocessor and
memory-access speed are the only constraints.
AV Drives
Drive-makers have tailored a special breed of hard disk to suit the needs of audio and
video recording and editing. Called AV drives because their primary application is audio
and video, these drives are optimized for transferring large blocks of data sequentially
(unlike normal hard disks, which must provide instant access to completely random data).
Audio and video files tend to be long and linear and therefore read sequentially for long
periods. Fast access time is not as important for such applications as is a high sustained
data-transfer rate. For example, most video production today requires data rates of
27MBps but uses compression ratios averaging about ten to one to produce a data stream
of about 2.7MBps that needs to be stored and retrieved. Most modern hard disks can
achieve the performance required by audio and video applications, but hard diskmakers
still offer a number of drives specially tailored to AV use.
To achieve the highest possible performance, these AV drives add extensive buffering and
may sacrifice or delay some data-integrity features. For example, they trade off absolute
Hard Disks
data security to eliminate interruptions in the high-speed flow of information. The hard
disk industry rationale for this design is that video data, unlike spreadsheets or databases,
tolerates errors well, so a short sequence of bad data wont hurt anything. After all, a single sector isnt even a third of a line of uncompressed video. The video subsystem can correct for such one-time errors in part of a line, masking them entirely.
One way to prevent interruptions in the flow of data is to alter error handling. Engineers
usually require that most hard disk drives attempt to reread the disk when they encounter
an error. Most manufacturers use elaborate algorithms to govern these rereads, often
minutely changing the head position or performing an entire thermal calibration. If these
are not successful, the drive may invoke its error-correction code to reconstruct the data.
AV drives alter these priorities. Because error correction is entirely electronic and imposes
no mechanical delays, AV drives use it first to attempt error recovery. Only after error
correction fails may the drive try to reread the data, often for a limited number of retries.
Advanced hard disks log the errors that they encounter so that they may be used later for
diagnostics. The logging operation itself takes time and slows disk performance. AV drives
delay error logging until it does not interrupt the data stream.
One feature that slows down conventional drives is sector remapping, which AV drives avoid
to sidestep its performance penalties. The remapping process imposes delays on the flow
of data because the read/write head of the drive must jump from one place to another to
write or retrieve the data from the remapped sector at the time of the data access. AV
drives often avoid auto-relocation to prevent interruption of the flow of high-speed data.
Drive Arrays
When you need more capacity than a single hard disk can provide, you have two choices:
Trim your needs or plug in more disks. But changing your needs means changing your
lifestyleforegoing instant access to all your files by deleting some from your disk, switching to data compression, or keeping a tighter watch on backup files and intermediary versions of projects under development. Of course, changing your lifestyle is about as easy as
teaching an old dog to change its spots. The one application with storage needs likely to
exceed the capacity of todays individual hard disks200GB and climbingis a network
server, and a total lifestyle change for a network server is about as probable as getting a platoon of toddlers to clean up a playroom littered with a near-infinite collection of toys.
Consequently, when the bytes run really low, youre left with the need for multiple disks.
In most single-user computers, each of these multiple drives acts independently and
appears as a separate drive letter (or group of drive letters) under common Windows.
Through software, such multiple-drive systems can even be made to emulate one large
disk with a total storage capacity equal to that of its constituent drives. Because Windows
95 and Windows 98 handle I/O seriallythey can do only one I/O task at a timesuch a
609
610
solution is satisfactory, but its hardly the optimum arrangement where reliability and providing dozens of users instant access is concerned. Instead of operating each disk independently, you can gain higher speeds, greater resistance to errors, and improved reliability by
linking the drives through hardware to make a drive arraywhat has come to be known as
a Redundant Array of Inexpensive Disks, or RAID.
Principles
The premise of the drive array is elementarycombine a number of individual hard disks
to create a massive virtual system. But a drive array is more than several hard disks connected to a single controller. In an array, the drives are coordinated, and the controller
specially allocates information between them using a program called Array Management
Software (AMS). The AMS controls all the physical hard disks in the array and makes
them appear to your computer as if they were one logical drive. For example, in some
drive arrays, the AMS ensures that the spin of each drive is synchronized and divides up
blocks of data to spread among several physical hard disks.
The obvious benefit of the drive array is the same as any multiple-disk installation
capacity. Two disks can hold more than one, and four more than two. But drive array
technology can also accelerate mass-storage performance and increase reliability.
Data Striping
The secret to both of these innovations is the way the various hard disks in the drive array
are combined. They are not arranged in a serial list, where the second drive takes over
once the capacity of the first is completely used up. Instead, data is split between drives at
the bit, byte, or block level. For example, in a four-drive system, two bits of every byte
might come from the first hard disk, the next two bits from the second drive, and so on.
The four drives could then pour a single byte into the data stream four times faster
moving all the information in the byte would only take as long as it would for a single
drive to move two bits. Alternatively, a four-byte storage cluster could be made from a
sector from each of the four drives. This technique of splitting data between several drives
is called data striping.
At this primitive level, data striping has a severe disadvantage: The failure of any drive in
the system results in the complete failure of the entire system. The reliability of the entire
array can be no greater than that of the least reliable drive in the array. The speed and
capacity of such a system are greater but so are the risks involved in using it.
Redundancy and Reliability
By sacrificing part of its potential capacity, an array of drives can yield a more reliable,
even fault-tolerant, storage system. The key is redundancy. Instead of a straight division of
the bits, bytes, and blocks each drive in the array stores, the information split between the
drives can overlap.
Hard Disks
For example, in the four-drive system, instead of each drive getting two bits of each byte,
each drive might store four. The first drive would take the first four bits of a given byte,
the second drive the third, fourth, fifth, and sixth bits; the third drive, the fifth, sixth, seventh, and eighth bits; and the fourth drive, the seventh, eighth, first, and second bits. This
digital overlap allows the correct information to be pulled from another drive when one
encounters an error. Better yet, if any single hard disk should fail, all the data it stored
could be reconstituted from the other drives.
This kind of system is said to be fault tolerant. That is, a single faultthe failure of one
hard diskwill be tolerated, meaning the system operates without the loss of any vital
function. Fault tolerance is extremely valuable in network applications because the crash
of a single hard disk does not bring down the network. A massive equipment failure therefore becomes a bother rather than a disaster.
The sample array represents the most primitive of drive array implementations, one that
is particularly wasteful of the available storage resources. Advanced information-coding
methods allow for higher efficiencies in storage, so a strict duplication of every bit is not
required. Moreover, advanced drive arrays even allow hot-swapping, a feature that permits
a failed drive to be replaced and the data that was stored upon it reconstructed without
interrupting the normal operation of the array. A network server with such a drive array
need not shut down even for disk repairs.
Implementations
In 1988, three researchers at the University of California at BerkeleyDavid A.
Patterson, Garth Gibson, and Randy H. Katzfirst outlined five disk array models in a
paper titled A Case for Redundant Arrays of Inexpensive Disks. They called their models
RAID Levels and labeled them as RAID 1 through 5, appropriately enough. Their numerical designations were arbitrary and were not meant to indicate that RAID 1 is better or
worse than RAID 5. The numbers simply provide a label for each technology that can be
readily understood by the cognoscenti.
In 1993, these levels were formalized in the first edition of the RAIDBook, published by
the RAID Advisory Board, an association of suppliers and consumers of RAID-related
mass storage products. The book is part of one of the RAID Advisory Boards principle
objectivesthe standardization of the terminology of RAID-related technology.
Although the board does not officially set standards, it does prepare them for submission to the recognized standards organizations. The board also tests the function and
performance of RAID products and verifies that they perform a basic set of functions
correctly.
The RAID Advisory Board currently recognizes nine RAID implementation levels. Five
of these conform to the original Berkeley RAID definitions. Beyond the five array levels
described by the Berkeley group, several other RAID terms are used and acknowledged
611
612
by the RAID Advisory Board. These include RAID Level 0, RAID Level 6, RAID
Level 10, and RAID Level 53.
The classification system is nonhierarchicala higher number does not imply a better or
more advanced technology. The numbers are no more than labels for quickly identifying
the technologies used. Because the common perceptionreally a misperceptionis that
the numbers do imply a ranking and that higher is better, some manufacturers have developed proprietary labels (RAID 7) or exploited non-Berkeley definitions (RAID 10 and 53)
with high numbers that hint they are somewhat better than the lower-numbered systems.
Although each level has its unique advantages (and disadvantages), no one RAID technology is better than any of the others for all applications.
In an attempt to avoid such confusion, the RAID Advisory Board now classifies disk array
products by what they accomplish in protecting data rather than by number alone. The
boards Web site includes both a description and listing of classified products.
RAID Level 0
Early workers used the term RAID Level 0 to refer to the absence of any array technology.
According to the RAID Advisory Board, however, the term refers to an array that simply
uses data striping to distribute data across several physical disks. Although RAID Level 0
offers no greater reliability than the worst of the physical drives making up the array, it
can improve the performance of the overall storage system. For example, reading data in
parallel from two drives can effectively double throughput.
RAID Level 1
The simplest of drive arrays, RAID Level 1, consists of two equal-capacity disks that mirror one another. One disk duplicates all the files of the other, essentially serving as a
backup copy. Should one of the drives fail, the other can serve in its stead.
This reliability is the chief advantage of RAID Level 1 technology. The entire system has
the same capacity as one of its drives alone. In other words, the RAID Level 1 system
yields only 50 percent of its potential storage capacity, making it the most expensive array
implementation. Performance depends on the sophistication of the array controller.
Simple systems deliver exactly the performance of one of the drives in the array. A more
sophisticated controller could potentially double data throughput by simultaneously reading alternate sectors from both drives. Upon the failure of one of the drives, performance
reverts to that of a single drive, but no information (and no network time) is lost.
RAID Level 2
The next step up in array sophistication is RAID Level 2, which interleaves bits or blocks
of data as explained earlier in the description of drive arrays. The individual drives in the
array operate in parallel, typically with their spindles synchronized.
Hard Disks
To improve reliability, RAID Level 2 systems use redundant disks to correct single-bit
errors and detect double-bit errors. The number of extra disks needed depends on
the error-correction algorithm used. For example, an array of eight data drives may use
three error-correction drives. High-end arrays with 32 data drives may use seven errorcorrection drives. The data, complete with error-detection code, is delivered directly to
the array controller. The controller can instantly recognize and correct for errors as they
occur, without slowing the speed at which information is read and transferred to the host
computer.
The RAID Level 2 design anticipates that disk errors occur often, almost regularly. At one
time, mass storage devices might have been error prone, but no longer. Consequently,
RAID Level 2 can be overkill except in the most critical of circumstances.
The principal benefit of RAID Level 2 is performance. Because of their pure parallel
nature, RAID Levels 2 and 3 are the best-performing array technologies, at least in systems that require a single, high-speed stream of data. In other words, RAID Level 2 yields
a high data-transfer rate. Depending on the number of drives in the array, an entire byte
or even a 32-bit double-word could be read in the same period it would take a single drive
to read one bit. Normal single-bit disk errors dont hinder this performance in any way
because of RAID Level 2s on-the-fly error correction.
The primary defect in the RAID Level 2 design arises from its basic storage unit being
multiple sectors. As with any hard disk, the smallest unit each drive in the array can store
is one sector. File sizes must increase in units of multiple sectorsone drawn from each
drive. In a ten-drive array, for example, even the tiniest two-byte file would steal ten sectors (5120 bytes) of disk space. (Under the Windows VFAT system, which uses clusters of
four sectors, the two-byte file would take a total of 20,480 bytes!) In actual applications
this drawback is not severe because systems that need the single-stream speed and instant
error correction of RAID Level 2 also tend to be those using large files (for example,
mainframes).
RAID Level 3
This level is one step down from RAID Level 2. Although RAID Level 3 still uses multiple drives operating in parallel, interleaving bits or blocks of data, instead of full error correction it allows only for parity checking. That is, errors can be detected but without the
guarantee of recovery.
Parity checking requires fewer extra drives in the arraytypically only one per array
making it a less expensive alternative. When a parity error is detected, the RAID Level 3
controller reads the entire array again to get it right. This rereading imposes a substantial
performance penaltythe disks must spin entirely around again, yielding a 17 millisecond
delay in reading the data. Of course, the delay appears only when disk errors are detected.
Modern hard disks offer such high reliability that the delays are rare. In effect, RAID
613
614
Level 3 compared to RAID Level 2 trades off fewer drives for a slight performance
penalty that occurs only rarely.
RAID Level 4
This level interleaves not bits or blocks but sectors. The sectors are read serially, as if the
drives in the array were functionally one large drive with more heads and platters. (Of
course, for higher performance, a controller with adequate buffering could read two or
more sectors at the same time, storing the later sectors in fast RAM and delivering them
immediately after the preceding sectors have been sent to the computer host.) For reliability, one drive in the array is dedicated to parity checking. RAID Level 4 earns favor
because it permits small arrays of as few as two drives, although larger arrays make more
efficient use of the available disk storage.
The dedicated parity drive is the biggest weakness of the RAID Level 4 scheme. In writing, RAID Level 4 maintains the parity drive by reading the data drives, updating the parity information, and then writing the update to the parity drive. This read-update-write
cycle adds a performance penalty to every write, although read operations are unhindered.
RAID Level 4 offers an extra benefit for operating systems that can process multiple data
requests simultaneously. An intelligent RAID Level 4 controller can process multiple
input/output requests, reorganize them, and read its drives in the most efficient manner,
perhaps even in parallel. For example, while a sector from one file is being read from one
drive, a sector from another file can read from another drive. This parallel operation can
improve the effective throughput of such operating systems.
RAID Level 5
This level eliminates the dedicated parity drive from the RAID Level 4 array and allows
the parity-check function to rotate through the various drives in the array. Error checking
is thus distributed across all disks in the array. In properly designed implementations,
enough redundancy can be built in to make the system fault tolerant.
RAID Level 5 is probably the most popular drive array technology currently in use
because it works with almost any number of drives, including arrays as small as two, yet
permits redundancy and fault tolerance to be built in.
RAID Level 6
To further improve the fault tolerance of RAID Level 5, the same Berkeley researchers
who developed the initial five RAID levels proposed one more, now known as RAID
Level 6. This level adds a second parity drive to the RAID Level 5 array. The chief benefit is that any two drives in the array can fail without the loss of data. This enables an
array to remain in active service while an individual physical drive is being repaired, yet
still remain fault tolerant. In effect, a RAID Level 6 array with a single failed physical disk
becomes a RAID Level 5 array. The drawback of the RAID Level 6 design is that it
Hard Disks
requires two parity blocks to be written during every write operation. Its write performance is extremely low, although read performance can achieve levels on par with RAID
Level 5.
RAID Level 10
Some arrays employ multiple RAID technologies. RAID Level 10 represents a layering of
RAID Levels 0 and 1 to combine the benefits of each. (Sometimes RAID Level 10 is called
RAID Level 0&1 to more specifically point at its origins.) To improve input/output performance, RAID Level 10 employs data striping, splitting data blocks between multiple drives.
Moreover, the Array Management Software can further speed read operations by filling
multiple operations simultaneously from the two mirrored arrays (at times when both
halves of the mirror are functional, of course). To improve reliability, the RAID level uses
mirroring so that the striped arrays are exactly duplicated. This technology achieves the
benefits of both of its individual layers. Its chief drawback is cost. As with simple mirroring
it doubles the amount of physical storage needed for a given amount of logical storage.
RAID Level 53
This level represents a layering of RAID Level 0 and RAID Level 3the incoming data is
striped between two RAID Level 3 arrays. The capacity of the RAID Level 53 array is the
total of the capacity of the individual underlying RAID Level 3 arrays. Input/output performance is enhanced by the striping between multiple arrays. Throughput is improved by
the underlying RAID Level 3 arrays. Because the simple striping of the top RAID Level 0
layer adds no redundant data, reliability falls. RAID Level 3 arrays, however, are inherently so fault tolerant that the overall reliability of the RAID Level 53 array far exceeds
that of an individual hard disk drive. As with a RAID Level 3 array, the failure of a single
drive will not adversely affect data integrity.
Which implementation is best depends on what you most want to achieve with a drive
array: Efficient use of drive capacity, fewest number of drives, greatest reliability, or quickest performance. For example, RAID 1 provides the greatest redundancy (thus reliability),
and RAID 2 the best performance (followed closely by RAID 3).
Parallel Access Arrays

In parallel access arrays, all the individual physical drives in the array participate in every
input and output operation of the array. In other words, all the drives operate in unison.
Systems that correspond to the RAID Level 2 or 3 design fit this definition. The drives in
independent access arrays can operate independently. In advanced arrays, several individual
drives may perform different input and output operations simultaneously, filling multiple
input and output requests at the same time. Systems that follow the RAID Level 4 or 5
design fit this definition. Although RAID Level 1 drives may operate either as parallel
access or independent access arrays, most practical systems operate RAID Level 1 drives
independently.
615
616
Interfacing
Just connecting four drives to a SCSI controller wont create a drive array. An array
requires special electronics to handle the digital coding and control of the individual
drives. Usually these special electronics take the form of a RAID controller. The controller
may be part of a standalone disk array in its own cabinet, in which case the array attaches
to the host computer as a single SCSI device. Alternatively, the RAID controller may be a
single expansion board that resembles a host adapter for a SCSI or AT Attachment interface but incorporates RAID electronics and usually megabytes of cache or buffer memory.
The disk drives for the array may be installed inside the computer host or in a separate
chassis.
Most disk arrays use SCSI hard disks because that interface allows multiple drives to share
a single connection. A growing number of manufacturers now offer AT Attachment (IDE)
array controllers, which allow you to take advantage of lower cost ATA hard disks. The
earliest of these supported only the four drives normally allowed in a single computer system under the ATA specifications. These typically allow you your choice of a RAID Level
0 or RAID Level 1 configuration. At least one manufacturer now offers a RAID Level 5
controller for ATA hard disk drives.
Floppy Disks
Since the first computer booted up, the floppy disk has been a blessing and a curse, subject of the same old saying usually reserved for kids, spouses, and governments: You cant
live with them, and you cant live without them. They were the first storage system for
personal computers, and theyve remained relatively unchanged ever since.
Microprocessors might be 10,000 times faster than they used to be. Hard disks might hold
10,000 times more bytes. But todays basic floppy disk hardly holds about the same
amount of data as the very first. Although the floppy disk has shrunk from eight inches to
three-and-a-half and grown a hard turtle-like shell, the basic disk still holds little more
than a megabyte, truly puny in a world of giga-everything.
Despite the best efforts of computer-makers, floppy disks arent dead yet. You can still
depend on them in the vilest of emergencies, when nothing else works to bring your computer to life. You (or a technician) can still troubleshoot your computer using its floppy
disk drive. And you can still exchange files with friends on the cheap, even disposable,
floppy disk medium. The floppy disk still has a role, although a modest one, in the life of
your computer.
The floppy disk provides a recording medium that has several positive qualities. The flat
disk surface allows an approximation of random access. As with hard disks, data is
arranged in tracks and sectors. The disk rotates the sectors under a read/write head, which
travels radially across the disk to mark off tracks. More importantly, the floppy disk is a
Floppy Disks
removable medium. You can shuffle dozens of floppies in and out of drives to extend your
storage capacity. The floppy disk in the drive provides online storage. Offline, you can
keep as many floppy disks as you want.
The term floppy disk is one of those amazingly descriptive terms that abound in this age of
genericisms. Inside its protective shell, the floppy disk medium is both a floppy (flexible)
and wide, flat disk. To protect it, the floppy resides in a shell. The first floppies had a truly
floppy protective shell, one made out of thicker but still flexible Mylar. Today, the floppy
fits into a hard case and overall is not very floppy. The disk inside, the one made from the
media, remains floppy so the name remains the sameuniquely accurate in a world of
computers inhabited by spinning fixed disks and recordable read-only memory.
History
When the floppy was first conceived, personal computers didnt exist. Its creator, IBM,
first used the floppy disk to store diagnostic programs and microcode for its large computer systeminstead of tape (too cumbersome, and many applications didnt require
such a large capacity) or memory chips (too expensive). It was the Dark Ages equivalent
of a CD-ROM. It was also big, about eight inches square, but for the time had a large
capacityabout 100KBand random-access capabilities.
By 1973, engineers had adapted the eight-inch floppy to a convenient read/write medium
suitable for both the original application and for storage for data-entry systems, such as
IBMs DisplayWriter word processing system. The 8-inch floppy disk had a number of
features going for it that made it desirable as a computer data-storage medium. The
floppy disk was compact (at least compared to the ream of paper that could hold the same
amount of information), convenient, and standardized. Above all, it was inexpensive to
produce and reliable enough to depend on.
In 1976, Shugart Associates introduced the 5.25-inch floppy disk, a timely creation that
exactly complimented the first commercial personal computers, introduced at about the
same time. (Both Apple Computer and Microsoft Corporation were founded in 1976,
although fledgling Microsoft offered a BASIC interpreter as its first productits operating system for floppy disks did not arrive until 1981.) Because these were smaller than the
older eight-inch variety, these 5.25-inch floppies were called diskettes by some. The irregularly used name later spread to even smaller sizes of the floppy disk.
In 1980, Sony Corporation introduced the 3.5-inch floppy disk of the same mechanical
construction that we know it today. After a lukewarm reception, the little disks gained a
foothold when Apple adopted them for its initial Macintosh in 1984. Portable computers
added more incentive for the move to 3.5-inch disks, as did IBMs adoption of the format
in 1987. The standard floppy disk has remained unchanged since then, despite further
attempts to improve it (such as the quad-density disks with double the normal capacity).
617
618
The first appearance of Mount Rainier technology in CD-RW drives (see Chapter 18,
Optical Storage (CD and DVD)) in late 2002 likely marked the end-of-the-line for the
floppy disk. This new industry initiative endows the CD with all the capabilities of a
floppy disk, in addition to its much greater capacity. As Mount Rainier drives become
widespread, floppy disk drives will become truly obsolete.
Medium
The traditional floppy disk medium itself is the thin, flexible disk inside the protective
shell. This disk is actually a three-layer sandwich, the meat of which is a polyester substrate that measures about 3.15 mils (thousandths of an inch), or 80 micrometers thick.
The bread is the magnetic recording medium itself, a coating less than one-thousandth of
an inch thick on each side of the substrate.
The floppy disk medium starts out as vast rolls of the substrate that are coated in a continuous process at high speed. The stamping machine cuts individual disks from the resulting
roll, or web, of medium, like a cookie cutter. After some further mechanical preparation
(for example, a metal hub is attached to the cookies of 3.5-inch disks), another machine
slides the disks into their protective shells.
Although retailers sell some floppy disks as single sided, a mixture of magnetic oxide and
binder coats both sides of the substrate of all floppy disks. Those sold as single sided are
only tested on one side, and their manufacturers only certify that one side for storing data.
By convention, the bottom surface of the disk is used in single-sided floppy disk drives.
Magnetic Properties
The thickness of the magnetic coating on the floppy disk substrate varies with the disk
type and storage density. In the most common disk types, it measures from 0.035 mil to
0.1 mil (that is, 0.9 to 2.5 micrometers). In general, the higher the storage density of the
disk, the thinner the magnetic coating. The individual particles are also finer grained.
Table 17.3 lists the coating thicknesses for both current and obsolete floppy disk types.
TA B L E 1 7 . 3
Floppy Disk Media Characteristics
Disk Type
Coating Thickness
Coercivity
5.25-inch double-density
2.5 micrometers
290 oersteds
5.25-inch high-density
1.3 micrometers
660 oersteds
3.5-inch double-density
1.9 micrometers
650 oersteds
3.5-inch high-density
0.9 micrometers
720 oersteds
Although all common floppy disk coatings use ferric oxide magnetic media, engineers
have tailored the magnetic particles in the mix to the storage density at which the disks
Floppy Disks
are used. The higher storage density media generally have higher coercivities, as shown in
Table 17.3.
Despite the names used by various floppy disk types, all floppy disk drives use Modified
Frequency Modulation (MFM) recording, which produces double-density recording. In
other words, both normal and high-density disks are double-density, even though many
manufacturers reserve the double-density term for lower-capacity disks.
Extra-high-density disks (that is, the 2.88MB floppies still occasionally sold) use a novel
technology called perpendicular recording. In conventional floppy disks, the grains of magnetic medium are aligned flat against the substrate, and the write process aligns them
along a radius of the disk. The particles of the medium used in perpendicular recording
are arranged vertically so that one end of each particle points toward the substrate. A special read/write head in the disk drive changed the vertical orientation of the magnetic field
of the media particles. In addition, extra-high-density disks use a high-coercivity bariumferrite recording medium.
Shell
The chief difference between a modern 3.5-inch floppy disk and the earliest disks is the
hard plastic shell. Magnetically and logically, the disks work the same as the first 8-inch
floppies. Drives, too, are basically the same, only miniaturized to suit the medium and
modern computers.
The shell protects the magnetic medium inside from whatever terrors lurk in your world,
be they teething infants, churning chair casters, or simple carelessness. The tough shell
allows you to put a label on the disk and then write whatever identification you want on it,
even with a ballpoint pen. The thinner shells of earlier disks could not protect the media
within from pen points, so writing on a label already affixed to an early floppy could
crease the medium and make it unreadable. Figure 17.2 shows the shell of a 3.5-inch
floppy disk and its notable external features.
FIGURE 17.2
Layout of 3.5-inch
floppy disk.
Write protect tab

Density sensing hole
Label area
Drive hub
Disk access hole
Shutter
Alignment notches
The shell of a 3.5-inch floppy disk measures 3.7 inches (94 millimeters) front to back, and
just over 3.5 inches (90 millimeters) wide. Each disk is a little more than one-eighth inch
(3.3 millimeters) thick. A single disk weighs just over three-quarters of an ounce, about
619
620
22 grams. Despite its 3.5-inch name, the media disk inside the shell actually measures
about 3.4 inches (nominally 86 millimeters) across.
To protect the medium inside from dirt, dust, and fingerprints, the head access opening
on a 3.5-inch floppy disk is covered with a spring-loaded sliding metal shield or shutter. It
opens automatically only when you insert a disk into a drive and slides closed when you
pop the disk from the drive. The shutter is an effective dust shield that prevents contaminants from collecting on the medium surface whenever the disk is out of its drive. This
protection means that 3.5-inch disks include all the protection they need and dont need
an additional sleeve or shuck that older floppy disks required.
Liners
As protective as the hard shell of 3.5-inch floppies may be, the medium needs protection
from it as well. A thin sheet of unwoven cloth akin a loose-woven paper towel cushions
each side of the medium from rubbing against the shell. The interlocked threads of the
liner serve as a slip-sheet and dust collector. Light contact with the strands of the liner let
the medium slide with little friction. At the same time, the material sweeps dust and particles from the surface of the medium so that it doesnt scrape against the read/write head
of the drive.
Hub
The creators of the 3.5-inch floppy disk learned that the one area of older floppies that
sustained the most damage was the hub hole. Every time you slide a disk into a drive,
drive hubs clamp onto the disk to hold and spin it. The hub entered the drive hole and
forced the entire disk into the proper position for proper reading and writing. At times,
the hub clamping down damaged the edges of the hole. After a disk suffered enough damage, it could become unreliable or unusable. The only protection to the hub hole offered
by older floppy disks was an optional reinforcing ring some manufacturers elected to add
to the perimeter of the hub hole.
The 3.5-inch floppy eliminates such problems by using a sturdy metal hub with a square
center spindle hole, which mates with the mechanism of the disk drive. The stamped steel
hub resists damage. The single rectangular cutout in the hub allows the drive mechanism
to unambiguously identify the radial alignment of the disk. The cutout serves as a mechanical reference for the position of data on the disk. The hub itself is glued to the disk media.
Insertion Key
A 3.5-inch floppy disk has four edges and two sides, giving you eight possible ways to try
to slide it into your disk drive, only one of which is correct. Although the over-square
design of the floppy shell prevents you from sliding a disk in sideways, you can still shove
one in backward or upside down. To ensure that you dont damage your disk drive by
improperly inserting a disk, the shell of 3.5-inch floppies is keyed by a notch in one
Floppy Disks
corner. A tab in the drive blocks this corner, so if the notch is not present you cannot slide
the disk all the way into the drive. The last few millimeters of sliding in the disk open the
shutter and load the heads against the disk, so the notch prevents the heads from ramming
against the floppys shell instead of the access area.
Molded into the plastic of the shell is a small arrow that serves as a visual reminder to you.
It points to the edge of the disk that you should slide into your disk drive, just in case the
shutter is not enough guidance.
Write Protection
The fundamental design of the floppy disk is to serve as a read/write medium so that you
can store information on it and read it back. Sometimes, however, you might want to protect the data on a floppy disk from change. For example, you might back up your archives
to floppy disk. Software vendors prefer protection against writing to the distribution disks
holding their programs so that you dont accidentally erase the code and become a support problem for them.
The 3.5-inch floppy design incorporates a write-protect tab that allows you to make any
floppy disk a read-only medium. The design uses a hole and a plastic slider. When the
slider blocks the hole, you can read, write, and format the disk. When the slider is moved
back to reveal the hole, an interlock on the drive prevents you from writing to the disk.
You can move the slider back and forth to write-protect a disk and then make it writable
again as often as you want.
Software vendors often remove the slider entirely. Without the slider, the write-protected
hole cannot be covered and the disk is permanently write-protected. You can, however,
circumvent even this permanent form of write-protection by blocking off the write-protect hole. The easiest way is to cover both sides of the hole with opaque tape to make a
distribution disk writable. This method is not without its dangers, however. If the tape
does not stick tightly to the disk, it can jam the drive mechanism, likely preventing you
from popping or pulling the disk out of the drive.
Density Key
In order for your disk drive to determine the type of magnetic medium on your disk so
that its electronics can be adjusted to match the disks coercivity, 3.5-inch floppy disks
incorporate a density key. This key is actually the presence or absence of a hole in one corner of the disk shell. Double-density disks lack a hole. High-density disks have a hole. An
extra-high-density notch marks disks with 2.88MB capacity. Higher-density disks also
have a visual indication of their capacityfor example, the stylized HD silk-screened
near the shutter of high-density disks.
The connections between many floppy disk drives and the host computer often do not
properly relay the density key information to the computer and its operating system.
621
622
When the density key information is not available, you can format a double-density disk
as high density. In years gone by, some enterprising manufacturers also offered holepunches designed to add a density-key hole to double-density disks so that you could format them as high density. Although the difference in coercivities between double- and
high-density media is modest, there are other differences in the formulation of the media
that make double-density disks unreliable at high-density capacities. Moreover, the holepunches often left residue in the form of small particles of plastic to contaminate the disks
(at the factory the density key is molded rather than punched). These contaminants can
shorten the life of the medium or damage your disk drive.
Format
Four formats are commonly used for 3.5-inch floppy disks, three of which are supported
on computers. (Computers do not support single-sided 3.5-inch floppy disk formats.) Your
disks drives and operating system automatically adjust to the format of the disks you
attempt to read, providing your drive is capable of reading the format. All higher-capacity
drives can read formats of lower capacity. Table 17.4 summarizes the essential characteristics of these four formats for 3.5-inch floppy disks.
TA B L E 1 7 . 4
The Characteristics of a 3.5-inch Floppy Disk
Capacity
Units
360KB
720KB
1.44MB
2.88MB
Sides
Number
Tracks
Number
80
80
80
80
Sectors per track
Number
18
36
Sector size
Bits
512
512
512
512
Rotation rate
RPM
300
300
300
300
Data-transfer rate
Kbps
500
500
500
1000
Bit density (max)
BPI
8717
8717
17,434
34,868
Track density
TPI
135
135
135
135
Coercivity
Oersteds
650
650
720
1200
The capacity of a floppy disk is set when the disk is formatted. Using the Format option
from the Windows menu associated with your floppy disk drive, you can select the capacity of new floppy disks (or reformat floppies to change their capacity). Preformatted disks
relieve you of the chore of formatting floppies yourself, although you can always format
over the factory format to change the capacity of a disk.
Floppy Disks
Double density is the starting format for 3.5-inch floppy disks. It uses 80 tracks with nine
sectors per track. The tracks are spaced 135 to the inch (about 5.3 to the millimeter). The
small diameter disk only allows a swath about 0.6 inch (15 millimeters) wide around the
disk for reading and writing. The high-density format merely doubles the number of sectors per track, packing the data in twice as tightly.
Because 3.5-inch floppy disks spin at a fixed speed of 300 RPM, doubling the density of
the data on each track also boosts the speed at which the information is read from the
disk. The basic reading speed of 250Kbps for double-density is doubled to 500Kbps with
high-density disks.
Extra-high-density disks again double the sector count on each track, to 36 sectors per
track, without changing the number of tracks or their spacing. Again, this increase in storage density also increases the reading and writing speed to 1000Kbps.
Note that the capacities of all floppy disks are given with the format in place. The formatting data steals away some of the usable storage area of the disk. Floppy diskmakers occasionally list the unformatted capacities of their products. The unformatted capacity of a
double-density (720KB formatted) disk is 1MB; that of a high-density (1.44MB formatted)
disk is 2MB; that of an extra-high-density (2.88MB formatted) disk is 4MB.
Before CD-ROM media took over the job of program distribution, software publishers
sought and developed ways of shoehorning more data onto every floppy disk. The leading
extra-capacity alternative was Microsofts Distribution Media Format (DMF). This variation
in the high-density design allowed Microsoft to fit 1,720,320 bytes on a standard highdensity 3.5-inch floppy disk in place of the more normal 1,474,560 bytes (nominal
1.44MB). The DMF format differed from the standard format in that it used 21 sectors
per track instead of the normal 18. DMF squeezed more sectors on each track by reducing
the interrecord gap (the space between sectors) down to nine bytes.
The differences went deeper, however. Each track used a 2:1 interleave factor so that sectors did not appear in order. This interleaving resulted in slower reading because the disk
had to spin around twice for each track to be read. The DMF format also skewed the sectors on adjacent tracks by three sectors so that Sector 1 on Track 4 sat next to Sector 1 on
Track 2.
The small interrecord gap made DMF disks difficult to write to with normal floppy disk
drives. In fact, Microsoft called DMF a read-only format. In any case, you cannot write to
DMF disks using ordinary software. However, several special utilities are available for
copying and even creating DMF disks. Note that Microsoft enforces a limit of 16 entries
in the root directory of a DMF disk by only allocating a single cluster to service as the
root, so DMF-formatted disks usually use a subdirectory structure for their contents.
623
624
Microsoft operating systems cannot ordinarily read DMF floppies. Consequently, DMF
floppies were used only when software required multiple floppy disks for its installation.
The first disk of the installation packageusually called the setup diskloaded software to
reprogram your floppy disk controller to read the DMF disks.
Zip Disks
The Zip disk, developed and first marketed by Iomega Corporation in 1995, was the first
floppy disk system with a capacity larger than 100MB. Although initially a proprietary system, Iomega has licensed Zip to other companies. Both drives and media are now available from multiple sources, although the format is totally under the control of Iomega.
To gain its large capacity, the Zip system uses optical technology that was first applied to a
little-remembered 20MB floppy disk system called Floptical. The Zip medium uses an
optically read servo track to allow repeatable head positioning in fine increments.
The first generation of Zip disks had listed nominal capacities of either 25MB (25,107,968
actual bytes) or 100MB (actually 100,431,872 bytes) per disk.
By increasing the density of storage on slightly modified media, in 1998 Iomega was able
to boost the capacity of the Zip system by a factor of two and a half without major modifications to the drive mechanism. As a result, Iomega was able to offer an improved system
with 250MB capacity at the same initial price as the original Zip drive.
Thanks to the increased lineal density of data on the 250MB Zip tracks, the data throughput of the new system is more than two times greater than the old, although access time
remains comparable. Actual data throughput often is limited by the drive interface. Highcapacity Zip drives use the same interfaces as the old, and performance through parallel
ports is compromised.
The shell of the Zip disk makes it a true cartridge. To achieve speeds in the hard disk
range, the medium must spin at a high rate, and the friction from rubbing against a liner
is an anathema to speed. The thicker cartridge gives the Zip disk spinning room and
enables it to rotate at 2968 RPM. In addition, the Zip disk drive has hard disklike access
speeds, with the first generation of drives having a 26 millisecond average access time.
The disk requires about three seconds to spin up to speed or spin down, which becomes a
factor only when youre exchanging cartridges.
The actual Zip media disk inside a cartridge measures true 3.5-inches across. Consequently,
the cartridge must be larger than conventional 3.5-inch floppies and MO cartridges, measuring 3.7 inches (94 millimeters) square and a quarter-inch (6.35 millimeters) thick. These
dimensions alone make the Zip disk incompatible with traditional floppy disks.
Instead of using a mechanical write-protect mechanism on the cartridge shell, Zip disks
are write-protected electronically. The Iomega system provides three protection modes
Floppy Disks
with optional password access limits as part of its ZipTools software. The three modes
include conventional write protection that prevents the inadvertent alteration of data on
the disk; read/write protection, which requires a password to access data on the disk; and
unprotect until eject, which lets you work with the data on the disk but protects the disk
when you remove it from the drive. The same software that adds write protection is
required to remove it. Passwords, however, are not recoverableeven by Iomega.
Zip disk drives read only Zip disks. Iomega has manufactured products using any of four
interfaces or portsAT Attachment (IDE), parallel port, SCSI, and USB. IDE is now
favored for internal drives, and USB 2.0 is favored for external drives.
The move to 250MB required changes to both the medium and drives, although superficially high-capacity Zip disks look almost identical to their older, low-capacity forebears.
The drive electronics were adapted to accommodate the higher read and write rates
required by the new media, and the magnetic medium has a higher coercivity to improve
high-density storage. As a consequence, older drives cannot be upgraded, and disks with
250MB capacity are rejected outright by older drives as incompatible.
Because of the laser/mechanical formatting of the disks and the differing media, lowcapacity Zip disks cannot be reformatted to higher capacities. In fact, the servo tracks cannot be erased by any means that would not destroy the cartridge.
On the other hand, high-capacity Zip drives are fully backward compatible with old
media. New 250MB drives can read and write 100MB disks. Old drives, however, can neither read nor write 250MB media.
In 2002, Iomega further refined the Zip drive to increase its capacity to 750MB per disk.
The chief trick was simply to increase the storage density, resulting in 4,780 tracks per
inch with data squeezed into each track at 137Kb per inch. The disk itself spins at 3676
RPM. The drives use the USB 2.0 interface and feature a data-transfer rate that peaks at
7.5Mbps. Although compatible with USB 1.0 connections, using the older interface will
severely degrade the performance of the drive.
The new 750MB disks can only be read and written by matching 750MB drives. However,
the drives are backward compatible with earlier Zip disk formats. A 750MB drive can both
read and write 250MB disks and read 100MB disks. It cannot write 100MB disks.
SuperDisks
Jointly developed by Compaq Computer Corporation, the storage products division of
3M Company (now Imation Corporation), Matsushita-Kotobuki Electronics Industries,
Ltd., and O. R. Technology, parent company of Optics Research, Inc., the SuperDisk was
supposed to be the next generation of floppy disk drive. First marketed in March, 1996 as
the LS-120 system, the first drives were made by MKE, installed in Compaq computers,
625
626
and used media manufactured by 3M. The optical technology used by the drives was
developed by ORT.
Outwardly a SuperDisk resembles an ordinary 3.5-inch floppy diskette. The only obvious
difference is the shape of the shield over the head slot. Instead of rectangular, the
SuperDisk wears a roughly triangular shield. Figure 17.3 shows an LS-120 diskette.
FIGURE 17.3
An LS-120 diskette.
Inside, however, the SuperDisk system uses a thinner substrate (2.5 mils) than traditional
floppies, one that is cut from a different plastic, polyethylene terathalate (PET). This substrate is more flexible than the traditional polyester to give the SuperDisk medium better
bend around the head for more reliable contact.
Format
Each SuperDisk diskette can store up to 125,829,120 bytes, or 120MB, accounting for the
designation of the drive. Not all of that capacity is usable. As with conventional floppy
disks, part of the total capacity must be devoted to the FAT and directory data.
The SuperDisk drive combines two technologies to increase its capacity: an opto-mechanical laser servo system and zoned recording.
The major increase comes from increasing the track density to 2490 per inch, achieved
using the laser servo system. The small diameter of the disk only allows 1736 tracks per side
at this density. Even so, it requires a special medium, a two-layer metal particle compound.
For compatibility with computer hardware and software, the track layout of the SuperDisk
medium is mapped to a logical format (the way the drive appears to your computer),
Floppy Disks
having 960 tracks per side. This mapping results in the drivewhich has two heads, one
for each disk sideappearing to your system as if it has eight heads.
The SuperDisk system uses zoned recording with 55 different zones across the radius of each
disk. The number of sectors per track varies from 51 on the innermost tracks to 92 at the
outer edge. Table 17.5 lists the physical and logical formats of an LS-120/SuperDisk diskette.
TA B L E 1 7 . 5
LS-120 Disk Format
Format Type
Cylinders
Heads
Sectors
Physical
1736
51 to 92 (zoned recording)
Logical
960
32
The SuperDisk medium spins at a constant 720 revolutions per minute. As a result of the
constant spin and varying sector count, a SuperDisk drive reads data from the disk at a
rate that changes with the track position of the read/write head. Near the center of the
disk, the transfer rate is lowest, about 400KBps. At the outer edge, the transfer rate
reaches 665KBps. Initial SuperDisk drives had an average access time of about 70 milliseconds. Table 17.6 lists the specifications of the SuperDisk medium and its format.
TA B L E 1 7 . 6
LS-120 Specifications
Feature
Specification
Media diameter
3.5 inches
Recording medium
Metal particle
Capacity (formatted)
120MB
Bytes per sector
512
Sectors per track
5192 (ZBR)
Track data capacity
26,11247,616KB
Tracks per surface
1736
Track density
2490 tracks per inch
Areal density
33,660 flux changes/inch
Recording surfaces (sides)
Track pitch
10.2 m
Peak transfer rate
3.20 to 5.33MBps
Throughput
313571
Estimated life
5 million passes
627
628
Technology
The stepper-motor-based open-loop mechanisms used by conventional floppy disk drives
cannot correctly position a read/write head with sufficient precision to reliably locate the
fine tracks used by the SuperDisk system. To achieve the necessary track density, the
SuperDisk system used an embedded-servo system. Servo data is etched into the disk during factory preparation. The SuperDisk drive then uses a laser to detect and read the
etched servo information from the disk and align the read/write head properly on each
track. In fact, the LS designation is an abbreviation for laser servo.
The read/write head has two gapsone used for the fine tracks used in high-density recording and one used for working with conventional double-density and high-density floppy disk
media. The laser servo design does not increase the capacity of conventional floppy disk
media.
Compatibility
All SuperDisk drives are backward compatible with standard 1.44MB floppy disks. The
drives can both read and write standard floppies, and the 1.44MB floppies written on a
SuperDisk drive are readable by conventional floppy disk drives. Because of the higher
spin speed of the SuperDisk drive, however, it can achieve a higher transfer rate with a
1.44MB floppyapproximately 2.5 times higherthan a conventional floppy disk drive.
The SuperDisk media are designed to appear as write-protected to conventional 1.44MB
floppy disk drives by virtue of a cutout that corresponds to the write-protect hole in a
conventional floppy. Consequently, there is no risk of damage to data by inadvertently
sliding an LS-120 disk into a conventional floppy disk drive.
The SuperDisk media are beyond the control capabilities of the chip on the controllers
for conventional floppy disk drives. The initial SuperDisk drives instead used an IDE
interface.
Mechanisms
As computer equipment goes, floppy disk drives are simple devices. The essential components are a spindle motor, which spins the disk, and a stepper-motor, which drives a metal
band in and out to position the read/write heads, and an assembly that is collectively
called, as with hard disks, the head actuator. A spring-loaded latch locks the disk in place
and ejects it when you press the button on the front of the drive.
After more than two decades of development, about the only refinement made to the conventional floppy disk system has been miniaturization. Drives now can be less than onethird the height of the first. No matter their size, however, all conventional floppy disk
drives work in essentially the same way.
Floppy Disks
To carry out their design purpose, all floppy disk drive mechanisms must be able to carry
out a few basic tasks. They have to spin the disks at a uniform speed. They must also
move their read/write heads with sufficient precision to locate each and every data track
on a given disk. And the open-loop head-positioner design requires a known starting
place, an index, which the drive must be able to locate reliably.
Speed Control
All the electronics packed onto the one or more circuit boards attached to the drive unit
merely control those simple disk drive operations. A servo system keeps the disk spinning
at the correct speed. Usually an optical sensor looks at a stroboscopic pattern of black dots
on a white disk on the spindle assembly. The electronics count the dots that pass the sensor in a given period to determine the speed at which it turns, adjusting it as necessary.
Some drives use similar sensors based on magnetism rather than optics, but they work in
essentially the same waycounting the number of passing magnetic pulses in a given
period to determine the speed of the drive.
Head Control
Other electronics control the radial position of the head assembly to the disk. The stepper-motor that moves the head reacts to voltage pulses by moving one or more discrete
steps of a few degrees (hence, the descriptive name of this type of motor). Signals from
the floppy disk controller card in the host computer tell the disk drive which track of the
disk to move its head to. The electronics on the drive then send the appropriate number
of pulses to the stepper-motor to move the head to the designated track.
The basic floppy disk mechanism receives no feedback on where the head is on the disk. It
merely assumes it gets to the right place because of the number of steps the actuator
makes. Because the drive does its best to remember the position of the head, hard reality
can leave the head other than in its expected place. For instance, you can reach in and
manually jostle the head mechanism. Or you might switch off your computer with the
head halfway across the disk. Once the power is off, all the circuitry forgets, and the location of the head becomes an unknown.
Head Indexing
So that the head can be put in the right place with assurance, the floppy disk drive resorts
to a process called indexing. That is, it moves the head as far as it will go toward the edge
of the disk. Once the head reaches this index position, it can travel no farther, no matter
how hard the actuator tries to move it. The drive electronics make sure that the actuator
moves the head a sufficient number of steps (a number greater than the width of the disk)
to ensure that the head will stop at the index position. After the head has reached the
index position, the control electronics can move it a given number of actuator steps and
know exactly where on the radius of the disk the head is located.
629
630
Tape
Tape was the first magnetic mass storage system used in computers, harking back to the
days of room-size Univacs and vacuum tubes. It first proved itself as a convenient alternative to punched cards and punched paper tapethe primary storage system used by mainframe computers. Later, information transfer became an important use for tape. Databases
could be moved between systems as easily as carting around one or more spools of tape.
After magnetic disks assumed the lead in primary storage, tape systems were adapted to
backing them up.
As a physical entity, tape is both straightforward and esoteric. It is straightforward in
design, providing the perfect sequential storage mediuma long, thin ribbon that can
hold orderly sequences of information. The esoteric part involves the materials used in its
construction.
The tape used by any system consists of two essential layersthe backing and the coating.
The backing provides the support strength needed to hold the tape together while it is
flung back and forth across the transport. Progress in the quality of the backing material
mirrors developments in the plastics industry. The first tape was based on paper. Shortly
after the introduction of commercial tape recorders at the beginning of the 1950s, cellulose acetate (the same plastic used in safety film in photography for three decades previously) was adopted. The state-of-the-art plastic is polyester, of double-knit leisure-suit
fame. In tape, polyester has a timeless style of its ownflexible and long-wearing with a
bit of stretch. It needs all those qualities to withstand the twists and turns of todays torturous mechanisms, fast shuttle speeds, and abrupt changes of direction. The typical tape
backing measures from one-quarter mil (thousandth of an inch) to one mil thick, about 10
to 40 microns.
The width of the backing varies with its intended application. Wider tapes offer more area
for storing data but are most costly and, after a point, become difficult to package. The
narrowest tape in common use, cassette tape, measures 0.150 inches (3.8 millimeters)
wide. The widest in general use for computing measures 0.5 inches (12.7 millimeters).
Equipment design and storage format determine the width of tape to be used.
Coatings have also evolved over the decades, as they have for all magnetic media. Where
once most tapes were coated with doped magnetic oxides, modern coatings include particles of pure metal in special binders and even vapor-plated metal films. Tape coatings are
governed by the same principles as other magnetic media; the form is different but the
composition remains the same. As with all magnetic storage systems, modern tape media
have higher coercivities and support higher storage densities.
Taken by itself, tape is pretty hard to get a handle on. Pick up any reasonable length of
tape, and youll have an instant snarl on your hands. The only place that tape is used by
Tape
itself is in endless loops (one end spliced to the other) in special bins used by audio and
video duplicating machines. In all other applications, the tape is packaged on reels or in
cartridges.
Reels came first. A simple spool onto which a length of tape gets wound, the reel is the
simplest possible tape carrier. In this form, tape is called open reel. Putting tape in a cartridge adds a permanent package that both provides protection to the delicate medium and
makes it more convenient to load. The most basic cartridge design simply packages a reel
of tape in a plastic shell and relies on an automatic threading mechanism in the drive itself.
All current computer-size tape systems use a more sophisticated designcassette-style cartridges that include both the supply and take-up reels in a single cartridge.
The basic cassette mechanism simply takes the two spools of the open-reel tape transport
and puts them inside a plastic shell. The shell protects the tape because the tape is always
attached to both spools, eliminating the need for threading across the read/write heads
and through a drive mechanism. The sides of the cassette shell serve as the sides of the
tape reelholding the tape in place so that the center of the spool doesnt pop out. This
function is augmented by a pair of Teflon slip-sheets, one on either side of the tape inside
the shell, that help to eliminate the friction of the tape against the shell. A clear plastic
window in either side of the shell enables you to look at how much tape is on either
spoolhow much is left to record on or play back.
The reels inside the cassette are merely hubs that the tape can wrap around. A small clip
that forms part of the perimeter of the hub holds the end of the tape to the hub. At various points around the inside of the shell, guides are provided to ensure that the tape travels in the correct path.
More recent tape cartridges have altered some of the physical aspects of the cassette
design but retain the underlying technologies.
Technologies
Tape systems are often described by how they workthat is, the way they record data
onto the tape. For example, although the term streaming tape thats appended to many tape
drives may conjure up images of a cassette gone awry and spewing its guts inside the dashboard of your card (and thence to the wind as you fling it out the window), it actually
describes a specific recording mode that requires an uninterrupted flow of data. At least
four of these termsstart-stop, streaming, parallel, and serpentinecrop up in the specifications of common tape systems for computers.
Start-Stop Tape
The fundamental difference between tape drives is how they move the tape. Early drives
operated in start-stop modethey handled data one block (ranging from 128 bytes to a few
631
632
kilobytes) at a time and wrote each block to the tape as it was received. Between blocks of
data, the drive stopped moving the tape and awaited the next block. The drive had to prepare the tape for each block, identifying the block so that the data could be properly
recovered. Watch an old movie with mainframe computers with jittering tape drives, and
youll see the physical embodiment of start-stop tape.
Streaming Tape
When your computer tape drive gets the data diet it needs, bytes flow to the drive in an
unbroken stream and the tape runs continuously. Engineers called this mode of operation
streaming tape. Drives using streaming tape technology can accept data and write it to tape
at a rate limited only by the speed the medium moves and the density at which bits are
packedthe linear density of the data on the tape. Because the tape does not have to stop
between blocks, the drive wastes no time.
Parallel Recording
Just as disk drives divide their platters into parallel tracks, the tape drive divides the tape
into multiple tracks across the width of the tape. The number of tracks varies with the
drive and the standard it follows.
The first tape machines used with computer systems recorded nine separate data tracks
across the width of the tape. The first of these machines used parallel recording, in which
they spread each byte across their tracks, one bit per track, with one track for parity. A
tape was good for only one pass across the read/write head, after which the tape needed to
be rewound for storage. Newer tape systems elaborate on this design by laying 18 or 36
tracks across a tape, corresponding to a digital word or double word, written in parallel.
Parallel recording provides a high transfer rate for a given tape speed because multiple
bits get written at a time but makes data retrieval time consumingfinding a given byte
might require fast forwarding across an entire tape.
Serpentine Recording
Most computer tape systems use multitrack drives but do not write tracks in parallel.
Instead, they convert the incoming data into serial form and write that to the tape. Serial
recording across multiple tracks results in a recording method called serpentine recording.
Serpentine cartridge drives write data bits sequentially across the tape in one direction on
one track at a time, continuing for the length of the tape. When the drive reaches the end
of the tape, it reverses the direction the tape travels and cogs its read/write head down one
step to the next track. At the end of that pass, the drive repeats the process until it runs
out of data or fills all the tracks. Figure 17.4 shows the layout of tracks across a tape using
four tracks of serpentine recording.
Tape
FIGURE 17.4
Layout of four tracks
using serpentine
recording.
1
2
3
4
A serpentine tape system can access data relatively quickly by jogging its head between
tracks, because it needs to scan only a fraction of the data on the tape for what you want.
Additionally, it requires only a single channel of electronics and a single pole in the
read/write head, thus lowering overall drive costs. Modern serpentine systems may use
over 50 tracks across a tape.
Helical Recording
The basic principle of all the preceding tape systems is that the tape moves past a stationary head. In a helical scan recording system, both the head and tape move. Usually multiple
rotating heads are mounted on a drum. The tape wraps around the drum outside its protective cartridge. Two arms pull the tape out of the cartridge and wrap it about halfway
around the drum (some systems, like unlamented Betamax, wrap tape nearly all the way
around the drum). So that the heads travel at an angle across the tape, the drum is canted
at a slight angle, about five degrees for eight-millimeter drives and about six degrees for
DAT. The result is that a helical tape has multiple parallel tracks that run diagonally
across the tape instead of parallel to its edges. These tracks tend to be quite finesome
helical systems put nearly 2000 of them in an inch.
In most helical systems, the diagonal tracks are accompanied by one or more tracks parallel to the tape edge used for storing servo control information. In video systems, one or
more parallel audio tracks may also run the length of the tape. Figure 17.5 shows how the
data and control tracks are arranged on a helical tape.
FIGURE 17.5
Helical scan recording
track layout.
Data tracks
Control track
Helical scan recording can take advantage of the entire tape surface. Conventional stationaryhead recording systems must leave blank areasguard bandsbetween the tracks containing
data. Helical systems can and do overlap tracks. Although current eight-millimeter systems
use guard bands, DAT writes the edges of tracks over one another.
This overlapping works because the rotating head drum actually has two (or more) heads
on it, and each head writes data at a different angular relationship (called the azimuth) to
the tracks on the tape. In reading data, the head responds strongly to the data written at
633
634
the same azimuth as the head and weakly at the other azimuth. In DAT machines, one
head is skewed 20 degrees forward from perpendicular to its track; the other head is
skewed backward an equal amount.
Formats
Each of these various recording methods, along with physical concerns such as tape width
and cartridge design, allow for a nearly infinite range of tape-recording systems. When
you try to make sense of backup systems, it often seems like all the possibilities have been
tried in commercial products.
Linear Recording Tape Systems

The most straightforward tape systems use simple linear recording. That is, they move
the tape past a stationary head just like the very first audio tape recorders. Linear recording is the old reliable in tape technology. The process and methods have been perfected
(as much as that is possible) over more than five decades of use.
Open-reel tape is still used in computer systems, but most personal computers use more
modern, more compact, and more convenient cassette-style cartridges. The most common
of these are patterned after a design the 3M Company first offered using quarter-inch tape
as a data-recording medium. First put on the market in 1972, these initial quarter-inch
cartridges were designed for telecommunications and data-acquisition applications calling
for the storage of serial data, such as programming private business telephone exchanges
and recording events. Compared to the cassette, the tape cartridge requires greater precision and smoother operation. To achieve that end, a new mechanical design was invented
by Robert von Behren of the 3M Company, who patented it in 1971the quarter-inch
cartridge mechanism.
Instead of using the capstan drive system like cassettes, the quarter-inch cartridge operates
with a belt drive system. A thin, isoelastic belt stretches throughout the cartridge mechanism, looping around (and making contact with) both the supply and take-up spools on
their outer perimeters. The belt also passes around a rubber drive wheel, which contacts a
capstan in the tape drive.
The capstan moves the belt but is cut away with a recess that prevents it from touching
the tape. The friction of the belt against the outside of the tape reels drives the tape. This
system is gentler to the tape because the driving pressure is spread evenly over a large area
of the tape instead of pinching the tape tightly between two rollers. In addition, it provides for smoother tape travel and packing of the tape on the spools. The tape is wound,
and the guide and other parts of the mechanism arranged so that the fragile magnetic surface of the tape touches nothing but the read/write head (see Figure 17.6).
Tape
FIGURE 17.6
Quarter-inch cartridge
mechanisms (DC-600
style cartridge is
shown; the DC2000 is
similar but smaller).
For sturdiness, the cartridge is built around an aluminum baseplate. The rest of the cartridge is transparent plastic, allowing the condition of the tape and the mechanism to be
readily viewed.
To try to lessen the chaos in the tape cartridge marketplace, a number of tape drive
manufacturersincluding DEI, Archive, Cipher Data, and Tandbergmet together
at the National Computer Conference in Houston in 1982. They decided to form a
committee to develop standards so that a uniform class of products could be introduced.
The organization took the name, Working Group for Quarter-Inch Cartridge Drive
Compatibility, often shortened into QIC committee. In November, 1987, the organization
was officially incorporated as Quarter-Inch Cartridge Standards, Inc.
QIC developed a number of standards for data cartridges. But as personal computers
became popular, most tape companies realized that the cartridges were just too big.
Squeezing a drive to handle a six-by-four cartridge into a standard 5.25-inch drive bay is a
challenge; fitting one in a modern 3.5-inch bay is impossible. Seeking a more compact
medium, quarter-inch cartridge makers cut their products down to size, reducing tape
capacity while preserving the proven drive mechanism. The result was the mini-cartridge.
QIC quickly developed standards for the new cartridge.
Because the 3M Company introduced the first mini-cartridge with the designation
DC2000, many people in the computer industry persist in calling all mini-cartridges
DC2000-style cartridges. Mini-cartridges actually come in a variety of designations. As
with the current model numbers of DC600-size cartridges, the model designations of
most mini-cartridges encode the cartridge capacity as their last digits. For example, a
DC2080 cartridge is designed for 80MB capacity; a DC2120 for 120MB.
One big advantage of the smaller cartridges is that drives for them easily fit into standard
3.5-inch bays. The mini-cartridge package measures just under 3.252.50.625 inches. As
originally developed, it held 205 feet of tape with the same nominal quarter-inch width
used by larger cartridges, hence the initial 2 in the designation.
After trying to exploit the compact size of the mini-cartridge for years, media-makers
decided they had gone too far. They cut too much capacity from their cartridges.
635
636
Therefore, they added some extra inches back on the back of the cartridges so that the
cartridges would still slide into smaller drives but could hold more tape. Two formats
developed: Travan and QIC-EX.
Travan
In 1995, tape- and drive-makers pushed mini-cartridges to yet higher capacities with the
simple expedient of making the cartridges bigger so that they could hold more tape. The
fruit of the labors of an industry group comprising Conner, Iomega, HP, 3M Company,
and Sony was called Travan technology.
The increase in cartridge size appears in three dimensions. The cartridges are both wider,
deeper, and taller. The added height allows the use of wider tape, the same eight-millimeter (0.315-inch) tape used by the QIC-Wide format, the format pioneered by Sony. The
wider tape permits an increase in track count by about 30 percent (from 28 to 36 in the
initial Travan format).
In addition, Travan adds an extra half-inch to the width and depth of cartridges. The
Travan cartridge measures 0.53.62.8 inches (HWD), smaller in the front (3.2 inches)
than the rear, where the tape spools reside. Internally, the Travan cartridge uses the same
3M mechanism as other quarter-inch cartridges and must be formatted before use.
Although the increase seems modest, it allows the tape capacity to more than double,
from 307 feet in a DC2120 cartridge to 750 feet in the Travan. The extra size gives
Travan a distinctive shape, basically rectangular with two corners curved in to make sliding in a cartridge easier. Figure 17.7 shows a Travan cartridge.
FIGURE 17.7
A Travan tape
cartridge.
The combination of more tracks and greater tape length was sufficient to boost single cartridge capacity to 400MB (uncompressed) in the initial Travan implementation. Nothing
about the Travan design impairs making further improvements, so whatever developments
add to the capacity of standard mini-cartridges can be directly reflected in increased
Travan capacities. In fact, since its introduction, Travan has been improved three times,
each change about doubling single-cartridge capacities.
The first change, creating TR-2, boosted the coercivity of the medium and allowed for a
50-percent increase in data density, both in a greater number of tracks and a higher linear
density on each track. The next change doubled linear density again, without altering the
Tape
track count, to create TR-3. Another increase in linear density and matching increase in
track count boosted the TR-4 Travan implementation to 4GB per cartridge. Table 17.7
summarizes the characteristics of the various Travan implementations.
TA B L E 1 7 . 7
Travan Tape Cartridge Specifications
Model
TR-1
TR-2
TR-3
TR-4
Capacity,
uncompressed
400MB
800MB
1.6GB
4GB
Capacity,
compressed
800MB
1.6GB
3.2GB
8GB
Minimum
transfer rate
62.5KBps
62.5KBps
125KBps
567KBps
Maximum
transfer rate
125KBps
125KBps
250KBps
567KBps
Media length
750 ft.
750 ft.
750 ft.
740 ft.
Media width
0.315 in.
0.315 in.
0.315 in.
0.315 in.
Media coercivity
550 Oe
900 Oe
900 Oe
900 Oe
Tracks
36
50
50
72
Data density
14,700 ftpi
22,125
ftpi
44,250
ftpi
50,800
ftpi
Interface
Floppy
Floppy
Floppy
SCSI/E-IDE
Read/write
compatibility
QIC-80
QIC-3010
3020/3010
3080/3095
Read-only
compatibility
QIC-40
QIC-80
QIC-80
QIC-3020
Perhaps the most notable part of the Travan design is its compatibility: A Travan drive
accepts standard DC2000 cartridges and QIC-Wide cartridges as well as its own native
mediathats two different cartridge sizes and tape widths in one drive. Compatibility
extends to both reading and writing all three cartridge styles. The various Travan implementations also feature backward compatibility. The more recent standards can read tapes
made under the earlier Travan standards.
QIC-EX
As long as the cartridge in a QIC drive sticks out like Travan, you might as well let it all
hang out. That dated clich underlies the philosophy of QIC-EX. By allowing a cartridge
to stick nearly three inches out of the drive, QIC-EX accommodates up to 1000 feet of
tape in the QIC format. Thats more than double the capacity of ordinary mini-cartridges.
637
638
Increase the tape width to QIC-Wides eight millimeters, and you can fit gigabytes onto a
standard tape cartridge, as shown in Figure 17.8.
FIGURE 17.8
Dimensions of a
QIC-EX tape cartridge.
Unlike Travan, which requires a wider drive with a wider throat to accommodate the
wider cartridges, the QIC-EX design is the same width as standard mini-cartridges. QICEX tapes consequently fit into most mini-cartridge drives to give an instant capacity
increase.
Helical-Scan Systems
Although helical-scan recording was originally designed for video signals, it is also an
excellent match for digital data. The high tape-to-head speed can translate to high datatransfer rates. Although the narrow data tracks devote but a small area of recording
medium to storing information, so small that it increases the noise content of the
recorded signal, digital technology handily avoids the noise. As with all digital signals,
the electronics of the helical-scan tape system simply ignore most of the noise, and error
correction takes care of anything that cant be ignored.
Helical-scan systems are the children of the digital age. They trade the mechanical precision required in the linear-scan systems for servo-controlled designs. Servo-mechanisms in the helical-scan recorder control both the tape speed and head speed to ensure
that the read/write head exactly tracks the position of the tracks on the tape. Although
servo-electronics can be built with other technologies, digital electronics and miniaturization make the required circuitry trivial, a single control chip. Moreover, servo-control automatically compensates for inaccuracies in the operation and even construction
of the tape-transport system. This design allows manufacturers to use less expensive fabrication processes and helps the mechanism age gracefully, adjusting itself for the wear
and tear of old age.
Three different helical-scan systems are currently used in tape systems for digital data:
eight-millimeter tape, four-millimeter tape (which is formally called the Digital Data
Standard), and a proprietary system sold by Pereos.
Tape
Eight Millimeter
Probably the most familiar incarnation of eight-millimeter tape is in miniaturized camcorders. Sony pioneered the medium as a compact, high-quality video-recording system.
The same tapes were later adapted to data recording by Exabyte Corporation and first
released on the computer market in 1987.
In the original Exabyte eight-millimeter digital recording system, the head drum rotated
at 1800 revolutions per minute while the tape traveled past it at 10.89 millimeters per second to achieve a track density of 819 per inch and a flux density of 54 kilobits per inch
enough to squeeze 2.5MB on a single cartridge. Improvements extended the capacity to
5MB without compression and up to 10GB with compression. The tape can be rapidly
shuttled forward and backward to find any given location (and block of data) within about
15 seconds.
Eight-millimeter drives tend to be quite expensive. Complete systems cost thousands of
dollars. A raw drive alone starts at over one thousand dollars. Its primary market is backing up file servers, machines that can benefit from its huge capacity.
Advanced Intelligent Tape
Sony developed an alternative eight-millimeter technology. By incorporating nonvolatile
RAM memory into the shell of an eight-millimeter tape, a tape drive can quickly retrieve
directory information without the need to scan the tape. It doesnt have to read the tape at
all to find out whats there.
This memory is what makes Sonys Advanced Intelligent Tape so smart. Each cartridge has
16KB for storing table of contents and file-location information. Each cartridge can store
25MB of uncompressed data. With data compression the AIT system can achieve transfer
rates of 6MBps and a capacity of 50MB per cartridge.
Despite being based on eight-millimeter technology, AIT is not backward compatible with
the older medium. AIT drives reject conventional eight-millimeter tapes as unsupported
media, and hardware write-protection prevents older drives from altering AIT tapes.
Digital Data Storage
Developed originally as a means to record music, the Digital Data Storage first saw commercial application as Digital Audio Tape (DAT), a name sometimes applied to the system
even when it is used for storing computer data. The technology was first released as a
computer storage medium in 1989. Using a tape thats nominally four millimeters wide
(actually 3.81 mm or 0.150 in, the same as cassette tape), the system is sometimes called
four-millimeter tape. The thin tape fits into tiny cartridges to store huge amounts of data.
The first DAT system could pack 1.3GB into a cassette measuring only 0.4 2.9 2.1
inches (HWD). The result was, at the time, the most dense storage of any current
639
640
computer tape medium, 114 megabits per square inch on special 1450 oersted metal particle tape (the same material as used by eight-millimeter digital tape systems). A cassette
held either 60 or 90 meters of this tape. The shorter tapes stored 1.3GB; the longer
tapes, 2.0GB.
Under the aegis of Hewlett-Packard and Sony Corporation, in 1990 this format was formalized as DDS, sometimes listed as DDS-1 (to distinguish it from its successors). The
same format underlies DDS-DC, which adds data compression to the drive to increase
capacity. Although DDS at first was listed in some literature (and in an earlier edition of
this book) as an abbreviation for Digital Data Standard, the DDS organization that coordinates the specification renders it officially as Digital Data Storage.
In 1993, the same companies updated the standard to DDS-2, doubling both the datatransfer speed and capacity of the system as well as introducing new 120-meter cassettes
using a tape coated with a metal powder medium.
In 1995, the standard got upped again to DDS-3, which can pack up to 12GB on each
cartridge or 24GB of compressed data. In addition, the DDS-3 standard allows for transfers at double the DDS-2 rate with speed potential of up to 1.5 megabits per second (discounting the effects of compression). DDS-3 uses the same tape media as its predecessor
format but increases the data density to 122 kilobits per inch. The key to this higher density is a new technology termed Partial Response Maximum Likelihood (PRML).
Announced in 1998, DDS-4 further enhances the four-millimeter format to provide 20GB
of storage on a single tape. Part of the capacity increase results from longer tapes, up to
155 meters per cartridge. In addition, DDS-4 reduces track width by 25 percent. Because
of its higher data density, DDS-4 also gains a bit of speed. The standard allows for transfer rates from 1 to 3Mbps.
DDS-5, originally planned for the year 2001, aimed for a native capacity of 40GB per cartridge but has yet to be formally released. The DDS Web site has not been updated since
1999. Table 17.8 summarizes the various DDS implementations.
TA B L E 1 7 . 8
Summary of DDS Specifications
Format
DDS
DDS-DC
DDS-2
DDS-3
DDS-4
DDS-5
Year introduced
1989
1991
1993
1995
1999
2001
Cartridge capacity
1.3GB
2.0GB
4.0GB
12GB
20GB
40GB
Transfer rate
183Kbps
183Kbps
360 to
750Kbps
720Kbps to
1.5Mbps
1 to
3Mbps
1 to
6Mbps
Tape length
60 m
90 m
120 m
125 m
155 m
N/A
Tape
In a DDS drive, the tape barely creeps along, requiring about three seconds to move an
incha tape speed of 8 millimeters per second. The head drum, however, spins rapidly at
2000 revolutions per minute, putting down 1869 tracks across a linear inch of tape with
flux transitions packed 61 kilobits per inch.
The first DAT systems shared tapes with audio DAT recorders. Since that time, however,
the later DDS standards have required improved magnetic media with higher coercivities.
As a result, to get the highest reliability you must use a tape made to the standard that
matches your drive. Although you may be able to read older tapes in a later DDS drive,
successful writing (where coercivity comes into play) is less probable. Similarly, DDS tape
will not work in your DAT recorder. The DDS Manufacturers Group, the organization
that coordinates DDS activities, has also developed a technology called Media Recognition
System that is able to detect audio DAT tapes. The DDS organization has developed distinctive logos for each level of the DDS hierarchy so that you can readily distinguish tapes
and drives. Figure 17.9 shows the logos for DDS-2 through DDS-4.
FIGURE 17.9
The official DDS-2,
DDS-3, and DDS-4
logos.
Hewlett-Packard and Sony hold intellectual property rights to DDS technology, and all
drives using DDS must be licensed by them. The DDS Manufacturers Group maintains a
Web site at www.dds-tape.com.
Alas, all tape technologies have lost their luster. They are all hard-pressed to keep up with
the huge increases in disk storage densities and capacities. The tape systems with adequate
capacity for backing up todays hard disk drives cost more than most personal computers.
Consequently, most individual users prefer other backup methods, such as extra hard
disks, networked storage (including backing up over the Internet), DVD recorders, and
simply ignoring the need for backing up. Data centers rely on a few super-capacity backup
systems that are priced well beyond the reach of individuals. Tape is therefore not quite
dead. You can use either a mini-cartridge or DDS system for backing up a few gigabytes
(about 20GB without compression with todays drives). That should be sufficient for keeping a spare copy of your personal files (including such megabyte-hogs as Computer-Aided
Design drawings, graphics, and videos). However, DVD is probably the best, lowest-cost
bet for backup today.
641
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
18
Optical Storage (CD and DVD)

Optical media are todays most important software distribution method,
for both computer programs and entertainment such as music and
videos. Add writing capabilities, and your optical drive can also back up
your hard disk or make discs to send files to friends or production
houses. You can even make your own CDs for your stereo system or turn
home movies into private-label DVDs. And a read/write optical drive
can even substitute as your primary mass storage system should your
hard disk fail and you have patience even a saint would envy. When the
new Mount Rainier system makes its way into operating systems, your
CD drive will likely fully take over the role formerly assigned to floppy
disks.
Although the CD and DVD are often treated as two different systems,
they share a common heritage and technology. The DVD is really only
the second generation of optical disc storage for computers. The major
technical differences are a result of improvements in technology over the
nearly two-decade history of the CD medium. No revolutionary changes
separate the two. In fact, the biggest changes only reflect new formats
created to take advantage of the larger capacity the technical refinements
permit.
History
Developed by the joint efforts of Philips and Sony Corporation in the
early 1980s, when the digital age was taking over the stereo industry, the
Compact Disc was first and foremost a high-fidelity delivery medium.
The CD was initially released in the United States in 1983 (in Japan the
644
CD got a one-year head start, officially released in 1982 according to Sony), and within
five years it had replaced the vinyl phonograph record as the premiere stereophonic
medium because of its wide range, lack of noise, near invulnerability to damage, and long
projected life.
The CD was designed chiefly for capacity and real-time playback of music. Engineers set
many of the practical aspects of the CD around the requirements of music recording. For
example, they selected the 70 or so minutes of music capacity as one of the core specifications in designing the system, because a primary design goal was to fit the entire
Beethovens Ninth Symphony, without interruption, on a single disc. As originally conceived, its storage was not reliable enough for computers. And computers had little use for
it. At the time, they were choking when confronted with a few megabytes. CDs wielded
hundreds of them.
But as the CD rose to prominence as the primary distribution medium for prerecorded
music, computer engineers began to look at the shiny medium with a covetous gleam in
their eyes. They saw the digital storage provided by the disc as a repository for more
megabytes than anyone had reason to use. After all, data is data (okay, data are data)
regardless of whether the bytes encode a symphony or an operating system. When someone got the idea that a plastic puck that cost a buck to make and retailed for $16.99 could
be filled with last years statistics and marketed for $249, the rush was on. The Compact
Disc became the CD-ROM (which stands for Compact Disc, Read-Only Memory), and
megabytes came to the masses.
Soon sound became only one of the applications of the Compact Disc medium. The original name had to be extended to distinguish musical CDs from all the others. To computer
people, the CD of the stereo system became the CD-DA, Compact Disc, Digital Audio.
Engineers tinkered with the storage format of the CD to stuff the discs with other kinds
of data. Philips optimized the medium for interactive applicationsmultimedia presentations and gamesto create CD-I, which stands for Compact Disc Interactive. Some even
thought that compressiona lot of itcould fit video on the discs. Compact Disc-Video
succeeded in squeezing video on the little discs, but not very much and not very well.
Viewable video had to await another string of developments and the new generation of
optical storage.
DVD was the needed innovation. The initials stand for Digital Versatile Disc, although
the system was first termed the Digital Video Disc before the adoption of the current
technical standards.
The roots of DVD go back to two competing proposals, both of which had the primary
intent of storing video on a disc the same size as a CD. The original developers of the
Compact Disc, Philips and Sony, backed a format they called MMCD for Multimedia
Compact Disc (they owned the CD name so they took advantage of it). The other camp,
CHAPTER 18 OPTICAL STORAGE (CD AND DVD)
History
led by Matsushita, Time Warner, and Toshiba, developed their own, incompatible format
they called SD.
For a while, the industry appeared poised for a repeat of the Beta/VHS debacle that put
two mutually incompatible videotape cassette formats on the market for nearly a decade.
In September, 1995, the industry appeared to come to its senses, hammering out a single
standard agreeable to both camps. To distinguish it from the earlier efforts and reflect the
expanded range of possibilities afforded by the new medium, the format was rechristened
with Versatile replacing the Video of the earlier proposals. Credit for developing the initial
standard is generally given to an industry consortium that included Hitachi, JVC,
Matsushita, Mitsubishi, Philips, Sony, Thompson, Time Warner, and Toshiba.
As with CDs, each application format for the DVD has its own subdesignation. These
include DVD-Video for video applications, such as the distribution of motion pictures;
DVD-Audio, as a high-quality audio disc with capabilities far beyond todays 16-bit discs;
and DVD-ROM for the distribution of computer software and other data.
When it came to developing a recordable format for DVD, however, the consensus fell
apart. As this is written, promoters advocate three read-only formats under the names
DVD-R(A), DVD-R(G), and DVD+R. Read/write systems are equally confused with four
competing standards, including DVD-Multi, DVD-RAM, DVD-RW, and DVD+RW.
Although the differences between these recordable standards is blurringeach one now
boasts a full 4.7 gigabytes-per-disc capacity as well as some degree of compatibility with
home DVD playersthe animosity is not fading. The industry is split into two camps, the
plus camp and the hyphen camp. The former, the promoters of the plus systems of
DVD+R and DVD+RW, include Dell, Hewlett-Packard, Mitsubishi, Philips, Ricoh, Sony,
and Thomson. The hyphen camp promotes the DVD-Multi format, which brings
together DVD-R(A) and DVD-R(G) as well as DVD-RAM and DVD-RW. The DVD
Forum, an organization that includes nearly all consumer electronics manufacturers (with
the notable exceptions of Dell and Hewlett-Packard but including the other plus camp
members), maintains the official standards.
Despite the wide variety of formats now used in computer optical storage systems, the
underlying technology for all remains the samea spinning disc is the target for the laser
beam.
Medium
The heart of both the Compact Disc and Digital Versatile Disc systems is the disc
medium itself. Its design, based on a pattern of dots that can be read optically but massproduced mechanically, makes optical storage the fastest and least expensive medium for
duplicating hundreds of megabytes of data.
645
646
The flat disc shape also offers a distinct advantage. Machines can mold copies of discs by
stamping them between dies instead of filling a three-dimensional mold with a casting liquid. This stamping process has a long history. It has been used for over a century in duplicating recorded musicfirst with shellac and clay to copy Emil Berliners first
phonograph records, then with vinyl for old-fashioned record albums. Duplicating CDs
and DVDs requires extra precision and a few extra steps, but it remains essentially the
same stamping process.
The disc-duplicating process begins with a disc master. A mastering machine equipped
with a high-powered laser blasts the pits in a blank discthe recording masterto make
an original mechanical recording. Then the master is made into a mold called a stamper. A
negative copy is electroplated onto the master, then separated from it, leaving the master
unscathed. One master can make many duplicate molds, each of which is then mounted in
a stamping machine. The machine heats the mold and injects a glob of plastic into it.
After giving the plastic a chance to cool, the stamping machine ejects the disc and takes
another gulp of plastic.
In making a CD or DVD, another machine takes the newly stamped disc and aluminizes
it so that it has a shiny, mirror-like finish. To protect the shine, the disc is laminated with
a clear plastic cover that guards the mechanical pattern from chemical and physical abuse
(oxidation and scratches). Finally, another machine silk-screens a label on the disc. It is
packaged, shrink-wrapped, and sent off to a warehouse or store.
The DVD process differs in that a disc can have multiple layers. In the current process, a
separate master is made for each layer. The layers are stamped out separately, each only
half as thick as a complete disc. The two complete layers are then fastened together with a
special transparent glue.
In theory, a disc could be any size, and setting a standard is essentially an exercise in pragmatism. Size is related to playing time. The bigger the disc, the more data it holds, all else
being equal. If you want to store a lot of data, a big disc has its allure. On the other hand,
a platter the size of a wading pool would win favor with no one but plastics manufacturers.
You can also increase the capacity of a disk by shrinking the size of every stored bit of digital code, but the practical capabilities of technology limit how small you can make a bit.
In trying to craft a standard, engineers have to balance the convenience of small size, the
maximum practical storage density, and a target for the amount of information they need
to store.
In the late 1970s when Philips and Sony were developing the CD, the maximum practical
storage density of the then-current technology was about 150 megabytes per square inch.
The arbitrary design goal of about 70 minutes per disc side (enough room for Beethovens
Ninth) dictated about 650MB at the data rate selected (which itself was a tradeoff between
data requirements and sound quality). The result was that the design engineers found a
History
120-millimeter (thats about 4.6 inches) platter to be their ideal compromise. A nice,
round 100 millimeters was just too small for Beethoven.
For portable applications, the engineers came up with a smaller form factor for discs, 80
mm (about 3.1 inches). Once plated and given its protective plastic coating, either size of
CD is about 1.2 mm (about 0.05 inch) thick.
For DVDs, the same sizes have been retained. This expedient allows the same equipment
used to make CDs with only minor modification to stamp out DVDs. In addition, the
same drives that read and write can also use CDs, whether in your home entertainment
system or inside your computer.
The DVD medium differs from conventional CDs in that discs can use both sides for
recording data and have multiple layers on each side. (Although two-sided CDs are possible, they have been commercially produced only rarely.) As noted earlier, current technology fabricates each DVD in two pieces, each 0.6 mm thick, which are later cemented
together. Cemented back-to-back, the disc gains two sides. Cemented so the face of one
butts the back of the other, the disc gains two layers. The latter configuration allows you
to play both recorded surfaces without flipping the disc over. To shift between layers, the
DVD player needs only to refocus its laser beam. The process is fastit takes only millisecondsand, with adequate buffering, can be completely invisible to your computer
and its software. Theres no pause in the data streaming from the drive.
Under the DVD standards, eight possible types of disc are currently defined, depending
on the size, number of sides, and number of layers. Table 18.1 summarizes the various
DVD disc types and their storage capacities.
TA B L E 1 8 . 1
DVD Disc Types and Capacities
Name
Diameter
Capacity
Sides
Layers
DVD-1
8 cm
1.36GB
DVD-2
8 cm
2.48GB
DVD-3
8 cm
2.72GB
DVD-4
8 cm
4.95GB
DVD-5
12 cm
4.38GB
DVD-9
12 cm
7.95GB
DVD-10
12 cm
8.75GB
DVD-18
12 cm
15.9GB
Almost all discs made thus far conform to the DVD-5 standard. This format was tailored
to the needs of the motion picture and videocassette industry. It allows a standard
647
648
Hollywood-style movie to fit on a single disc. Unlike videocassettes, however, the movie
will have digital quality images and soundand not just stereo sound but full eight-channel surround. In that the cost of duplicating DVDs is a fraction of that of videocassettes,
the software industry will be urging you into the new medium as fast as it can. In that you
should easily be able to see and hear the difference, you shouldnt need too much encouragement.
The DVD format allows your disc drive to read the two layers in either of two ways,
determined when the disc is recorded. Parallel tracking path (PTP) tracks are read in parallel. Your drive reads both layers at nearly the same time, in parallel. In practice, the two
tracks may contain the same material in different formatsone might hold a standard
aspect-ratio image and the other a widescreen imageallowing you to switch between
them instantly. Opposite tracking path (OTP) tracks are read sequentially. The drive reads
one layer to its end, then reverses and reads the second layer in the opposite direction. To
minimize changeover time when the read head reverses direction, the drive starts reading
the bottom layer from the inside out and then switches to reading the inside layer from
the outside in. Discs that use OTP tracks are sometimes describe as reverse-spiral duallayer (RSDL) discs. Some disc producers prefer them because they offer a longer (nearly)
continuous playing time.
The first DVDs were mostly two-sided, single-layer DVD-10 discs, usually a motion picture in full-frame format on one side and widescreen on the other. One reason was that
drives capable of reading two-layer discs required more development, but all DVD drives
can handle two-sided discs, providing you physically flip the disc to access the other side.
A modern DVD drive distinguishes the separate layers of a multiple-layer disc by selective
focus. To read the bottom layer, it simple focuses on its pits. To read the upper layer, it
focuses through the semitransparent lower layer to the upper layer. The separation
between the two layers, about 20 to 70 microns, is sufficient for each layer to have a distinct-enough focus for them to be distinguished. The pattern in the layer that is out of
focus blurs together so that individual bits blur together and dont register.
Four-layer DVDs, which have two layers of data on each of their two sides, require a notquite-perfected manufacturing process that writes two layers simultaneously on each of
the two halves that are cemented to make the completed disc.
Materials
The recording materials used by read-only, recordable, and rewritable optical disc systems
are quite different because each system operates on different physical principles. Readonly media depend on fast, mechanical reproduction of discs. Recordable media require a
one-use material with a long life. Rewritable media require a recording material that can
be cycled between states so that it can be erased and reused.
History
Read-Only
Every Compact Disc is a three-layer sandwich. The bulk of the disc is a transparent polycarbonate plastic substrate called the carrier. It is made from the material injected into the
stamping pressthe machine smashes a warm glob of raw plastic flat and embosses the
spiral pattern of pits encoding the digital data into one of its surfaces. The aluminum
coating that makes the disc reflective (except for the data pits) is vapor-plated on the polycarbonate substrate to a thickness of about one-tenth micron. To protect the aluminum
from oxidation (which would darken it and make the pattern of pits difficult or impossible
to read), it is protected with a final layer of lacquer about 200 microns thick.
The laser in a disc drive reads from the reflective side of the disc. That is, the laser reads
through the thick carrier to the aluminized layer. The label of the disc gets silk-screened
onto the thin protective lacquer coating. Note that the coating is about one thousand
times thinner than the carrier on the reflective side. As a result, a compact disc is more
vulnerable to damage on the label side. Although a scratch here may not interfere with the
reading of the disc, it may allow air to come into contact with the aluminize layer and oxidize it, thus ruining the disc.
The construction of multilayer DVDs is similar, except the polycarbonate carrier is only
half as thick. Double-sided and double-layer discs cement two half-thickness discs
together. DVD-5 discs (single-sided, single-layer) cement a dummy half-thickness disc to
the back of the one active layer. In any case, in the center of the disk, the protective plastic
layer is replaced with a transparent plastic cement, which seals the aluminize layer on the
lower half of the disc and binds the upper half to it.
Compared to vinyl phonograph records or magnetic discs, CDs and DVDs offer a storage
medium that is long-lived and immune to most abuse. The protective clear plastic layer
resists physical tortures (in fact, Compact Discs and one-sided DVDs are more vulnerable
to scratches on their label side than the side that is scanned with the playback laser). The
data pits are sealed within layers of the disc itself and are never touched by anything other
than a light beam. These discs never wear out and acquire errors only when you abuse
them purposely or carelessly (for example, by scratching them against one another when
not storing them in their plastic jewel boxes). Although error correction prevents errors
from showing up in the data, a bad scratch can prevent a disc from being read at all. The
smaller features used by the DVD system for storing data are more vulnerable (one
scratch can wipe out more stored data), but the DVD system has a more robust error-correction system that more than compensates.
Recordable
Discs used in CD recorders differ in two ways from those used by conventional CD playersbesides being blank when they leave the factory. CD-R discs require a recordable
649
650
surface, something that the laser in the CD recorder can alter to write data. This surface
takes the form of an extra layer of dye on the CD-R disc. Recordable CDs also have a formatting spiral permanently stamped into each disc.
As with other CDs, a recordable disc has a protective bottom layer or carrier of clear
polycarbonate plastic that gives the disc its strength. A thin reflective layer is plated on the
polycarbonate to deflect the CD beam back so that it can be detected by the drive.
Between this reflective layer and the normal protective top lacquer layer of the disc, a
CD-R disc has a special dye layer. The dye is photoreactive and changes its reflectivity in
response to the high-power mode of the CD recorders laser. Figure 18.1 shows a crosssection of a typical CD-R disc.
FIGURE 18.1
Cross-section of
recordable CD media
using cyanine dye
(not to scale).
Silkscreened label
Protective lacquer
Reflective gold coating
Photoreactive green layer
Clear polycarbonate base
The CD-R medium records information by burning the dye layer in the disc. By increasing the power of the laser in the drive from 4 to 11 milliwatts, its beam heats the dye layer
to about 250 degrees (Celsius). At this temperature, the dye-layer melts and the carrier
expands to take its place, creating a nonreflective pit within the disc.
Colored Substrate CD-Rs, sometimes simply called Color CD-Rs, dye the polycarbonate substrate a pleasing tint. The most popular colors include black, blue, orange, purple, and
red. The dye has absolutely no effect on the digital code thats written to the disc, so it
does not affect the data or music stored on a CD.
The color of the substrate should not be confused with the color of the recording
medium. Three different compounds are commonly used for photoreactive dyes used by
CD-R discs, and each of these dyes has a characteristic color. These include green, gold,
and blue.
Green. The dye used in green CD-R discs is based on a cyanine compound. The
Taiyo Yuden company developed this photoreactive dye, which was used for the first
CD-R discs, including those used during the development of the CD-R standards.
Even now green CD-R discs are believed to be more forgiving laser power variations during the read and write processes. The green cyanine dye is believed to be
permanent enough to give green CD-R discs a useful life of about 75 years. In addition to Taiyo Yuden, several companies including Kodak, Ricoh, TDK, and
Verbatim make or have made green CD-R discs.
Gold. Gold CD-R discs used a phthalocyanine dye developed by Mitsui Toatsu
Chemicals. The chief advantage of gold over green discs is longer life, because the
History
dye is less sensitive to bleaching by ambient light. If it were on a dress or shirt, it

would be more colorfast. Gold CD-R discs are believed to have a useful life of about
100 years. Some people believe that gold discs are also better for high-speed (2 or
4) recording than are green discs. Mitsui Toatsu and Kodak manufacture most gold
CD-R discs.
Blue. The most recent of the CD shades is blue, a color that results from using cyanine with an alloyed silver substrate. The material is proprietary and patented by
Verbatim. According to some reports, it is more resistant to ultraviolet radiation
than either green or gold dyes and makes reliable discs with low block error rates.
Some manufacturers use multiple layers of dyes on their discs, sometimes even using two
different dyes. The multiple-layer CD-R discs are often described as green-green, goldgold, or green-gold, depending on the colors of the various layers.
Additionally, the reflective layers of recordable CDs also vary in color. They may be silver
or gold, which subtly alters the appearance of the dye. The basic reflective layer looks silver because it is made from aluminum. Gold discs have actual gold added to the mix to
increase the lifespan of the medium, which also makes the discs look bettergolden.
NOTE
At one time, manufacturers believed gold discs could endure for 100
years, compared to 30 years for mere silver. The makers of silver-based
discs now also claim a 100-year span. In that the medium itself is less
than 20 years old, no one knows for sure.
With current CD drives, there is no functional difference between the different CD-R
colors, be they a result of tinting the substrate, the dye, or the reflective layerall appear
the same hue to the monochromatic laser of a CD drive that glows at a wavelength of 780
nanometers. All the current CD-R materials reliably yield approximately the same degree
of detectable optical change. In the past, however, early CD-ROM readers had varying
sensitivities to the materials used in CD-R discs and would reliably read one color but not
another. Consequently, lore about one color or kind of disc being better overall than the
others arose. The major differences in discs result from mechanical tolerances (how well
the manufacturer maintains the perfect spiral of the track) than they do from the color of
the disc. Poorly performing discs are simply badly made, regardless of color.
No matter the dye used, recordable CD media are not as durable as commercially
stamped CDs. They require a greater degree of care. They are photosensitive, so you
should not expose them to direct sunlight or other strong light sources. The risk of damage increases with exposure. The label side of recordable CDs is often protected only by a
thin lacquer coating. This coating is susceptible to damage from solvents such as acetone
(finger nail polish remover) and alcohol. Many felt-tip markers use such solvents for their
651
652
inks, so you should never use them for marking on recordable CDs. The primary culprits
are so-called permanent markers, which you can usually identify by the strong aroma of
their solvents. Most fine-point pen-style markers use aqueous inks, which are generally
safe on CD surfaces. Do not use ballpoint pens, fountain pens, pencils, or other sharptipped markers on recordable CDs because they may scratch through the lacquer surface
and damage the data medium.
The safest means of labeling a recordable CD is using a label specifically made for the
recordable CD medium. Using other labels is not recommended because they may contain solvents that will attack the lacquer surface of the CD. Larger labels may also unbalance the disc and make reading it difficult for some CD players. In any case, once you put
a label on a recordable CD, do not attempt to remove it. Peeling off the label likely will
tear off the protective lacquer and damage the data medium.
Eraseable/Rewritable
The CD-RW system is based on phase-change media. That is, the reflective layer in the
disc is made from a material that changes in reflectivity depending on whether it is in an
amorphous or crystalline state. The most common medium is an alloy of antimony,
indium, silver, and tellurium, which has an overall silver color.
In its crystalline state, the medium has a reflectivity of about 15 to 25 percent. In its
amorphous state, the reflectivity falls a few percentenough to be reliably detected by the
laser-based disc-reading system.
A blank disc has all its reflective medium in its crystalline state. To record data, the drive
increases laser power to between 8 and 15 milliwatts and heats the medium to above its
500 to 700-degree (Celsius) melting point. The operating is straightforward and equivalent to the CD-R writing process, except for laser power.
CD-RW discs typically use a fine metal alloy for their phase-change medium, coating it
on the substrate in 8 to 12 layers. Some sources state that the phase-change media are
more stable than the one-use CD-R media and therefore can preserve your data for
decades longer.
Erasing the CD-RW disc complicates the recording process. To completely erase a disc
and restore it to its original crystalline state, the disc must be annealed. The reflective
layer is heated to about 200 degrees Celsius and held at that temperature while the material recrystallizes. The first CD-RW systems required that you entirely erase a disc to reuse
it, a time-consuming process. Although the heating and recrystallization took only a tiny
fraction of a second for each spot of data, erasing an entire disc required about 37 minutes. Most modern drives use on-the-fly erasing, selectively annealing small areas of the
disc with the laser at moderate power as they need to be reused. The annealed areas may
be rewritten with higher laser power.
Physical Format
Physical Format
The medium provides space to store information, but a complete storage system requires
something more. It needs a standardized method of organizing that information. That
organization may take several levelsa physical level and one or more logical levels. The
physical level or format determines how densely information fits onto the medium and
how it is accessed. The logical level determines how it is addressed and used.
The physical format of the Compact Disc and Digital Versatile Disc shares a characteristic
with hard disks thats the same with most disc-based mediathe information is arranged
around the disc in long arcs. But whereas hard disks and floppy disks make those arcs into
concentric tracks, the CD and DVD turn the arc into a long spiral. As a result, CDs dont
have tracks like hard disks. They have one continuous track. DVDs have a multiple tracks
but only one on each layer.
The difference is that hard disks are meant to be a random-access medium. CDs and
DVDs are designed for sequential access. The CD was optimized for storing music, which
usually is a continuous and uninterrupted stream of data. The DVD was primarily optimized for video, which also requires a continuous and uninterrupted stream. In scanning a
single spiral track, the reading mechanism never has to pause to jog to a new position. As
long as it follows the track, its where it needs to be.
CDs and DVDs also differ from hard and floppy disks in that they do not spin at a constant rate. Drives adjust the speed of the spin so that they read the same number of data
pits of the same size in any given period. Technically speaking, hard and floppy disks spin
with constant angular velocity. In any given period (say, a millisecond), the disk always
rotates by the same-size angle. In contrast, CDs and DVDs are a constant linear velocity
medium. In any given period, they spin the same linear length of track past the reading
mechanism. As a result, near the center of a CD or DVD, where the diameter is small, the
disc must spin faster to present the same length of track to the reading mechanism.
The actual linear velocity used by the original Compact Disc system was 1.2 meters per
second. As a result, the spin varied from about 400 revolutions per minute (RPM) at the
inner diameter to about 200 RPM at the outside edge. Higher-speed drives, discussed
later, spin at multiples of this rate. Under the basic Digital Versatile Disc standard, a single-layer disc spins at a constant linear velocity of 3.49 meters per second. As a result, a
discs spin rate must vary from 600 RPM at the outer edge to about 1200 RPM at the
inner edge of the recordable area. Multilayer discs spin somewhat faster, 3.68 meters per
second. Although DVD drives in computers may spin at faster rates, the DVD players you
connect to your video system are locked to this speed.
With each spin of a disc, the track advances outward from the center of the disc, a distance called the track pitch. In the case of CDs, the track pitch is 1.6 micrometers. The
653
654
individual pits on the track that encode data bits are at least 0.83 micrometers long. The
raw data rate at the basic speed of the CD system is 4.3218 megabits per second. In original form as an audio playback device, the actual data throughput was 150 kilobytes per
second (1.2 megabits per sec) after demodulating, decoding, error-correcting, and formatting the raw signal.
DVD storage is considerably denser. Most significantly, all measurements can be smaller
because the DVD system uses a shorter wavelength laser (650 to 680 nanometers compared to 780 nanometers used by CDs). The standard DVD track pitch measures 0.74
microns. On a single-layer disc, each data pit is about 0.40 micron long. On double-layer
discs, pits measure 0.44 micron. The basic data rate for DVD is 26.16 megabits per second. Subtract out the overhead, and the net rate is about 11.08 megabits per second.
Table 18.2 summarizes the physical format of CDs and DVDs.
TA B L E 1 8 . 2
CD and DVD Physical Formats
CD
Single-LayerDVD
Double-LayerDVD
Disc diameter
120 or 80 mm
120 or 80 mm
120 or 80 mm
Track pitch
1.6 nm
0.74 nm.
0.74 nm
Standard read velocity
1.2 m/sec
3.49 m/sec
3.68 m/sec
Minimum pit length
0.83 micron
0.4 micron
0.44 micron
Laser wavelength
780 nm
680 nm
680 nm
Basic read rate
4.3218Mbits/sec
26.16Mbits/sec
26.16Mbits/sec
Both CD-R and CD-RW blank discs (as well as their DVD equivalents) have a physical
format already laid down on them. Each has one continuous spiral track at the standard
pitch (1.6 microns for CDs). In addition, this track wobbles with a period of 22.05 kilohertz
and an excursion of 0.3 microns. That is, the smooth spiral actually acts as a superimposed
S-shape. The CD drive detects this wobble and uses it to control the rotation rate of the
disc. The wobble is frequency-modulated by a one kilohertz signal that provides an
absolute timebase for the drive.
Drives
The hundreds or thousands of megabytes on your CDs and DVDs would do you no good
without a means for reading them. That means is the optical disc drive. A CD drive or
CD-ROM reader gives you access only to CDs, be they audio or data discs. All DVD
drives can read both CD and DVD discs, although early drives often cannot read discs
written with CD-R and CD-RW drives.
Drives
With the price difference between CD and DVD drives disappearing, theres less reason
than ever to consider a CD drive. A DVD drive will do everything the CD drive does (but
likely better). Performance and compatibility are no longer issues. Consequently, in any
new computer, a DVD drive is the preferred option.
All optical disc drives in computers can play back audio CDs. The multimedia standards
require that CD-ROM drives on multimedia computers have both a front panel headphone jack and volume control. Of course, the multimedia PC standards date back to a
time when decent speakers connected to a computer were as rare as bagpipes in a symphony. Todays drives get around needing the extra jacks because all drives have an audio
connector on their rear panels for linking to your computers sound system. This connector provides analog audiothat is, the sound from a disc that has already been decoded
from its original digital format. Drives also have standard mass storage interface jacks
(typically AT Attachment or SCSI) to send digital data to your computer.
In theory, you can recover digital audio through this connection. Drives differ in their
ability to deliver usable audio in digital form, however, a feature called digital audio extraction (DAE). For example, if you wanted to capture a track from an audio CD to convert
into an MP3 file, you would need a drive that supports digital audio extraction. All current
drives support DAE. Older CD drives with the AT Attachment interface often had difficulty with DAE (if they supported the feature at all). You may encounter problems if you
try to use one of these older drives to capture audio.
Dedicated home DVD players have video jacks but the DVD drives in computers do not.
At one time, decent video playback in your computer required that the DVD reader plug
into an MPEG decoder that slid into an expansion slot and directly linked to your computers video system. As the microprocessors and video systems in computers gained
speed, however, the complex decoding job was readily assumed by software, a special
MPEG video codec.
Speed
When Philips and Sony originally propounded the Compact Disc system, memory was
expensive and computers were only for hobbyists. Players had to read audio data from
each disc at the same rate as it was to be played. Higher speeds were irrelevant. Once
computers adopted the CD, however, audio speed became a horrendous holdup. Audio
CD speed was lower than a modern floppy disk, and floppy disks were too slow.
Data, unlike music, need not abide by the constraints of real-time playback. It wasnt long
before computer engineers took advantage of a simple expedient to speed up the CD
drives connected to computers. They made the discs spin fasterand fasterand faster.
From humble beginnings of only twice as fast as ordinary audio CDs, they have revved up
to as high as 50 times faster.
655
656
Drive-makers describe the speed of their products using the x-factor. The base speed of an
audio CD drive is 1. A 2 drive spins twice as fast. A 12 drive spins 12 times faster, and
so on.
After about 12, the rotation rates for CDs become awesome. When reading from the
inside of the disk, a 12 rate amounts to about 3600 RPM. At the outer edge, the disc
spins at half that rate. Higher speed factors (for example, 20, 36, or 50) are comparably faster. Not only is spinning the disc at these rates a challenge, altering the spin speed
from the inner to outer reaches of the disc is tricky and time consuming. The need to
change speeds can really slow down random access.
In data applications, a constant data rate is unnecessary. With that gem of wisdom in
mind, drive-makers developed CD players with a constant spin rate that, with a logic
peculiar to the computer market, are known as variable-speed drives. Although the spin rate
is constant, the data rate the drives deliver is not. It is twice as high at the outer edge of
the disc as it is at the inner edge. The x-factor of the drive varies with the data rate.
As any cynic would expect, drive manufacturers invariably quote the fastest speed that variable-speed drives can possibly deliver. In practical use, drives almost never achieve this
rate. Because these multispeed drives actually spin their discs at a constant rate (that is,
at a constant angular velocity), at smaller diameters, the data rate (and the x-factor) is
lower than at larger diameters near the outside edge of the disc. But the data on optical
discs starts at the inside of the track spiral and continues to the outside. Consequently,
variable-speed drives read (or write) at their top speed only at the very end of a completely full disc. Most discs arent filled to capacity, so the highest speed areas of the disc
arent often usedand variable speed drives rarely achieve the high speed at which they
are rated. A few more reputable computer-makers describe variable-speed drives by listing
both their slowest and fastest speed, such as 12/24.
On the other hand, variable-speed drives have an advantage in random access. Because
they spin your discs at a constant rate, you dont have to wait for the drive to alter its
speed to match the rate required at a particular place on the disc.
Note, too, that high-speed drives impose an access penalty. Drives with faster speed ratings require more time to spin your discs up to the proper speed. That means you have to
wait longer after you slide a disc into the drive before you can read it. The problem is particularly severe with portable computers, which stop their discs from spinning after a
minute or so to conserve power. Each time the drive powers down, you have to wait for it
to spin the disc back up to speed before you can read from it.
The x-factor for DVD drives is not the same as for CD drives. As with the CD, the DVD
speed ratings are pegged to the base data rate of the original video-only DVD system.
The linear velocity of the DVD system is about 2.98 times faster than the linear velocity
Drives
of the CD. But drive electronics dont care about linear speed. They lock onto the data
rate. The data format of the CD has larger features than those of DVDs. The minimum
pit length, for example, is 0.83 microns for CDs and only 0.4 for DVDs. For a given data
rate, the CD must spin about twice as fast as a DVD to produce the same data rate, the
26.16Mbps basic read rate of DVDs. Add the increased linear velocity to the change
required by feature size and read rate, and the result is that a 1 DVD drive best reads
data at about six times the base CD rate. Consequently, every DVD is worth about six
from a CD drive. Table 18.3 gives a comparison of some common CD and DVD
speeds (assuming a constant data rate).
TA B L E 1 8 . 3
Comparison of CD and DVD Drive Speed Factors
CD Factor
DVD Factor
Raw Data Rate
0.17
4.32Mbits/sec
0.33
8.64Mbits/sec
0.67
17.3Mbits/sec
26.2Mbits/sec
12
52.4Mbits/sec
18
78.6Mbits/sec
20
3.3
86.4Mbits/sec
24
105Mbits/sec
30
131Mbits/sec
36
157Mbits/sec
40
6.7
173Mbits/sec
Because the coding system used by DVDs is more efficient than that used by CDs, a
DVD drive extracts more information at a given raw data rate. The user data rate from a
1 DVD drive reading a DVD is about nine times that of a 1 CD drive reading a CD.
In general, a faster drive is a better driveproviding you can put up with the extended
spin-up times of really quick disc drives. Some software makes particular speed demands
from CD drives. The minimum speed requirement usually is in the range of 4 to 6. All
CD drives in new computers deliver at least 12, so you should encounter no problem
with software speed compatibility. Similarly, the base DVD rate is fast enough for most
applications, although you may prefer the quicker load times afforded by a 4 or faster
DVD drive.
When you play audio from CDs or movies from DVDs, high drive speeds and x-factors
make no difference. A faster drive will not produce a higher-quality signal from these
657
658
isosynchronous sources. Only when you read computer data from disc do the x-factors
matter. For example, play a CD so you can listen to it, and your drive will operate at 1.
When you attempt to extract digital data from the same audio CD to create an MP3 file,
however, the drive may shift to a higher speedif your computer can process the data fast
enough. Extraction need not occur in real time.
Access Time
Compared to magnetic hard disks, all optical disc readers are laggardly beasts. Mass is the
reason. The read head of an optical drive is substantially more massive than the flyweight
mechanisms used in hard disks. Instead of a delicate read/write head, the CD or DVD
drive has a large optical assembly that typically moves on a track. The assembly has more
mass to move, which translates into a longer wait for the head to settle into place. Optical
drives consequently have hefty access timeswhere the typical hard disk drive now has an
average access time of 9 milliseconds or less, a quick optical drive is about one-tenth the
speed (with an average access time about 90 to 100 milliseconds).
As noted earlier, the constant linear velocity system used by some drives also slows the
access speed. Because the spin rate of the disc platter varies depending on how far the
read/write head is located from the center of the disc, as the head moves from track to
track, the spin rate of the disc changes. With music, which is normally played sequentially,
thats no problem. The speed difference between tracks is tiny, and the drive can quickly
adjust for it. Make the optical drive into a random-access mechanism, and suddenly speed
changes become a big issue. The drive might have to move its head from the innermost to
outermost track, requiring a drastic speed change. The inertia of the disc spin guarantees
a wait while the disc spins up or down.
Controls
Nearly all optical disc drives have an eject button on the front panel. Pressing this button
causes the drive to spin down the disc inside the drive (if there is one) and then slide out
the tray or pop out the caddy should one be used. Pressing the button again slides the tray
back in. Usually pressing on the tray open will also trigger the drive to slide it all the way
in.
The various Multimedia PC (MPC) specifications require a volume control on the front
panel of any CD-ROM drive you have in your multimedia computer. This control is useful if you decide to use your drive for playing back music while you work. You can plug
headphones into the jack on the front of the drive (also required by the MPC standards)
and use the volume control to adjust the loudness of the playback independent of the CD
control software you run on your computer. The front panel volume control usually does
not control the output on the audio connector on the rear of the drive.
Standards
Some CD and DVD drives have extended control panels, usually with the standard
motion controls as found on dedicated disc players: stop, play, fast forward, rewind, track
forward, and track back. Although these controls are not required, they can be handy
when you use the DVD player in your computer for playing video through an external
monitor.
With the exception of the volume control, all the front panel control functions (including
eject) can be operated through suitable software.
Compatibility
DVD drives are required to be able to play back audio CDs made under the Red Book
standard. Although as a digital system that can read CD-ROM data, DVD could be compatible with any CD, this is not always the case. Most DVD systems cannot handle interactive Green Book CDs. The dye used in many CD-R media is invisible to DVD
wavelengths, so early DVD drives may be unable to read the CD-R discs you make.
Drive-makers have adopted a number of strategies to enable their newer DVD drives to
read CD-Rs properly. Of course, different CD-R media use different dyes, so you might
find your DVD drive works with some CD-Rs and not with others. CD-RW media complicate matters because the medium is not as reflective as that used by prerecorded CDs
and DVDs. Manufacturers have developed a MultiRead label to assure you that the
drive will read CD-ROM, CD-R, and CD-RW media. DVD drives are not required to be
able to read CD-Video disc (White Book), but most canit takes no great technical feat
because the DVD drive can read the CD medium, and the MPEG-2 circuitry in the DVD
drive also handles the MPEG-1 of CD-Video. DVD players can read data from Enhanced
CDs made under the Blue Book standard, including music CDs.
Standards
Because both the Compact Disc and Digital Versatile Disc were conceived as distribution
media for softwarein this case, software meaning audio for the CD and video for the
DVDstandardization has been of the utmost importance. Without hard and fast standards, you cannot be sure that the disc you buy will play back in your drive. Consequently,
every format of disc has an official standard that guides both drive- and disc-makers so
their products can be compatible.
The Sony-Philips standards are published in books that are commonly referred to by the
colors of their covers. This small rainbow covers all currently recognized CD standards
and was enlarged to embrace the new Super Audio CD. The DVD standards, promulgated by the DVD Forum and published by Toshiba, wear more prosaic identifying names
based on a simple letter designation. The five standards recognized at the time of this
writing include those for audio, video, read-only data, write-once data, and read/write
data. Table 18.4 summarizes these standards.
659
660
TA B L E 1 8 . 4
CD and DVD Standards
Book
Name
Application
Standard
Book A
DVD-ROM
Data distribution
Book B
DVD-Video
Consumer video playback
Book C
DVD-Audio
Consumer audio playback
Book D
DVD-WO (DVD-R)
Write-one data storage
Book E
DVD-E (DVD-RW)
Rewritable data storage
Red Book
CD-DA
Consumer audio playback
ISO-10149
Scarlet Book
SA-CD
Super Audio CD
Pending
Orange Book
CD-R, CD-RW
Writable and rewritable

datastorage
Yellow Book
CD-ROM
Data distribution
Green Book
CD-i
Interactive CDs
Blue Book
CD-Extra, CD-Plus
Stamped multisession CDs
White Book
CD-V
Consumer video playback
ISO 10149:1989 (E)
The primary DVD standard is maintained by Toshiba Corporation. It is freely available

but hardly free. Anyone can obtain a copy from the DVD Forum at http://www.dvdforum.org after paying a $5,000 fee and signing a nondisclosure agreement.
Each of the most widely used of the CD and DVD standards is separately discussed in the
following sections.
Note that the original format for CDs was audio. The data format, CD-ROM, was separately developed around the audio parameters. In DVD, the original format was DVDROM, with the first application, DVD-Video, being a subset of the DVD-ROM format.
CD-DA
CD-DA, which stands for Compact Disc, Digital Audio, is standardized in the Red Book.
It was the original Compact Disc application, storing audio information in digital form.
The name Red Book refers to the international standard (ISO 10149), which was published
as a book with a red cover and specifies the digitization and sampling rate details, including the data-transfer rate and the exact type of pulse code modulation used.
Under the standard, a CD-DA disc holds up to 74 minutes of stereo music with a range
equivalent to todays FM radio stationthe high end goes just beyond 15KHz (depending
on the filtering in the playback device); at the low frequency end, nearly to DC, zero
hertz. The system stores audio data with a resolution of 16 bits, so each analog audio level
Standards
is quantified as one of 65,536 levels. With linear encoding, thats sufficient for a dynamic
range of 96 decibels. To accommodate an upper frequency limit of 15KHz with adequate
roll-off for practical antialiasing filters, the system uses a sampling rate of 44.1KHz.
Under the Red Book standard, this digital data is restructured into 24-byte blocks,
arranged as six samples, each of a pair of stereophonic channels (each of which has a depth
of 16 bits). These 24 bytes are encoded along with control and subchannel information
into the 588 optical bits of a small frame, each of which stores about 136 microseconds of
music. Ninety-eight of these small frames are grouped together in a large frame, and 75
large frames make one second of recorded sound.
In CD-DA systems, the large frame lacks the sync field, header, and error-correction code
used in CD-ROM storage, discussed later. Instead, the error-correction and control information is encoded in the small frames. The necessary information to identify each large
frame is spread through all 98 bits of subchannel Q in a given large frame. One bit of the
subchannel Q data is drawn from each small frame.
From the subchannel Q data, a sector is identified by its ordinary playing time location (in
minutes, seconds, and frame from the beginning of the disc). The 98 bits of the subchannel Q signal spread across the large frame is structured into nine separate parts: a two-bit
synchronization field; a four-bit address field to identify the format of the subchannel Q
data; a four-bit control field with more data about the format; an eight-bit track number;
an eight-bit index number; a 24-bit address, counting up from the beginning of the track
(counting down from the beginning of the track in the pre-gap area); eight reserved bits; a
24-bit absolute address from the start of the disc; and 16 bits of error-correction code. At
least nine of ten consecutive large frames must have their subchannel Q signals in this
format.
In the remaining large sectors, two more subchannel Q formats are optional. If used, they
must occur in at least 1 out of 100 consecutive large frames. One is a disc catalog number
that remains unchanged for the duration of the disc; the other is a special recording code
that is specific and unchanging to each track.
Super Audio CD
The Scarlet Book covers the Super Audio CD, the DVD/CD hybrid developed jointly by
Philips and Sony to compete with DVD-Audio. The format specifications were first
released in March 1999.
The standard allows for three formsa single-layer disc that only supports Super Audio, a
disc with two Super Audio layers, and a hybrid disc that mates a Super Audio layer with a
conventional CD-A layer for compatibility with old equipment. A hybrid disc works and
sounds like a conventional CD in a conventional CD drive but plays back Super Audio in
an SA drive.
661
662
To store all the data necessary for Super Audio, all Super Audio layers use basic DVD
technology. The data on these layers takes the form of Sonys proprietary Direct Stream
Digital (DSD) encoding technology (discussed later). The CD layer uses conventional CD
technology (16-bit audio sampled at 44.1KHz).
The DSD encoding system provides frequency response to 100KHz and a dynamic range
in excess of 120 dB. Although few people can hear anything beyond 20KHz, the highest
quality analog master tapes recorded at a speed of 30 inches per second often have information out to 50KHz. The high-frequency range of DSD will finally put to rest complaints that digital doesnt sound as good as analog.
The DSD system uses one-bit sampling at a 2.8224MHz rate, the same as used in a highquality pulse-code modulation (PCM) encoding system that uses 64 oversampling.
Instead of converting the resulting code to PCM, the DSD system records the one-bit
sample values. Delta sigma modulation determines the one-bit value for each samplethat
is, the sample represents the sum (sigma) of the changes (delta) of the signal. The system
maintains a running total of the bits representing the strength of the analog waveform. At
each sampling interval, it compares the present value with the previous value. If the new
value is higher than the previous value, the system adds a logical 1 to the code stream. If
the current value is lower than the previous value, the system adds a logical 0 to the
code stream. If a value does not change, it is nevertheless evaluated and will produce an
alternating pattern of 1s and 0s that correct one another.
The overall result is that a rising (positive) waveform will be a dense string of 1s, and a
full falling (negative) waveform will be a dense string of 0s. Consequently, this form of
modulation is often termed Pulse Density Modulation (PDM). As with all digital signals, the
PDM output is resistant to noise and distortionand alteration to the signal smaller than
an entire pulse will be ignored in processing and excluded from the reconstructed signal.
Unlike coded digital signals, however, PDM signals (and the signals of a similar digital
modulation system, called Pulse Width Modulation) are closely allied to their analog equivalent. The pulses in the signal mimic the analog signal in strength and frequency. In fact, a
simple low-pass filter can convert the digital signals into analog form.
In the case of Sonys DSD, the simple analog conversion isnt very good. Digital artifacts
make the simple conversion noisy. The Sony system uses high-order filtering to move the
noise out of the audio band.
The CD-compatible layer of a Super Audio CD is created during mastering through a
process Sony calls super bit mapping direct downstream conversion. It is essentially the same
process used in mastering ordinary audio CDs from 64 oversampled masters.
Although the standard Super Audio CD format uses two audio channels of DSD signals,
the specification allows for up to six such channels for future applications. In addition,
Standards
storage space on the outside of the audio signal area is reserved for text, graphics, or
videosfor example, to accompany the audio presentation.
The Super Audio CD also includes provisions for watermarking (invisibly indicating the
origin of the media) discs through coding embedded in the wobble of the track. This feature can be used for tracking the origins of a particular disc (for example, for controlling
disc piracy) or in copy-protection schemes.
According to Sony, a hybrid Super Audio CD will play in any driveCD, DVD, or SACD. In a CD drive, it will deliver better than CD quality thanks to the super bit mapping
downstream conversion. Current DVD drives will also play the discs with CD quality.
The true beauty of the disc will come out only on special SA-CD players or a future generation of DVD drives that have the proper playback algorithms.
CD-ROM
The Yellow Book, first introduced in 1984, describes the data format standards for CDROM discs and includes CD-XA, which adds compressed audio information to other CDROM data. The Yellow Book divides CD-ROM operation into two modes. Mode 1 is
meant for ordinary computer data. Mode 2 handles compressed audio and video data.
Because Yellow Book discs can contain audio, video, and data in their two modes, they are
often termed mixed mode discs. Yellow Book is the standard that first enabled multimedia
CDs. It is now an internationally recognized standard as ISO 10149:1989 (E).
As the full name implies, Compact Disc Read-Only Memory is fundamentally an adaptation of the Compact Disc for storing digital informationRock and Roll comes to computer storage. Contrary to the implications of the name, however, you can write to
CD-ROM discs with your computer, providing you buy the right (which means expensive)
equipment. For most applications, however, the CD-ROM is true to its designationit
delivers data from elsewhere into your computer. Once a CD-ROM disc is pressed, the
data it holds cannot be altered. Its pits are present for eternity.
In the beginning, CD-ROM was an entity into itself, a storage medium that mimicked
other mass storage devices. It used its own storage format. The kind of data that the CDROM lent itself to was unlike that of other storage systems, however. The CD-ROM supplied an excellent means for distributing sounds and images for multimedia systems,
consequently engineers adapted its storage format to better suit a mixture of data types.
The original CD-ROM format was extended to cover these additional kinds of data with
its Extended Architecture. The result was the Yellow Book standard.
Logical Format
As with other disc media, the CDs capacity is divided into short segments called sectors. In
the CD-ROM realm, however, these sectors are also called large frames and are the basic
unit of addressing. Because of the long spiral track, the number of sectors or large frames
663
664
per track is meaninglessits simply the total number of sectors on the drive. The number
varies but can reach about 315,000 (for example, for 74 minutes of music) or about
340,000 for newer, 80-minute discs.
Large frames define the physical format of a Compact Disc and are defined by the CDROM media standards to contain 2352 bytes. (Other configurations can put 2048, 2052,
2056, 2324, 2332, 2340, or 2352 bytes in a large frame.) The CD-ROM media standards
allow for several data formats within each large frame, dependent on the application for
which the CD-ROM is meant. In simple data-storage applications, Data Mode 1, 2048
bytes in a 2352-byte large frame actually store data. The remaining 304 bytes are divided
among a synchronization field (12 bytes), a sector address tag field (4 bytes), and an auxiliary field (288 bytes). In Data Mode 2, which was designed for less critical applications not
requiring heavy-duty error correction, some of the bytes in the auxiliary field may also be
used for data storage, providing 2336 bytes of useful storage in each large frame. Other
storage systems allocate storage bytes differently but in the same large-frame structure.
The four bytes of the sector address tag field identify each large frame unambiguously.
The identification method hints at the musical origins of the CD-ROM systemeach
large frame bears an identification by minute, second, and frame, which corresponds to
the playing time of a musical disc. One byte each is provided for storing the minute count,
second count, and frame count in binary coded decimal (BCD) form. BCD storage allows
up to 100 values per byte, more than enough to encode 75 frames per second, 60 seconds
per minute, and the 74 minute maximum playing time of a Compact Disc (as audio storage). The fourth byte is a flag that indicates the data storage mode of the frame.
In Data Mode 1, the auxiliary field is used for error detection and correction. The first
four bytes of the field store a primary error-detection code and are followed by eight bytes
of zeros. The last 276 hold a layered error-correction code. This layered code is sufficient
for detecting and repairing multiple-bit errors in the data field.
Extended architecture rearranges the byte assignment of these data modes to suit multisession applications. In XA Mode 2 Form 1, the 12 bytes of sync and four of header are
followed by an eight-byte subheader that helps identify the contents of the data bytes,
2048 of which follow. The frame ends with an auxiliary field storing four bytes of errordetection code and 276 bytes of error-correction code. In XA Mode 2 Form 2, the auxiliary field shrinks to four bytes; the leftover bytes extending the data contents to 2324
bytes.
Data Coding
The bytes of the large frame do not directly correspond to the bit-pattern of pits that are
blasted into the surface of the CD-ROM. Much as hard discs use different forms of modulation to optimize both the capacity and integrity of their storage, the Compact Disc uses
Standards
a special data-to-optical translation code. Circuitry inside the Compact Disc system converts the data stream of a large frame into a bit-pattern made from 98 small frames.
Each small frame stores 24 bytes of data (thus 98 of them equal a 2352-byte large frame)
but consists of 588 optical bits. Besides the main data channel, each small frame includes
an invisible data byte called the subchannel and its own error-correction code. Each byte of
this information is translated into 14 bits of optical code. To these 14 bits, the signal-processing circuitry adds three merging bits, the values of which are chosen to minimize the
low-frequency content of the signal and optimize the performance of the phase-lock loop
circuit used in recovering data from the disc.
The optical bits of a small frame are functionally divided into four sections. The first 27
bits comprise a synchronization pattern. They are followed by the byte of subchannel
data, which is translated into 17 bits (14-bit data code plus three merging bits). Next
comes the 24 data bytes (translated in 408 bits), followed by eight bytes of error-correction code (translated into 136 bits).
The subchannel byte actually encodes eight separate subchannels, designated with letters
P through W. Each bit has its own function. For example, the P subchannel is a flag used
to control audio muting. The Q subchannel is used to identify large frames in audio
recording.
As with a hard disk, this deep structure is hidden from your normal application software.
The only concern of your application is to determine how the 2048 (or so) bytes of active
storage in each large frame are divided up and used. The CD-ROM drive translates the
block requests made by the SCSI interface (or other interface) into the correct values in
the synchronization field to find data.
Sessions
A session is a single recorded segment on a CD and may comprise multiple tracks. The session is normally recorded all at once in a single session, hence the name. Under the
Orange Book standard, a session can contain data, audio, or images.
On the disc, each session begins with a lead-in, which provides space for a table of contents for the session. The lead-in length is fixed at 4500 sectors, equivalent to one minute
of audio or 9MB of data. When you start writing a session, the lead-in is left blank and is
filled in only when you close the session.
At the end of the session on the disc is a lead-out, which contains no data but signals to
the CD player that it has reached the end of the active data area. The first lead-in on a
disc measures 6750 sectors long, the equivalent of 1.5 minutes of audio or 13MB of data.
Any subsequent lead-outs on a single disc last for 2250 sectors (half a minute, or about
4MB of data).
665
666
Addressing
The basic addressing scheme of the Compact Disc is the track, but CD tracks are not the
same as hard disk tracks. Instead of indicating a head position or cylinder, the track on a
CD is a logical structure akin to the individual tracks or cuts on a phonograph record.
A single Compact Disc is organized as one of up to 99 tracks. Although a single CD can
accommodate a mix of audio, video, and digital data, each track must be purely one of the
three. Consequently, a disc mixing audio, video, and data would need to have at least three
tracks.
The tracks on a disc are contiguous and sequentially numbered, although the first track
containing information may have a value greater than one. Each track consists of at least
300 large frames (thats four seconds of audio playing time). Part of each track is a transition area called pre-gap and post-gap areas (for data discs) or pause areas (for audio discs).
Each disc has a lead-in area and a lead-out area corresponding to the lead-in and lead-out
of phonograph records. The lead-in area is designated track zero, and the lead-out area is
track 0AA(hex). Neither is reported as part of the capacity of the disc, although the subchannel of the lead-in contains the table of contents of the disc. The table of contents lists
every track and its address (given in the format of minutes, seconds, and frames).
Tracks are subdivided into up to 99 indices by values encoded in the subchannel byte of
nine out of ten small frames. An index is a point of reference thats internal to the track.
The number and location of each index is not stored in the table of contents. The pre-gap
area is assigned an index value of zero.
Capacity
The nominal maximum capacity of a CD amounts to 74 minutes of music recording time
or about 650MB when used for storing data. With 80-minute discs, the data capacity
extends to about 700MB. These capacities are only approximate, however. A number of
factors control the total capacity of a given disc. For example, mass-produced audio CDs
sometimes contain more than 74 minutes of sound because disc-makers can cram more
onto each disc by squeezing the track on the glass master disc into a tighter, longer spiral.
This technique is the secret to extending the playing time 80-minute discs.
The special CDs that you can write on with your computer cannot benefit from this
tighter-track strategy because their spiral is put in place when the discs are manufactured.
The standard formats yield four capacity levels on two different sizes of disc, as discussed
in the upcoming CD-R section. In any case, these numbers represent the maximum storage capacity of a recordable CD. Nearly anything you do when making a CD cuts into
that capacity.
Standards
Format
The Yellow Book describes how to put information on a CD-ROM disc. It does not, however, define how to organize that data into files. In the DOS world, two file standards
have been popular. The first was called High Sierra format. Later this format was
upgraded to the current standard, the ISO 9660 specification.
The only practical difference between these two standards is that the driver software supplied with some CD-ROM players, particularly older ones, meant for use with High
Sierraformatted discs may not recognize ISO 9660 discs. Youre likely to get an error
message that says something like Disc not High Sierra. The problem is that the old version of the Microsoft CD-ROM extensionsthe driver that adapts your CD-ROM player
to work with DOScannot recognize ISO 9660 discs.
To meld CD-ROM technology with DOS, Microsoft Corporation created a standard bit
of operating code to add onto DOS to make the players work. These are called the DOS
CD-ROM extensions, and several versions have been written. The CD-ROM extensions
before Version 2.0 exhibit the incompatibility problem between High Sierra and ISO
9660, noted earlier. The solution is to buy a software upgrade to the CD-ROM extensions
that came with your CD-ROM player from the vendor who sold you the equipment. A
better solution is to avoid the problem and ensure any CD-ROM player you purchase
comes with Version 2.0 or later of the Microsoft CD-ROM extensions.
ISO 9660 embraces all forms of data youre likely to use with your computer. Compatible
discs can hold files for data as well as audio and video information.
For Windows 95, Microsoft created another set of extensions to ISO 9660. Called the
Joliet CD ROM Recording Specification, these extensions add support for longer file names
but to 128 characters instead of the 255-character maximum of Windows 95as well as
nesting of directories beyond eight levels, allowing directory names to use extensions, and
broadening the character set. To maintain compatibility with ISO 9660, the extra Joliet
data must fit in a 240-character limit, foreclosing on the possibility of encoding all
Windows 95 directory data.
CD-Recordable
The Orange Book is the official tome that describes the needs and standards for Compact
Disc-Recordable (CD-R) systems. It turns the otherwise read-only medium into a writeonce medium so that you can make your own CDs. Introduced in 1992, the Orange Book
standard introduced multisession technology. A multisession disc can contain blocks of
data written at different times (sessions). Each session has its own lead-in track and table
of contents.
667
668
Developed jointly by Philips and Sony (sound familiar?), the Orange Book defines both
the physical structure of recordable CDs and how various parts of the data area on the
disc must be used. These include the program area, which holds the actual data the disc is
meant to store; the program memory area, which records the track information for the
whole disc and all the sessions it contains; the lead-in and lead-out areas; and a power calibration area thats used to calibrate the power of the record laser.
The nature of the CD-ROM medium and the operation of CD recorders make the creation and writing of a CD-ROM a more complex operation than simply copying files to a
hard disk drive. Because CD-ROMs are essentially sequentially recorded media, the CD
recorder wants to receive data and write it to disc as a continuous stream. In most CD
recorders, the stream of data cannot be interrupted once it starts. An interruption in the
data flow can result in an error in recording. Moreover, to obtain the highest capacity possible from a given CD, you want to limit the number of sessions into which you divide the
disc. As noted earlier, each session steals at least 13MB from disc capacity for the overhead
of the sessions lead-in and lead-out.
If your system cannot supply information to your CD recorder fast enough, the result is a
buffer underrun error. When you see such an error message on your screen, it means your
CD recorder has exhausted the software buffer and run out of data to write to the disc.
You can prevent this error by increasing the size of the buffer if your software allows it.
Or you can better prepare your files for transfer to CD. In particular, build a CD image
on a hard disk that can be copied on the fly to the CD.
The best strategy is to give over your computer to the CD-writing process, unloading any
TSR programs, background processes, or additional tasks in a multitasking system.
Screensavers, pop-up reminders, and in-coming communications (your modem answering
the phone for data or a fax) can interrupt your CD session and cause you to waste your
time, a session, or an entire disc.
Your system needs to be able to find the files it needs to copy to your CD-ROM as efficiently as possible. Copying multiple short files can be a challenge, particularly if your
hard disk is older and slower or fragmented. CD recordermakers recommend discs with
access times faster than about 19 milliseconds. An AV-style hard disk is preferable because
such drives are designed for the smooth, continuous transfer of data and dont interrupt
the flow with housekeeping functions, such a thermal calibration. Youll also want to be
sure your files are not fragmented before transferring them to CD. Run your defrag utility
before writing to your CD.
Depending on the manufacturer of your CD recorder and the software accompanying it,
you may have a choice of more than one mode for copying data to CD. In general, you
have two choices: building a CD image on your hard disk and copying that image intact to
your CD. Some manufacturers call this process writing on the fly. From a hardware
Standards
standpoint, this is the easiest for your system and CD recorder to cope with because the
disc image is already in the form of a single huge file with all the directory structures
needed for the final CD in their proper places. Your system needs to only ready your hard
disk and send a steady stream of data to the CD recorder.
The alternative method is to create the CD structure in its final form on the CD itself.
Some manufacturers call this writing a virtual image. In making a CD by this method,
your CD recorders software must follow a script or database to find which files it should
include on the disc and locate the files on your hard disk. The program must allocate the
space on your CD, dividing it into sectors and tracks, while at the same time reading the
hard disk and transferring the data to the CD.
Capacity Issues
With a read-only medium, you normally dont have to concern yourself with the issue of
storage capacity. Thats for the disc-maker to worry aboutthe publisher has to be sure
everything fits. With about 650 megabytes of room on the typical CD and many products
requiring only a few megabytes for code, the big problem for publishers is finding enough
stuff to put on the disc so that you think youre getting your moneys worth.
The advent of recordable CDs changes things entirely. With CDs offering convenient
long-term storage for important files such as graphic archives, youll be sorely tempted to
fill your CDs to the brim. Youll need to plan ahead to make all your files fit.
CD-ROMs have substantial overhead that cuts into their available capacity. If you dont
plan for this overhead, you may be surprised when your files dont fit.
Raw Capacity
CD-ROM capacities are measured in minutes, seconds, and sectors, based on the audio
format from which engineers derived the medium. Recordable CDs come in five capacities: 18- and 21-minute discs are 80 millimeters in diameter; 63-, 74-, and 80-minute discs
are 120 millimeters in diameter.
Two kinds of file overhead affect the number of bytes available on a given recordable CD,
which can actually be used for storage. One is familiar from other mass storage devices,
resulting from the need to allocate data in fixed-size blocks. The other results from the
format structure required by the CD standards.
Logical Block Padding

As with most hard and floppy discs, CD-ROMs allocate their storage in increments called
logical blocks. Although logical block sizes of 512, 1024, and 2048 bytes are possible with
todays CD drives, only the 2048-byte logical block format is in wide use. If a file is
smaller than a logical block, it is padded out to fill a logical block. If a file is larger than
one logical block, it fills all its logical blocks except the last, which is then padded out to
be completely filled. As a result of this allocation method, all files except those that are an
669
670
exact multiple of the logical block size require more disc space than their actual size. In
addition, all directories on a CD require at least one logical block of storage.
Format Overhead
In addition to the block-based overhead shared with most mass storage devices, CDROMs have their own format overhead that is unique to the CD system. These are remnants of the audio origins of the CD medium.
Because audio CDs require lead-in and lead-out tracks, the Yellow Book standard for CDROM makes a similar allowance. The specifications require that data on a CD-ROM
begin after a two-second pause, followed by a lead-in track 6500-sectors long.
Consequently, the first two seconds of storage space and the lead-in area on a CD are not
usable for data. These two seconds comprise a total of 150 sectors, each holding 2048
bytes, which trims the capacity of the disc by 307,200 bytes. The 6500-sector lead-in consumes another 13,312,000 bytes. The lead-out gap at the end of a storage session and the
pre-gap that allows for a subsequent session consume another 4650 sectors or 9,523,200
bytes.
The ISO 9660 file structure also eats away at the total disc capacity. The standard
reserved the first 16 sectors of the data areathats 32,768 bytesfor system use. Various
elements of the disc format also swallow up space. The root file, primary volume descriptor, and volume descriptor set terminator each require a minimum of one sector. The path
tables require at least two sectors. The required elements consequently take another five
sectors or 10,120 bytes of space. Discs with complex file structures may exceed these minima and lose further storage space.
The more sessions you divide a given CD into, the less space that will be available for
your data. Each session on a multisession CD requires its own lead-in. Consequently, each
session requires at least 13MB of space in addition to the file structure overhead.
Operation
Creating a CD is a complete process. The drive doesnt just copy down data blocks as
your computer pushes them out. Every disc, even every session, requires its own control
areas to be written to the disc. Your CD-R drive doesnt know enough to handle these
processes automatically because the disc data structure depends on your data and your
intentions. Your CD-R drive cannot fathom either of these. The job falls to the software
you use to create your CD-R discs.
Your CD-creation software organizes the data for your disc. As it sends the information to
your CD-R drive, it also adds the control information required for making the proper disc
format. After youve completed writing to your disc, the software fixates the disc so that it
can be played. The last job is left to youlabeling the disc so you can identify the one
you need from a stack more chaotic than the pot of an all-night poker game.
Standards
Speed
As with ordinary CD-ROM, the speed of CD-R drives is the transfer rate of the drive
measured in multiples of the basic audio CD speed, 150KBps. The very first CD
recorders operated at 1 speed, and each new generation has doubled that speed. The
fastest drives currently operate at 4, although technical innovation can increase that just
as it has improved basic CD speed.
Most CD recorders have two speed ratingsone for writing and one for reading. The
writing speed is invariably the same or less than the reading speed. Advertisements usually
describe drives using two numbers, the writing speed (lower number) first. The most
common speed combinations are 11, single-speed read and write; 12, single-speed write
and double-speed read; 22 double-speed writing and reading; 24 double-speed writing
and quadruple-speed reading; and 44 quadruple-speed in both writing and reading.
How fast a CD recorder writes is only one factor in determining how long making one or
more CDs will take. Other variables include your system, writing mode (whether you try
to put files together for a CD session on the fly or try to write a disc image as one interrupted file), and the number of drives.
Your system and writing mode go hand in hand. As noted later in this section, a CD
recorder requires a constant, uninterrupted stream of data to make a disc. The speed at
which your computer can maintain that data flow can constrain the maximum writing
speed of a CD-R drive. Factors that determine the rate of data flow include the speed of
the source of the data (your hard disk), the fragmentation of the data, and the interfaces
between the source disc and your CD recorder.
Most CD recorders have built-in buffers to bridge across temporary slowdowns in the
data supply, such as may be involved when your hard disks read/write head repeatedly
moves from track to track to gather a highly fragmented file or when an older, non-A/V
drive performs a thermal calibration. Even with this bridge action, however, such hard
disk slowdowns reduce the net flow of data to the CD recorder. If you try to create a CD
by gathering together hundreds of short hard disk files on the fly, your hard disk may not
be able to keep up with the data needs of a 4 CD recorder. In fact, if the files are many
and small, the hard disk may not even be able to maintain 1 speed, forcing you to resort
to making an image file before writing to the disc.
On the other hand, one manufacturer (Mitsumi) reports that higher writing speeds produce more reliable CDs. At the 1 writing speed, the laser remains focused on a given disc
area longer, possibly overheating it. In other words, you may want to avoid 1 speed
unless the performance of your system and its software requires it. Although early software, drives, and computers often could not keep up with speeds in excess of 1, most
current products do not have difficulties at higher speeds.
671
672
When you have to produce a large number of CDs quickly, one of the best strategies is to
use multiple drives. Five drives writing simultaneously cuts the net creation time of an
individual CD by 80 percent. For moderate-volume applications, stacks of CD writers can
make a lot of senseand CDs. For large-volume applications (generally more than a few
hundred), pressing CDs is the most cost-effective means of duplication, albeit one that
requires waiting a few days for mastering and pressing.
Disc-Writing Modes
Depending on your CD-R drive and your CD-creation software, you may have your
choice of the mode you use for writing to your CD. The mode determines what you can
write to your discs and when. Typically you dont have to worry about the writing mode
because your software takes care of the details automatically. However, some drives and
software may be limited to the modes under which they can operate.
The basic CD-writing modes are four: track-at-once, multisession, disc-at-once, and
incremental writing. Each has its own requirements, limitations, and applications. A new
standard, Mount Rainier, discussed separately, ensures compatible on incrementally written discs.
Track-at-Once
The most basic writing method for CDs is the creation of a single track. A track can be in
any format that your CD-R drive can write (for example, a CD-ROM compatible disc or
a CD-DA disc for your stereo system). The track-at-once process writes an entire track in
a single operation. A track must be larger than 300 blocks and smaller than the total
capacity of the disc minus its overhead.
Writing track-at-once requires only that you designate what files you want to put on a
CD. Your CD-creation software takes over and handles the entire writing process.
Originally the big limitation of track-at-once writing was that you could write only one
track on a disc in a single session. Consequently, unless you had a lot to write to your disc
already prepared beforehand, this process was wasteful of disc space. Some modern CD
systems can add one track at a time to a disc within a single session, even allowing you to
remove the disc from the drive and try it in another in the middle of the process.
Each track has overhead totaling 150 blocks for run-in, run-out, pre-gap and linking. CD
standards allow 99 tracks per disc. Consequently, if your tracks are small, you may waste
substantial capacity. Writing the maximum number of blocks of minimal size (300 blocks
plus 150 blocks of overhead each) will only about half-fill the smallest, 18-minute CD disc
(44,550 blocks on a 81,000 block disc).
Track Multisession
Sometimes called track incremental mode, track multisession mode is the most common
means of allowing you to take advantage of the full capacity of CDs. Track multisession
Standards
writing allows you to add to CDs as you have the need for it by dividing the capacity of
the disc into multiple sessions, up to about 50 of them. Each session has many of the characteristics of a complete CD, including its own lead-in and lead-out areas as well as a table
of contents.
In fact, the need for these special formatting areas for each session is what limits the number of sessions on the disc. The lead-in and lead-out areas together require about 13.5MB
of disc space. Consequently, CDs with a total capacity of 680MB can hold no more than
about 50 sessions.
When the CD standards were first created, engineers didnt even consider the possibility
that individual consumers would ever be able to write their own discs. Consequently, they
assumed that all discs would be factory mastered in a single session. They designed early
CD drives to recognize only one session on a disc. Many older CD-ROM drives (particularly those with 1 and 2 speed ratings) were single-session models and cannot handle
multisession discs written in track multisession mode. Single-session drives generally read
only the first session on a disc and ignore the rest.
Another problem that may arise with multisession discs is the mixing of formats. Many
CD players are incapable of handling discs on which CD-ROM Mode 1 or 2 sessions are
mixed with XA sessions. The dangerous aspect of this problem is that some CD-mastering
software (and CD drives) allow you to freely mix formats in different sessions. You may
create a disc that works when you read it on your CD drive that cannot function in other
CD drives. The moral is not to mix formats on a disc. (Dont confuse format with data
type. You can freely mix audio, video, and pure data as long as they are written in the
same format, providing the one you choose is compatible with all three data types.)
Most modern CD-R machines allow you to write more than one track in a given session.
The advantage of this technique is the elimination of most of the 13.5MB session overhead. Instead of lead-in and lead-out tracks, each pair of tracks is separated by 150 blocks
(two seconds) of pre-gapoverhead of only about 300KB. The entire session must, of
course, be framed by its own lead-in, table of contents, and lead-out areas.
In multisession discs, the drive writes to the lead-in area after it finishes with the data on
the disc. The lead-in contains the table of contents for the session as well as an indication
of the remaining writable area on the disc. The lead-in of the last session on the disc indicates that no more sessions are present, closing the disc.
Disc-at-Once
Old-fashioned vinyl phonograph records were cut as a single, continuous process. From
the moment the cutting stylus plunked down on the master disc until it finished the disc,
spinning around in the capture track, the mastering process had to be free of interruptions. After all, any gap in the spiral track of the phonograph record would stall your
record player. To cut a master record, the engineers prepared a master tape that was
673
674
complete in every detail of everything that was to go on the final disc, including blank
tape for the gaps between cuts on the final disc.
The CD equivalent to making such a master disc is the disc-at-once process. As with cutting a master record, the disc-at-once process must be completely free from interruption
from the beginning of the lead-in area to the completion of the lead-out area. The table
of contents, all tracks, and the Q channel must all be prepared before the writing process
begins. The entire disc will be written in one swoop so that the formatting data will
appear on the disc (for example, the lead-in will be written before the data). Typically, to
make a CD using disc-at-once writing, youll prepare an exact image of the CD and store
it on a hard disc. The hard disc must be A/V rated so that it does not interrupt the data
stream for thermal calibration or other housekeeping and thus cause buffer underrun (see
the section titled Underrun, later in the chapter).
In effect, disc-at-once is a combination of track-at-once and multisession writing that simply extends across the entire CD (or as much of it as will ever be used).
Disc-at-once is the recording method that must be used when you prepare a disc to serve
as the master for making mass-produced CDs. Because the laser never turns off, a disc
recorded using the disc-at-once mode contains no link blocks.
Packet Writing
If you could make a CD-R work like a conventional hard disc, it would be capable of
incremental writing. That is, you could add data to your disc whenever you needed to simply by saving a file. In CD terminology, this is called packet writing. With appropriate software drivers, you can drag and drop files to your CD recorder as if it were a hard disk
drive.
In this context, a packet is a block of data smaller than a track. Your drive accepts the
packet and writes it to the disc, identifying it with four blocks of run-in information, two
of run-out information, and a link block. Each packet thus suffers seven blocks or about
15KB of overhead in addition to that required for directory information.
The ISO 9660 file system comes up short in packet writing. It requires that all the file
information be written in the table of contents when you create a session. Multisession
discs sidestep this problem by creating a new file system every time you write a new session, with all the overhead of a complete file system (whoops, there goes another
13.5MB). Packet writing therefore requires drives and software that follow the Universal
Data Format (UDF) system, discussed under DVD-ROM, later in the chapter.
Mount Rainier
Whereas packet writing requires software added as an application on top of an operating
system, the Mount Rainier standard incorporates the same functionality (and more) within
the operating system. Jointly developed by Compaq (now Hewlett-Packard), Microsoft,
Standards
Philips Electronics, and Sony, the Mount Rainier specification obsoletes packet writing
with a similar drag-and-drop interface to allow random writing to CD and DVD drives.
In addition, the specification requires that drives be able to access data in 2KB allocation
units, like those used on magnetic disks.
One of the most important parts of the Mount Rainier specification is shifting responsibility for managing disc defects from the packet-writing software to the disc drive itself. The
drive maintains a map of bad sections of the disc in a special table carved from the user
data area of the disc. Using the table, the drive can skip over bad areas of the disc when
writing data without any intervention from the operating systemwithout the operating
system even knowing the bad areas exist.
This feature alone requires new drive designs. Conventional CD and DVD drives cannot
be upgraded to Mount Rainier technology. On the other hand, conventional media work
with Mount Rainier drives, and Mount Rainier drives can read discs written under earlier
standards.
Mount Rainier eliminates much of the hassle of formatting CDs by moving the process to
the background. Although you can use preformatted discs with a Mount Rainier drive,
using unformatted discs imposes no penalty. Slide a new disc in the drive, and the drive
automatically starts formatting it even as you write data.
Discs created using Mount Rainier technology cannot be read by conventional drives
without special software drivers. Make a Mount Rainier disc in one computer, and you
wont be able to read it on another machine that does not have a driver that supports the
technology. You can enable an older system to read (but not write) Mount Rainier discs by
installing new driver software (when it becomes available).
The common name for Mount Rainier is EasyWrite technology. The promoters of the
new format have developed a certification program and a logo that lets you quickly identify drives that correctly implement the technology.
The Mount Rainier specification was first published on July 30, 2002. The Web site
www.mt-rainier.org provides access to the full specification.
Underrun
No matter the mode, the CD-writing process is continuous, start to finish. The laser
switches on at the beginning of a session and remains in continuous operation until that
session is finished. The CD format requires the interleaving of data between blocks during the writing process to help ensure data integrity. To properly sort the interleaved data,
the drive needs an overview of the data. To gain this overview, the drive has a data buffer
from which it draws the data to write.
For the laser in a CD-R drive to operate continuously, it must have a continuous supply of
data to keep its buffer filled with enough information to properly perform the
675
676
interleaving. If at any time it runs out of data to write, the writing process is interrupted.
Unlike hard disks, the CD drive cant pick up where it left off on the next spin of the disc.
The error resulting from the interruption of the data flow is termed buffer underrun.
CD players see the interrupted session as an error (which it is) that may render the disc
unplayable. In other words, buffer underrun ruins a disc. Better CD-R drives allow you to
close the interrupted session and recover the remaining space on the disc for other sessions.
Testing
To prevent you from wasting discs with inadvertent data underruns, most CD-R mastering software makes a trial run or test of the recording session before actually committing
your data to disc. The test involves performing exactly the same steps as the actual write
operationincluding operating the laser in the drive in its write modebut keeps the
power of the laser at read level. The CD-R drive runs through the entire write operation,
but the lower power of the laser prevents it from affecting (and potentially ruining) a disc.
If the recording software discovers a problem during recording that would cause an
underrun or other problem, it will advise you how to sidestep the problem, typically by
stepping down to a lower writing speed on your CD-R drive or, as a last resort, defragmenting your hard disk.
The only problem with pre-write testing is that the trial run takes as long as writing
everything to your disc, essentially doubling the write time of every disc you make. Most
CD-mastering programs allow you to switch off this pre-write testing. Although you do
this at your own peril (and the expense of ruined CDs), if youre making a batch of discs it
is a viable timesaving option. In general, if you can write the first disc successfully, you can
run through dozens of additional copies without worry.
Fixation
Before a CD that you write can be read by a CD-ROM drive or the audio CD player in
your stereo system, it must have an overall table of contents that follows the ISO 9660
standard. The process of finishing the disc for reading is termed fixation. In the process of
fixation, the disc is finalized when your CD-R drive writes an overall absolute lead-in area
and absolute lead-out area for the entire disc.
Multisession drives also can create discs that are fixated for appending. The individual sessions each have their own table of contents that reflects the sessions actually written on
the disc, but the disc lacks the overall lead-in and lead-out areas. When youve added the
last session to the disc, the finalization process writes an indication on the disc that no further sessions are present, then writes the overall disc lead-in and lead-out areas, completing a table of contents compatible with the ISO 9660 standard. Most CD-mastering
programs refer to this finalization process as closing the disc.
Standards
Software that performs packet writingfor example, Sonys Compact Disc Recordable
File System (CDRFS)may require a process termed freezing the disc before you can use
packet-written discs in ordinary CD players. The freezing process writes lead-in and leadout areas on the disc. After a disc has been frozen, you can still write additional sessions
onto it, providing, of course, additional capacity is available. The freeze process only subtracts from the available capacity, draining away the 13MB of overhead required by any
single session.
CD-Rewritable
CD-RW stands for CD-Rewritable, meaning that these drives can create new CDs that
you can erase and use again. In fact, you can treat a rewritable CD as if it were a big
floppy disk drive or slow hard disk drive. All CD-RW drives also function as CD-R drives,
and they are standardized under the Orange Book just as are CD-R drives.
The difference between CD-R and CD-RW is in the media. Put a blank CD-R disc in a
CD-RW drive, and you make a permanent record that you cannot change. With a CDRW disc, you can rewrite and reuse disc space thanks to its phase-change medium.
CD-RW drives are actually the third incarnation of phase-change technology. The first
drives, under the Phase Change Recordable banner (or PCR), were made by Toray
Industries. It used a medium slightly larger than CDs, 130 millimeters in diameter as
opposed to the CDs 120 mm, that was consequently physically incompatible with CD
drives. Both sides of the disc had a recordable surface, allowing for a total capacity of
1.5GB per disc. Panasonics PD discs reduced the size of the disc to the same as CDs and
modified the storage format. The actual writing format uses sectors of 512 bytes versus
the 2048 byte or larger sectors used by CD, so PD discs are also logically incompatible
with CDs. You cannot duplicate a CD on the PD medium. Moreover, the logical format
of the PD system limits its capacity to 650MB per disc, as opposed to the 680MB total of
CDs. Further, the phase-change material used by the Panasonic PD drives is not compatible with the optical heads and electronics of CD drives. Although the electronics of the
Panasonic drives adapted to handle either phase-change or conventional CD media, PD
discs work only in PD drives.
Because of the lower reflectivity of CD-RW media, phase-change discs often are unreadable in early (pre-1998) CD-ROM and CD-R drives. Newer drives have compensatory
circuitry built in called automatic gain control.
In operation, a CD-RW drive can function more like a conventional hard disk than a CDR. The drive can update the disc table of contents at any time, so you can add files and
tracks without additional session overhead. Under Windows, you typically drag and drop
files to your CD-RW drive just like you would with any other disk. The Mount Rainier
standard, discussed earlier, formalizes a system with this capability and more.
677
678
Discs written by CD-RW drives made before the Mount Rainier standard are not entirely
compatible with all CD drives. The format used during CD-RW operation in preMount
Rainier drives is usually different from that of conventional CDs. To read a CD-RW disc
in a CD-ROM or CD-R drive, the disc must be closed, an operation that effectively reorganizes its format. In the typical implementation, the reorganization process requires
blank space on the disc, so you cannot fill one of these CD-RW discs with data and expect
to later use it in another drive.
DVD-ROM
To your computer, the storage on a DVD looks much like that of any disk system.
Information is organized into 2KB blocks that correspond to the clusters on disk systems.
The file structure takes the form of the Micro UDF/ISO Bridge format. In effect it
bridges two storage formats.
Universal Data Format (UDF) was designed by the Optical Storage Technology
Association (OSTA), a group of companies involved in optical data storage, to make data
stored on optical discs independent of any operating system (hence, universal). The goal
was to allow you to write an optical disc on your computer and read it on any other computer in the world, regardless of operating system, microprocessor, or even whether it was
powered by electricity or steam. UDF defines the data structures (partitions, files, sectors),
error correction, character sets, and read/write method of the DVD system. ISO 9660
defines the tree-oriented directory structure (the same as on computer CDs) compatible
with Windows and other popular operating systems. The overall structure of the disc fits
the UDF format with the ISO 9660 structure on top, the intent being to eventually eliminate ISO 9660 support.
The UDF specification has incremented up to version 2.01, the level that OSTA recommends DVD-ROM publishers follow. DVD-Video discs and players are locked to the
UDF version 1.02 specifications. You can download both versions of the specification in
their entirety from www.osta.org.
The Micro UDF/ISO Bridge format imposes some limits on DVDs. One oft-quoted limit
is a maximum file size of 1GB. This constraint applied only to DVD-Video discs.
Although it seems incompatible with two hours of continuous video playback, the DVD
system was designed to agglomerate multiple small files (including video, audio, and control information), process them together, and output a single continuous video stream.
On the disc itself, the block is chopped and scattered to help in error recovery. The tiny
size of the pits on the disc means that a splotch or scratch will likely span a considerable
storage area. Spreading the data out scatters potential read errors into small pieces in different storage units so that they can be more readily detected and corrected.
Standards
Each 2KB block is translated into a 2064-byte physical sector for storage on the disc. The
sector gets further subdivided into 12 rows of 172 bytes each. The central ten rows store
only data. The first starts with a 12-byte sector header to identify the storage unit. Four
bytes provide the actual ID information with two additional bytes used for error correction dedicated to the ID data. The remaining six bytes in the header are reserved. The
following 160 bytes in the first row contain data. The last row of each sector ends with
four bytes of error-detection and error-correction information for the data area of the sector.
Sixteen sectors are then interleaved together to form a larger storage unit, the block. Ten
bytes of error-correction code are added to each row in the block, and the resulting overall block gains another 16 rows of error-correction code. The result is a block thats
37,856 bytes, with about 15 percent of its contents devoted to error-correction information. These blocks are then written sequentially to the disc.
The DVD-ROM format allows for both constant linear velocity and constant angular
velocity recording. The former favors capacity in sequential access applications (audio and
video). The latter improves random access speed for data applications but reduces the
capacity of each layer. Although it is an inherent part of the DVD-ROM specification, the
chief application of CAV recording in DVD has been the various rewritable formats.
The required support to read UDF-based DVD-ROM discs is built in to Windows 98
and later Windows versions. Earlier Windows versions will require add-on drivers, which
are usually included with DVD drives. In addition to decoder software, playback of DVDVideo requires DirectShow 5.2 or newer, which is included with Windows 98. More
recent versions can be downloaded from Microsofts Web site at www.microsoft.com.
Again, DVD drives usually include the necessary decoder (as software or as a separate
hardware MPEG-decoder board) as well as DirectShow.
DVD-Video
The DVD-Video system devotes one track to video information that may use either
MPEG-1 or MPEG-2 encoding. All commercial discs use MPEG-2 because it simply
looks (and works) better. A dedicated DVD player can render either into video for your
television or monitor. With current technologies, a DVD-ROM player in a computer
works best with a separate hardware-based MPEG-2 decoder. Only the fastest computer
processor can decode MPEG-2 in real time, and even these dont work as well as dedicated hardware decoders. Moreover, because your computer must devote nearly all its
power to decoding the video, theres little left for doing other work at the same time.
DVD-Video would not be possible without compression (nor would any other digital
video system be practical). DVD-Video data originates with a bit-rate of 124Mbps, and it
must be compressed down to the maximum rate permitted under the DVD standard,
9.6Mbps. The average bit rate is about 3.5Mbps. Despite the heavy-duty compression,
679
680
DVD still delivers about twice the resolution of VHS videocassettes. A typical DVDVideo system produces horizontal resolution of about 500 lines, compared to less than 240
lines for a VHS tape.
DVD-Video goes far beyond todays VHS and CD-Video systems. It allows both conventional-style images with the 4:3 aspect ratio as well as those with the 16:9 ratio favored by
High-Definition Television systems. DVD players are required to have built-in filters to
translate 16:9 images into the full-width of a 4:3 aspect ratio screenin other words,
built-in letterbox format translation. The DVD players will also allow you to zoom in to
fill the screen height with a 16:9 image and pan to either side of the picture. The MPEG2 encoding delivers about four times the spatial resolution as MPEG-1 (used by some CD
systems) and allows a high-quality display with 480 lines of 720 pixels, each to fit into a
4MBps data stream. As with any video-compression technique, the exact data rate depends
on the complexity of the image. Typically a high-quality image requires less data than a
low-quality one plagued by noise. The onscreen resolution produced by a DVD system is
in the range of 480500 horizontal lines. Unlike other video systemsVCRs and laserdisc
systemsDVD stores video images in component format (separate red, green, and blue
images). Other consumer formats use composite video.
DVD-Video also introduces the concept of subpictures, which are additional images of limited color depth that can be multiplexed with the main audio and video data. The DVD
standard allows for up to 32 subpictures, which typically will be menus for control systems, subtitles for foreign language films, or production credits. Each subpicture can measure as large as 720 by 480 pixels in four colors.
Note that DVD-Video is not High-Definition Television (HDTV). About the only thing
in common between the two is the 16:9 aspect ratio supported by both. DVD-Video is
more closely aligned with standard NTSC video, offering quality similar to that in the
television studio. An HDTV image has about five times the number of pixels as the
DVD-Video format. Its compressed format requires about twice the data rate (about
19.4Mbps). Certainly the DVD-18 medium has enough capacity to store HDTV, but it
will require a new storage and playback format to cope with the HDTV data rate.
DVD mimics HDTV by offering a wide aspect ratio format. The DVD standard allows
for both the old television and video aspect ratio of 4:3 and the HDTV aspect ratio of
16:9. The wide aspect ratio images have the same number of pixels as 4:3 images. The
image is compressed anamorphically to fit. On playback the ostensibly square pixels get
stretched horizontally to the wider aspect ratio. The standard allows the display of wide
aspect ratio images in three different ways, which you can select when playing back a disc.
These include the following:
Letterbox mode. This mode fills the width of the screen with the full width of the
image, leaving bands 60 lines high at the top and bottom of the screen black.
Standards
Pan and Scan mode. This mode fills the full narrow screen with a window into the
wide image. The window is dynamic. The disc stores cues determined by the director of the film or producer of the DVD that guide the window to follow the action
in the image.
Widescreen mode. This mode provides a full-width image on a 16:9 aspect ratio
screen.
The audio accompanying DVD-Video can take any of many forms. The most common
(used on nearly all releases of theatrical movies) is Dolby Digital. Although the Dolby
Digital system can accommodate up to 5.1 channels of PCM audio, the standard embraces
lesser configurations as wellincluding simple monophonic and stereophonic recordings.
In other words, a label proclaiming Dolby Digital on a movie box does not guarantee
anything in the way of true multichannel sound.
The required audio support depends on the standard followed by the recorded video.
Discs containing NTSC video (the standard in North America and Japan) are required to
use Dolby Digital. Discs containing PAL video (Europe and most of the rest of the world)
must use MPEG-2 audio. Other audio formats may optionally accompany either video
standard.
The DVD-Video standard accommodates eight tracks of audio, each track being a single
data stream that may comprise one or more audio channels. Each of these channels may
use any of five encoding systems. In addition to Dolby Digital (with up to 5.1 channels
per track), all DVD drives must also be able to decode PCM audio, up to eight channels
per track (potentially 64 channels per disc, but real data rates ordain the channel count be
lower), and MPEG audio (up to 7.1 channels per track). Optional decoding systems
which may or may not be included within the circuitry of a given DVD drive but can be
attached as an accessoryinclude Digital Theater Sound (DTS) and Sony Dynamic
Digital Sound (SDDS). Chapter 25, Audio Systems, discusses these systems in more
detail.
Video DVDs are burdened with several layers of copy protection that are entwined both
with hardware and operating system software. Copy-protection occurs at three levels:
Analog copy protection. Alters the video output so that the signal appears corrupted to VCRs when you attempt to record it. The current Analog Protection System
(APS) adds a rapid modulation to the colorburst signal (called a colorstripe) and pulses
in the vertical blanking signal (termed AGC because it is meant to confuse the automatic gain control of VCRs). Computer video boards with conventional analog video
outputs (composite or S-video) must incorporate APS. DVD-Video discs themselves
control whether the APS system is used on playback by signaling to the disc player
to switch APS on or off.
681
682
Serial copy protection. Protects by encoding control information in the signal that
determines whether it can be copied. The copy generation management system adds
information to line 21 of the NTSC video signal to tell equipment whether copying
is permitted. Although DVD-Video encodes line 21 differently, the information is
regenerated into the analog video output of DVD players and computer video
boards.
Digital encoding. Encrypts the digital form of media files requiring the video
player to know the key necessary for decrypting the code. The DIVX licensing system (abandoned as a DVD format on June 16, 1999) works through digital encoding.
In addition, DVDs are marked with a regional code, a number that specifies the part of
the world in which playback of the DVDs content is permitted. The DVD player checks
to see whether a region code on the software matches that encoded into its hardware. If
the two dont match, the disc wont play. DVD media boxes are marked with the region
code as a number on a globe icon, the number corresponding to one of six regions, as
listed in Table 18.5.
TA B L E 1 8 . 5
DVD Regional Codes
Code Number
Region
Canada and United States (including U.S. territories)
Egypt, Europe, Japan, Middle East, and South Africa
East and Southeast Asia
Central and South America, Caribbean, Australia, New Zealand, and Pacific
Islands
Africa, Former Soviet Union, Indian subcontinent, Mongolia, and North

Korea
China
DVD-Audio
Ever since the introduction of the CD-DA format, audio purists have insisted that it was
not good enough. They could hear a definite digital sound (whatever that might be) that
interfered with their enjoyment of music. They were quick to point out that tradeoffs
made in the design of the CD-DA format slighted sound quality. The relatively low sampling rate required to pack enough information on a disc required high-order low-pass filtering to keep digital artifacts out of the audio, and even then the upper cutoff frequency
must be too low to accommodate everything that old analog tape recorders could capture.
DVD-Audio
The huge capacity of the DVD system eliminates the need for the tradeoffs of the
CD-DA system and unleashes the potential for superb audio qualitygood enough to
satisfy listeners with 24-karat ears. The audio-only implementation of DVDtermed
DVD-Audioisnt just for purists, however. The system can use the multigigabyte storage
of the medium for super-quality audio, additional channels, or both.
Whereas CDs are built around a 44.1KHz sampling rate, DVD supports both this rate
(and the whole Red Book standard) as well as 48KHz, the same as professional audio systems, and a super-high quality 96KHz sampling rate. In addition to todays 16-bit digital
audio, DVD will also support 24-bit audio as well as several compressed multichannel formats to accompany video. The multichannel audio standards vary with the video standard
used, with Dolby AC-3 (eight-channel audio) for NTSC video.
The options allowed the producer for linear PCM are wide. The systems support up to six
channels with bit-depths of 16, 20, and 24. Recording may use either of two bit-rate familiesthe 44.1KHz of Red Book audio, supplemented by 88.2 and 176.4KHz multiples, or
a 48KHz base rate supplemented by 96 and 192KHz multiples. The highest sampling rate
and bit-depth (192KHz at 24 bits) allows response to nearly 96KHz, with a dynamic range
approaching 144 dB. The maximum data rate of the system, 9.6Mbps, constrains the system to two channels of these highest-quality signals, and the storage capacity of the DVD
system limits playing time (at this quality level) to about 67 minutes.
To extend playing time, the DVD-Audio standard allows for data compression of various
sorts, including MPEG. Most intriguing is MLP, which stands for Meridian Lossless
Packing. As the name implies, this system works like file compression and reduces data
redundancies in the digital audio but allows the original signal to be perfectly reconstructed during decoding. Compression extends playing time enough to allow up to about
135 minutes of six-channel 24-bit audio sampled at the 96KHz rate.
To make six-channel audio compatible with two-channel stereo sound systems, DVD-Audio
incorporates a special System-Managed Audio Resource Technique, or SMART (entirely different from the S.M.A.R.T. hard disk failure-prediction technology). This system allows the
producer of an audio DVD to determine the optimum mix-down that combines the channels together for each selection on each disc. The mixdown information is stored as a table
of coefficientsthe level assigned to each channel in the mixand each selection on the
disc can have a particular table assigned to it. Each disc can accommodate up to 16 tables.
DVD-Audio discs also accommodate other kinds of data in addition to digital audio.
The standard allows for including up to 16 still images in each track and for synchronizing a display of lyrics with the music. In addition, the DVD-Audio system provides for
computer-style navigation through a screen-oriented menu system (or a simplified control
system for audio players without computer displays).
683
684
Write-Once DVD Formats

The three write-once DVD systems are all based on the same technology as CD-R discs.
They use an organic-dye medium that is essentially burnt by a high-power laser to create
darkened pits on a disc.
DVD-R(A)
The A in the name stands for authoring. DVD-R(A) was the first rewritable DVD format,
marketed by Pioneer in 1997. The initial drive stored only 3.95GB per disc and was
priced at $17,000hardly a mass-market item. The format was revised in 1999 to accommodate the full 4.7GB of other DVD systems. Both the 3.95GB and 4.7GB formats
remain in use.
The DVD-R system is aimed at professional applications. It has been widely deployed to
store images of documents used in commerce and for mastering DVDs, where it replaces
the Digital Linear Tape systems once widely used for the purpose. Using the new Cutting
Master Format on 4.7GB discs allows the use of part of the lead-in track to hold the Disc
Description Protocol header information that was used on DLT tapes.
The DVD-R(A) system is centered around a 635-nanometer laser thats used for writing
discs. The dye media are tuned to this wavelength, so only discs designed for DVD-R(A)
can be recorded in DVD-R(A) writers. The resulting discs can, however, be read by most
DVD players, including both DVD-R(A) and DVD-R(G) drives.
The DVD-R(A) standard is sanctioned by the DVD Forum and is an official standard of
ECMA International (which changed its name from the European Computer
Manufacturers Association in 1994) titled ECMA-279.
DVD-R(G)
The G in the name stands for general purposes. DVD-R(G) is both an offshoot and refinement of the older DVD-R(A) technology. The chief change is the substitution of a lessexpensive 650-nanometer laser for the shorter wavelength used in the DVD-R(A) system.
This laser allows for less expensive drives and media but inhibits compatibility. DVDR(G) discs can be written only in DVD-R(G) drives, although they can be read in almost
any DVD player, including those designed for DVD-R(A) discs. The change in laser does
not affect capacity. DVD-R(G) discs hold up to 4.7GB.
Because DVD-R(G) is a consumer-level medium, it incorporates content protection to the
extent that a DVD-R(G) drive cannot make a copy of an ordinary DVDfor example, a
motion picturethat uses CSS encryption. Moreover, the DVD-R(G) system does not
support the CRF system used in mastering DVDs. The DVD-R(G) standard is sanctioned
by the DVD Forum and is part of the DVD-Multi specification.
DVD-Audio
DVD+R
DVD+R is a relatively recent addition to the write-once repertory. An offshoot of
DVD+RW technology (hence the use of the plus sign in the name), DVD+R discs can be
created in any DVD+RW drive manufactured since the introduction of the format in
April 2002.
As with CDs, the write-once DVD+R media are less expensive than the rewritable media,
so DVD+R is a more affordable way of making permanent records. In addition, the dyebased medium is more reflective than the rewritable medium, making it more easily read
in commercial DVD players. Whereas DVD+RW playback may be marginal in some
early DVD players, DVD+R discs will likely work better.
The foremost consideration in developing the DVD+R format was compatibility with
both DAT drives and video drives. As a consequence, the media hold the same 4.7GB as
commercial DVDs, and there are no plans for alternate formats. The DVD+R specifications do allow for two-sided media, but the technology precludes the use of multilayer
discs.
The DVD+R specification is sanctioned by the DVD+RW Alliance, and the official standard is published (along with DVD+RW) as ECMA-274.
Rewritable DVD Formats

The same phase-change technology that makes CD-RW possible has been applied to
DVD to make a number of rewritable formats. The first of these came from the DVD
Forum (and all were delivered months later than promised).
DVD-RAM
The first rewritable format, DVD-RAM, was aimed particularly at data applications. It
uses a combination of phase-change and magneto-optical technologies to produce a longlife medium with the ruggedness further enhanced by putting each disc in its own cartridge/carrier, much like a floppy disk shell. As a result, the DVD-RAM system produces
discs projected not only to be reliable data archives for more than 30 years, but also the
medium itself should endure more than 100,000 write/rewrite cycles.
When DVD-RAM drives first appeared in June, 1998 each disc packed only 2.6GB per
side. Two-sided discs doubled that capacity but added the need to flip the cartridge over
to access the second side. By October 1999, the specification had been extended to a
capacity of 4.7GB per disc side with the first drives arriving in June 2000. The revised
specification also included cartridges using discs 80 millimeters in diameter, aimed particularly at high-end digital video camcorders.
The DVD-RAM system uses a wobbled groove to generate the clock used for synchronizing data writing with the spin of the disc. The wobble, a periodic side-to-side twist of the
685
686
spiral track, is pressed into the blank disc along with fixed sector headers during manufacture. Initially all DVD-RAM discs were encased in cartridges, but the latest design makes
the disc removable from the cartridge so you can use it in other formats of DVD drive.
Drives can write only on discs encased in their cartridges. Most sources list the DVDRAM system as incompatible with most other drive formats. DVD-RAM drives write only
DVD-RAM discs, and reading the discs in other kinds of drives is chancy. DVD-RAM
drives, on the other hand, readily read all other DVD formats.
DVD-RAM drives have built-in defect management, making them the best choice for
computer data. In professional camcorders, DVD-RAM has proved impressive, and the
defect-management system built in to the drives makes them a good choice for randomaccess use, their original design intent. The chief handicap of the DVD-RAM system is its
inability to make discs that are playable on other kinds of DVD drives.
The DVD-RAM specification is maintained by the DVD Forum, and the standard is published by ECMA as ECMA-272 and ECMA-273.
DVD-RW
Developed originally as a companion to and offshoot of the DVD-R(A) format, DVDRW is the DVD Forums choice for a sequential read/write DVD medium. It uses about
the same track pitch, data length, and speed control (constant linear velocity) as both
DVD-R formats and differs chiefly in medium. DVD-RW uses a phase-change medium,
which has somewhat less reflectivity than do the write-once dyes, 18 to 30 percent for
DVD-RW, compared to 45 to 80 percent for single-layer DVD-ROM. As a consequence,
DVD-RWs initial slow start was further impeded by incompatibilities. Ordinary DVD
drives had difficulty reading DVD-RW discs, often mistaking the lower reflectivity as
two-layer media in that the discs 18-to-30-percent reflectivity is the same as that of twolayer DVD-ROMs.
Blank DVD-RW discs actually have two kinds of data prewritten to them. To synchronize
the spinning disc with data being written, the disc has address information written on the
land area between the grooves of the spiral. In addition, each disc has a prewritten lead-in
track thats used by copy-protection systems to prevent the pirating of movies. This leadin data takes three forms:
Version 1.0 discs have it physically embossed on the disc during manufacturing
Version 1.1 discs (the most common) have the data written normally (that is, by a
special DVD-RW drive) on the lead-in
Variation B of Version 1.1 discs also have a unique 64-bit disc identification barcode etched near the hub (in the burst cutting area) thats required by some serial
copy management systems.
DVD-Audio
The first DVD-RW systems were designed specifically for recording video. They operated as sequential media and supported only disc-at-once writing. Although editing was
possible, it was cumbersome because the entire disc required rewriting for a single simple
change. Initial drives operated only in real timethat is, 1 speedalthough faster drives
are now available. Discs must be formatted before use, although preformatted discs and a
quick-format process help minimize the pain.
The expected lifetime of a DVD-RW disc is about 1000 write/rewrite cycles. DVD-RW is
sanctioned by the DVD Forum and is part of the DVD-Multi specification.
DVD+RW
From the start, DVD+RW was designed to be a random-access data medium. At the same
time, the system was designed to be capable of producing discs compatible with most
DVD players. The DVD+RW format is compatible with standard DVDs, and the same
650-nanometer laser reads and writes the medium. The only significant difference is the
medium itself, a phase-change compound with somewhat lower reflectivity than standard
DVD discs, comparable to that of DVD-RW. As a result, some older DVD players may
mistake DVD+RW media for two-layer discs.
Although DVD+RW discs are compatible during reading, the system incorporates a number of enhancements to make DVD+RW useful in data storage. The standard allows
drives to use constant linear velocity recording (as with DVD-RW) as well as constant
angular velocity recording (constant spin rate) for faster random access. Discs made with
either spin-control method are playable on standard DVD players (which use CLV).
For synchronizing disc spin and writing speed, DVD+RW uses wobble tracksa periodic
radial shift to the spiral groove. The wobble produces a frequency of 817KHz during 1
recording, correspondingly higher at faster write speeds. The wobble clock provides four
addresses for each 32KB block of data on the disc, allowing the drive to accurately locate
any 2KB data cluster. This addressability allows for lossless linking, a term used by the
DVD+RW promoters to describe the ability of a DVD+RW drive to resume writing
within one micron of the place it left off whenever it is interrupted for whatever reason.
This ability to locate accurately any 2KB block of data makes the DVD+RW system truly
random access. It can change any data cluster independently of the others without the
need to rewrite the entire disc. This ability fits with the Mount Rainier drag-and-drop
writing process, so making Mount Rainier DVD+RW drives requires little more than new
firmware and certification.
Initially DVD+RW allowed for only one kind of disc, a 12-centimeter disc holding
4.7GB. The same disc can be used both in computer drives for storing data and in dedicated DVD recorders for motion pictures. In August, 2002, the DVD+RW alliance
approved 8-centimeter discs holding 1.46GB for applications such as digital camcorders.
687
688
DVD+RW discs require formatting before use, but the DVD+RW system supports automatic background formatting. As soon as you insert a new disc into a drive, it starts formatting the lead-in. This way, the drive is immediately ready to use the disc. If you
interrupt the formatting process (for example, by removing a partially formatted disc), the
drive will resume formatting the disc the next time you insert it. DVD+RW drives support
automatic defect management, hiding bad storage areas from your computer and operating system.
The DVD+RW system allows for standalone DVD video recorders. In such applications,
it supports four writing modes or quality levels, differing chiefly by bit-rate and playing
time. Table 18.6 summarizes these speeds.
TA B L E 1 8 . 6
DVD+RW Video Recorder Quality Levels
Designation
Name
Recording Time
Data Rate
Resolution
HQ
High Quality
1 hour
9.72Mbits/sec
720 480 pixels
SP
Standard Play
2 hours
5.07Mbits/sec
720 480 pixels
LP
Long Play
3 hours
3.38Mbits/sec
360 480 pixels
EP
Extended Play
4 hours
2.64Mbits/sec
360 480 pixels
Initial DVD+RW drives for computers operated at up to 2.4 normal DVD speed and
allowed writing at their highest speeds. In August 2002, the DVD+RW Alliance
announced the approval of 4 drives.
Although not supported by the DVD Forum, the DVD+RW format is standardized at
ECMA-247.
Blu-ray
In February 2002, a consortium of nine electronics firms (none Americanthe list
includes Hitachi, LG Electronics, Matsushita, Philips Electronics, Pioneer, Samsung,
Sharp, Sony, and Thomson Multimedia) announced it had agreed on the specifications for
the successor to the DVD. Called the Blu-ray Disc, the innovation resembles nothing
more than a DVDits the same size and color and uses the same basic technology. The
disc spins in a drive and a laser reads tiny digital spots to play back movies or music. The
difference is capacity. The initial Blu-ray discs will be able to store up to 27GB of data,
about 5.75 times more than todays DVDs. Moreover, plans include more than doubling
that capacity with two-layer discs (a technology already used by DVDs). The specification
also allows for discs holding 25 or 23.3GB with relaxed pit-length requirements.
Although initial products will likely only play commercially recorded material, Blu-ray
technology allows for erasing and rewriting discs, much as computers already do with
DVD-Audio
CDs. Existing DVD drives will not be able to play Blu-ray media, although Blu-ray drives
should be backward compatible with current formats.
The Blu-ray name explains the secret to the new technology. The laser that reads the disc
uses shorter wavelengths (at 405 nanometers, its closer to violet than blue) that can pick
out digital dots that are closer together than can the yellowish laser used by DVDs or the
red laser used by CDs. The tracking pitchthe distance between adjacent grooves in the
single-track spiralis only 0.32 micrometers, and the shortest data pits are only 0.138
micrometers long.
All Blu-ray discs use a phase-change medium coated with a 0.1 millimeter optical transmittance protection layer. To further ruggedize the discs, they are housed in a tough
plastic cartridge that measures 5.1 by 5.2 inches and about one-quarter-inch thick. (Exact
dimensions: 129 by 131 by 7 millimeters.)
Blu-ray is specifically aimed at putting high-definition television onto DVD-style discs.
The system encodes video into MPEG-2 data streams and writes data at 36Mbps. A single
disc can hold over two hours of HDTV or 13 hours of VHS-quality video at a 3.8Mbps
rate.
As of this writing, Blu-ray is not yet an official standard. The optical community is currently exploring an alternate blue-laser system that uses a 0.8-millimeter protective layer.
689
PART 5
Human Interface
Cyborgscreatures half man, half machine that inhabit the nether reaches of space and the
imaginations of viewers of Star Trek: The Next Generationdiffer from the union of you and
your computer only in a matter of degree. They represent the intimate interconnection of
human neurons and post-silicon circuitry, where human-like bodies receive direct signals
from a master wireless network, including modest directives of the nature to subjugate,
enslave, assimilate, or destroy Earth. Cyborgs are most scary because they are imaginable,
understandable, andwe shudder with growing awarenesspossible sooner than we think.
Scary, too, because youre already directly communicating with your computer. Youve used
your thoughts to guide and control your computer, to tell it what to do. And it commands
you, too. Think of that last trip you took when you surfed over to Mapquest.com for directions. A giant computer network told you what to do. Although you were free to ignore the
commands, you probably followed them for fear of hearing your wheels slowly grinding
themselves down to the axles in mud, lost in a world so dark, gloomy, and far from civilization that even your cell phone wont work.
The difference between you at your computer and a cyborg is only the communication channel and code. The cyborg technology of science fiction has mastered the interface between
biological electrical circuitry and electronics, and it has cracked the code your body uses for
nerve impulses and the thoughts inside your brain.
The first time you tried to communicate with your computer such a direct connection probably sounded pretty good, especially if you had no previous skills with a keyboard. Direct
thought control of your computer would eliminate all of that finger-torture and even speed
everything upat least until some of your inner thoughts started to muck things up. Worse
yet, your brain would be flooded by the computers responses to your commands directly
altering your own thoughts. Certainly it would be great for the computer to fill your brain
with the right directions when you want to go someplace, but the computer might just as well
flood your gray matter with the belief that you actually wanted to go somewhere. It might
even make you think youre already there or create a world of entire fantasy. Such ideas have
given science fiction writers a field dayand a wealth of plots to explorewhich would be
wonderful if you werent the subject of the experiments.
In other words, the interface between you and your computer is yet another case of be
careful what you wish for. Todays communications systems are primitive, indeed, but they
work, and what you really want may not be what you really want. You may long for something betterthe grass is always greener in the other guys photo-editing programbut
todays connections between you and your computer do a pretty good job of bringing disparate worlds together.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
19
Principles: Connecting to You

Chapter 1, Computers, noted that programmability is one of the defining features of a computer. But you need some way of getting a program
into your computer. After all, the program is how you take command
and control your computer.
Engineers programmed the first computers with wires and plug-boards.
The wiring defined patterns that formed the digital code of the program.
Although such hard-wired programs are the ultimate in nonvolatile
memory, they are a bit inconvenient. Each bit in a program is a connection you have to make by hand, and all too often youre apt to make a
few mistakes as you plug in thousands of wires.
Next, engineers tried banks of switches to enter programs. Although
switches are a bit easier to change than hard wiring, they are just as
prone to error, and its time consuming to have to enter each bit, one at a
time.
Teletypes
Banks of switches were hardly the way to get big programs of hundreds
or thousands of bytes into a computer. What engineers longed for was a
device that could directly generate digital codes with a familiar interface,
a familiar means for people to use it. Fortuitously, exactly what they
needed was already widely used in the communications industry. The
teletype machine traces its roots back to 1902 when researchers finally
cracked one of the toughest problems plaguing the printing telegraph
since Samuel Morse created the first one in 1845. With the creation of
the start-stop code (which lives on today in the RS-232C port, see
694
Chapter 11, Ports) in 1908, Charles and Howard Krum produced the first practical
automatic printer connected to a telegraph line.
In 1919, the father-son duo created a keyboard-based transmitter, and the foundation for
the teletype system was in place. In 1925, they merged their interests with those of a rival
developer, Edward Kleinschmidt, and the company took the name Teletype Corporation
in 1929 and was purchased by the Bell System a year later.
Although the printer was the most important side of the invention for its creators, computer designers eagerly adapted the transmitter to their machines. The transmitter
became what is essentially the first computer keyboardyou typed into a typewriter-style
keyboard and produced digital code (in the form of a five-bit international code invented
by Emile Baudot in 1870 and named after him).
Keypresses on the teletype keyboard produced punched tape that, when fed into a reader,
generated the Baudot code. Makers of tabulators, such as IBM, found the same technology could make the punch-cards they used. This punch-card technology became the input
and output system of the first commercial computer, ENIAC. Two years later, the Binac
computer used an electrically controlled typewriter keyboard to write magnetic code on
tape.
By 1964, Bell Laboratories and General Electric created the first electronic computer terminal for the Multics computer system. With the introduction of this first video data terminal (VDT), all the intermediary steps between the keystrokes and final digital code
disappeared. Each keypress produced a code in electronic form that could be immediately
processed by the computer.
Regardless of whether punched tape serves as an intermediary or a terminal creates digital
codes directly, all these technologies owe one particular bit of heritage to the teletype. You
need to type in each letter the computer receives in exactly the same order you want the
computer to receive it. The stream of characters is a serial sequence. Communication is
reduced to a single, unidirectional stream, much like telling a story in a novel.
From the cyborg viewpoint, all this text comes from the language processing part of your
brain. Your ideas must be fully digested and turned into words, and the words into text,
before you communicate them with your computer.
This same technology survives today in all computers. The interface between man and
machine in this process is your computers keyboard. The first major personal computer
operating system used the keyboard as its primary (and only) input device. Your only
means of control and data input was the keyboard. Even the longest, most intricate programs required that someone, somewhere, enter all the commands and data used in writing them through the keyboard.
CHAPTER 19 PRINCIPLES: CONNECTING TO YOU
Teletypes
Pointing Devices
To many people, the keyboard is the most formidable and forbidding aspect of a computer. The keys might as well be teeth ready to chomp down on their fingers as soon as
they try to type. Typing just isnt something that comes naturally to most people.
Learning to type takes months or years of practicepractice thats about as welcome as a
piano lesson on a sunny afternoon when the rest of the neighborhood kids are playing in
the pool outside.
Imagine trying to drive your car by typing in commandsturn right 15 degrees, increase
speed to 55 miles per hour, stop before driving off that cliffoh, well. Brace yourself for
the landing. Although the keyboard provides an excellent way to move your properly formatted language-based thoughts into your computer, it is inefficient at more sophisticated
means of control. Even if you could type with your fingers moving at the speed of light,
control through the keyboard would still be slow because your brain needs to perform
some heavy-duty processing of the data first, at biologic rather than electronic speeds.
Teletype-style input poses another problem. Control systems built from it are command
driven. You have to type in individual commands, which in turn requires that you know
which commands to type. If you dont know the commands, you sit there with your fingers knotted and nothing happening.
One of Douglas C. Engelbarts jobs at the Stanford Research Institute between 1957 and
1977 was to find ways of making computers more accessible and usable by ordinary people. One of his ideas was to put graphics on the computer screen and use a handheld
device that would point at different places on the screen as you moved the device across
your physical desktop. He made his first model of the device in 1964 and received a patent
in 1970 for the first computer mouse. He called it that because it had a tail coming out
the end, the interface wire.
Engelbarts concept of combining the pointing device coupled with a graphical/menudriven onscreen user interface was later developed at the Palo Alto Research Laboratory
of Xerox Corporation in its Alto workstation in 1973. Xerox first commercially offered
these concepts in its 1981 Star workstation, which inspired the Apple Lisa and Macintosh
computers but was in itself not a marketing success.
The underlying concept was to allow you to indicate what function you want your computer to carry out by selecting from a list of commands presented as a menu. You point at
the menu selection by physically moving the pointing device, which causes a corresponding onscreen movement of the cursor. One or more buttons atop the device enable you to
indicate that you want to select a menu itema process much easier to do than describe.
The mouse was meant to be small enough to fit under the palm of a hand with the button
under a fingertip. The whole process of moving the mouse and its onscreen representation is termed dragging the mouse.
695
696
Apple Computer, understanding the achievements made at SRI and Xerox with the mouse
and graphical interface, incorporated both into its Macintosh computer in 1984. Although
you could obtain a mouse and software to use it for Intel-architecture computers at about
the same time, widespread use of a graphical interface did not become popular with Intel
machines until the introduction of Windows 95.
Graphic Input Devices

A mouse is an indicator rather than a full-fledged input device. When you want to put
graphic images into your computer, a mouse works only if you want to draw them anew.
Capturing an existing image is more complex because your computer needs a way to represent it.
That need had already been filled by bitmapped display systems (see Two-Dimensional
Graphics, later in the chapter), which break an image into visual units small enough that
they blend together in the eye to form a single, solid image. The units used to represent
the image are termed pixels, short for picture elements. The process of converting an
image or the representation of an object into pixels is called pixelization.
The process is not easy because it must deal with two worlds, the optical and electronic,
converting the optical representation of an image or object into an electronic signal with a
recognizable digital format. Several devices can make this conversion, including the scanner and digital camera.
The scanner can convert anything you have on paperor for that matter, anything reasonably flatinto computer-compatible electronic form. Dot by dot, a scanner can reproduce photos, line drawings, even collages in detail, sharper than your laser printer can
duplicate. Better yet, equip your computer with optical character recognition software,
and the images your scanner captures of typed or printed text can be converted into
ASCII files for your word processor, database, or publishing system. Just as the computer
opened a new world of information management to you, a scanner opens a new world of
images and data to your computer.
The digital camera captures what you see in the real world. It grabs a view of not only
three-dimensional objects but also entire scenes in their full splendor. As the name
implies, the digital camera is the computer equivalent of that old Kodak, one that produces files instead of film. It captures images in a flashor without one in bright daylightand requires no processing other than what you do with your photo-editing
software. It tops the list of most wanted computer peripherals because its not only a useful tool but also a neat toy that turns anyone into an artist and can add a good dose of fun
to an otherwise drab day at the computer.
Output
Output
At the other end of the computer, its output, where it needs to communicate to you, the
history is much the same as the input side. The first computers used hardcopy printed
output to express their answers or displeasure. Even Charles Babbage envisioned printed
paper as the basic output for his Analytical Engine. After all, paper has an enduring advantage. You can see your results for as long as you want, and more importantly, you can see
your mistakes. When you can hold the evidence of your mistakes in your hand, you can
repair your errors more quickly.
But putting all output on paper has a big disadvantage. If youve ever waited for an important job to print, you know the frustration of watching each line slowly appear down the
paper. Printing is a slow, mechanical process. Trying to run an interactive computing session while relying on your printer for output would be like running a relay race with a
glacier anchoring your team.
Much as the video data terminal changed computer input, it revolutionized computer output. Without it, interactive computing would be no more than a dream. But the computer
terminal didnt appear on the market like Athena, fully armed and prepared for battle. It
entered this world as a teletype machine that wrote with lighted phosphors instead of ink
(and at the same speed). Over the years, the computer terminal picked up momentum,
graphics, colors, and finally windows.
At first, the personal computer simply split the video system from the keyboard and
inserted the computer in the middle. Communications with the display followed the same
routine as the terminal and the teletypethe slow, serial stream of text characters. In a
quest for speed, engineers sidestepped their own best plans and revealed the magic of
direct access and painting pictures directly on the screen. Todays systems add a second
computer to your computer, one solely aimed at making images on your screen faster than
you can blink.
Teletype to Terminal
When engineers ushered in Harvard Mark I in 1943, television was little more than a
bluish dream in the minds of broadcasters, barely off to a flickery black-and-white start.
The first commercial licenses went into effect in July, 1941, and then went on hold for the
years of World War II. Putting text on the screen meant holding up a printed card. The
only output devices available to the first computer were mechanical, and the one most
readable by humans was printed paper, courtesy of the teletype machine.
The early connection between computer and teletypes lingers on in the lingo. Even today,
the data thats sent by a computer to its output device as a string of printable characters is
still termed teletype output. The character string, if converted to the correct code, would
run a mechanical teletype happily through reams of coarse paper.
697
698
The first terminals were little more than teletypes that printed on the screen instead of
onto paper. Essentially the same minimal design survives today at the dumb terminal,
named not for the engineers who design it but for how it processes what you see. A dumb
terminal is not smart enough to do data processing on its own. It puts each character on
its screen exactly as it is received through the umbilical cable linking it to its computer
host. Compared to the mechanical teletype, refinements are few. Instead of rattling off the
edge of the paper, a too-long electronic line more likely will wrap or scroll down to the
line below. The terminal never runs out of paperit seemingly has a fresh supply of blank
screen below, rolling upward as necessary to receive each additional line. But the output it
generates is even more tenuous than the flimsiest tissue and disappears at the top of the
screen, perchance never to be seen again.
In the electronic form of the computer terminal, the teletype method of text-handling
means that when one character changes on the screen, a whole new screen full of text
must be generated and sent to the terminal. The system cannot back up to change the one
character, so it must rush headlong forward, reworking the whole display along the way.
Add brains to a dumb terminal and you get a smart terminal. A microprocessor inside lets
the smart terminal recognize special commands for formatting its display and may even be
able to do some computer-like functions on its own.
BIOS Support
Computers act like dumb terminals in their most basic display functions. They put characters on the screen like teletypes to produce teletype output under the guidance of their
BIOSs. In fact, the basic computer BIOS gives several layers of teletype output. In the
most primitive, a program must load one character at a time into a microprocessor register and issue a video interruptspecifically, interrupt 010(hex). The BIOS then takes over,
instructing the microprocessor to check where to put the character (a several-step process
in itself) and then pushing the character into the appropriate place in memory. The BIOS
then returns control back to the program to process the next character. The most
advanced teletype mode lets a program put an entire line of text on the screen through a
similar, equally cumbersome process.
In basic teletype mode, characters are written on the screen from left to right, from screen
top to bottom, merely scrolling after each line is full or ended with a carriage return.
More advanced display technologies are able to write anywhere on the monitor screen
using formatting instructions much as smart terminals do. For example, commands in the
computer BIOS let your programs locate each character anywhere on the screen.
HTML
The great advance made in communications by the Internet has at its heart teletype technology. Your computer displays basic Web text in teletype styleone character at a time
Output
as if it came as part of a string from a teletype. The chief difference between what came
across the teletype wire and what comes through your Internet connection is the code.
Teletypes used Baudot; the Web uses Hypertext Markup Language (HTML), layered upon
the ordinary ASCII code that specifies individual letters.
HTML uses the processing power of the computer to improve on basic teletype technology. Certainly its quieter. Unlike a teletype machine, your screen does not chatter like the
teeth of a freezing epileptic. More importantly, the HTML character stream includes formatting codes that allow it to specify type size and faces, even colors. Although your computer still pops each character on your screen in the order it appears in the HTML string,
it can make changes in the appearance of each character based on the in-stream codes.
The codes act as switches. When one appears in the data stream, it switches on a text featurefor example, turning the text red. All later text gets displayed in red until another
HTML code turns it off or sets another color.
Some HTML codes affect the formatting of characters displayed on your computers
screen (for example, indenting lines). This formatting is relative rather than absolute. Its
effects vary depending on the size and shape of your Web display. For example, a simple
tab might take up half a column on a narrow window or a few character widths in a wide
one. In this way, HTML operates like a teletypethe characters in the HTML stream
come out without regard to the size or shape of the display medium.
More recent revisions to HTML allow the Web to go beyond minimally formatted text.
Tables allow you to create more structure in your text displays. In addition, you can
embed streams of images with your text characters. Even these elements, however, are
handled like the letters and numbers relayed to an old teletype display. In other words, old
ideas dont die. They just go online.
Character Technologies
The most notable aspect of teletype technology, be it on your screen or on the Web, is
that it is character oriented. The smallest unit of information it deals with is a text character. At the times when your computer steps back to this technology of yesteryearfor
example, the first few moments of booting up, before it spins its disks to load the basic
code of your operating system, or when you read your emailyour computer lets you see
its world only in terms of letters and numbers.
Somehow your computer must organize the characters it wants to display. Teletype
machines handle this matter mechanically. They simply stamp each character on a piece of
paper. The typing automatically organizes the text and the inky image preserves it. Your
computer isnt so lucky. It has to organize the text in electronic form, it has to be able to
move the character representations around inside its circuitry, and it has to keep the characters you see glowing on a screen that would rather remain permanently black.
699
700
Your system has two alternative ways of dealing with the characters it wants to put on
your screen. It can store and manipulate them in the form of individual characters, or it
can break the characters down into tiny pieces called a bitmap, in which each piece representing one of the bright dots appearing on your computer display. Your computer uses
both of these technologies every day. Although Windows treats your entire display screen
as one giant bitmap, applications running under Windows often send information to the
operating system to display in character form.
Character-Mapping
When your computer deals with characters as a fixed unit, one code per indivisible character, it uses a technology termed character-mapping. The name refers to the character map,
a special range of addresses thats sometimes called screen memory or display memory. The
memory of the character map is reserved for storing the characters that will appear on the
screen. Simple programs such as your computers bootup BIOS routines write text on the
screen by pushing bytes into the proper places in that memory. Just as a street on a
roadmap corresponds to the location of a real street, each byte of display memory corresponds to a character position on the screen.
The most common operating mode of the character-mapped display systems used by
computers when they boot up divides the screen into a matrix (essentially a set of pigeonholes, with each hole corresponding to one position on the screen) that measures 80 characters wide and 25 high. To display a character on the screen, a program loads the
corresponding code into the memory location associated with its matrix cell. To put the
image on the screen, the display system reads the entire matrix, translates into a serial data
stream that scans across the monitor screen, and moves the data to the video output. In
other words, it creates the exact bit-pattern that will appear on the screen on the fly, computing each nanosecond of the video signal in real time. From there, the signal is the
monitors problem.
For your programs, writing characters to the screen is simply a matter of writing directly
to screen memory. Consequently, this display technique is often called direct writing. It is
the fastest way to put information on a computer screen. Character-mapping is also more
versatile than teletype technology. Programs can push characters into any screen location
in any order they pleasetop, bottom, left, or right, even lobbing one letter atop another,
overwriting the transitory existence of each.
Once an advanced operating system such as Windows loads, however, your computer
steps away from character-mapping. The operating system imposes itself between your
programs and the BIOS. The operating system captures the characters your text-oriented
programs attempt to fling directly at the screen. The operating system can then recompute the map, making it larger or smaller and, in the latter case, moving it to a designated
area of the screen.
Output
Direct Writing
This quick way of putting characters on the screen is often termed direct writing, because
programs simply move the code assigned to a character directly to the memory location
corresponding to that characters screen positiona one-step process that requires only
one microprocessor instruction.
Direct writing makes a specific demand from programs using it: They need to know the
exact location of each screen memory address. For all applications to work on all computers, the addresses used by each system must be the sameor your software needs some
means of determining what addresses your computer uses. In the first personal computers,
the design engineers reserved two memory areas to serve as character buffers, one for
color text and one for monochrome. These memory areas became standard for all computers, and your system still uses them for the text it displays on the screen while it is
booting up. To determine which character buffer to use, your software can check the video
mode flag at a special location in memory, which indicates the proper buffer. Although this
dual-buffer feature remains part of every personal computer, its hardly necessary because
of the universal use of color display systems.
Character Boxes
In text modes, the display memory addresses hold codes that have nothing to do with the
shapes appearing on the monitor screen except as a point of reference. The actual patterns
of each character that appears on the screen are stored in a special ROM chip called the
character ROM thats part of the video circuitry of the computer. The code value that
defines the character is used by the video circuitry to look up the character pattern that
matches it. The bit-pattern from the character ROM is scanned and sent to the screen to
produce the final image.
Modern display adapters allow you to download your own fonts (typefaces) into onboard
RAM thats reserved from the same block that would serve as the character map. These
downloaded fonts can be used as if they were located in ROM with the same ease of
manipulation as ROM-based fonts. Downloaded fonts appear just the same, whether
pushed on the screen through the teletype or direct-access technique.
Each onscreen character is made from an array of dots, much like the text output of a
teletype or dot-matrix printer. Computer and display adapter manufacturers use several
video standards to build individual characters out of different size dot arrays. The framework in which the dots of an individual character are laid out, called the character box, is a
matrix like a crossword puzzle. The character box is measured by the number of dots or
cells composing its width and its height. For example, Figure 19.1 shows a series of characters formed in character boxes measuring 15 by 9 cells.
701
702
The text modes used by various early display standards all had their own, distinctive character boxes. The standard Video Graphics Array (VGA) text screen uses a 16-by-9
character box. Each character takes up a space on the screen measuring 16 dots high and 9
dots wide.
FIGURE 19.1
Characters formed in
15-by-9-cell boxes.
The last vestige of character mode display technology that remains under Windows is the
DOS box, or command mode. You can select the height and width of the character box used
in the command mode to adjust the size of the windows in which your character-based
applications run in text mode using the properties sheet for command mode.
The size of the character box does not exactly describe how large each character is or how
many dots are used in forming it. To improve readability, individual characters do not
necessarily take up the entire area that a character box affords. For instance, text characters on most monochrome displays keep one row of dots above and one below those used
by each character to provide visible separation between two adjacent lines of text on the
screen.
Video Attributes
The character-mapped displays of most computer video systems do not store each letter
adjacent to the next. Instead, each onscreen character position corresponds to every other
byte in screen memory; the intervening bytes are used as attribute bytes. Even-numbered
bytes store character information; odd bytes, attributes.
The attribute byte determines the highlighting or color of a displayed character thats
stored in the preceding memory byte. The codes used in monochrome and color displays
are different. Monochrome characters are allowed the following attributes: normal, highlighted (brighter onscreen characters), underlined, and reverse-video characters (dark on
light instead of the normal light on dark). The different attributes can be combined,
although in the normal scheme of things highlighted reverse-video characters make the
character background brighter instead of highlighting the character shape itself. These
monochrome display attributes are listed in Table 19.1.
Output
TA B L E 1 9 . 1
Monochrome Display Attributes
Byte Value
Attribute
00
Non-display
01
Underline
07
Normal
09
Intensified underline
0F
Intensified
71
Reverse video underline
77
Reverse video
79
Reverse video intensified underline
7F
Reverse video intensified
81
Blinking underline
87
Blinking normal
89
Blinking intensified underline
8F
Blinking intensified
F1
Blinking reverse video underline
F7
Blinking reverse video
F9
Blinking intensified reverse video underline
FF
Blinking intensified reverse video
Color systems store two individual character hues in the attribute byte. The first half of
the byte (the most significant bits of the digital code of the byte) code the color of the
character itself. The latter half of the attribute (the least significant bits) code the background color. Because four bits are available for storing each of these colors, this system
can encode 16 foreground and 16 background colors for each character (with black and
white considered two of these colors). In normal operation, however, one bit of the background color code indicates a special character attributeblinking. This attribute allows
any color combination to blink, but it also cuts the number of hues available for backgrounds in half (to eight colors, with all intensified color choices eliminated). When you
or your software needs to be able to display all 16 background colors, a status bit allows
the character-flashing feature to be defeated. Color display attributes are shown in
Table 19.2.
703
704
TA B L E 1 9 . 2
Color Display Attributes
Nibble Value
Foreground Color
Background Color
Flashing
Black
Black
No
Blue
Blue
No
Green
Green
No
Red
Red
No
Cyan
Cyan
No
Magenta
Magenta
No
Brown
Brown
No
Light gray
Light gray
No
Dark gray
Black
Yes
Bright blue
Blue
Yes
Bright green
Green
Yes
Pink
Red
Yes
Bright cyan
Cyan
Yes
Bright magenta
Magenta
Yes
Yellow
Brown
Yes
White
Light gray
Yes
Because each character on the screen requires two bytes of storage, a full 80-character column by 25-character row of text (a total of 2000 characters) requires 4000 bytes of storage. In the basic computer monochrome video system, 16KB are allotted to store
character information. The basic (and basically obsolete) color system reserves 64KB for
this purpose.
Video Pages
The additional memory does not go to waste, however. It can be used to store more than
one screen of text at a time, with each separate screen called a video page. Either basic
video system is designed to quickly switch between these video pages so that onscreen
images can be changed almost instantly. Switching quickly allows a limited degree of animation. The technique is so useful that even todays most advanced 3D graphics boards
use it, although with pictures instead of text.
Output
Two-Dimensional Graphics
The only graphics available on the first computers were block graphics, akin in more than
name to the first toys of toddlers, mere playthings that you wouldnt want to use for serious work. The first computer graphics systems made television look good, which in any
other context would be an insurmountable challenge to the imagination. The foundation
of the new display systemscalled bitmapped graphicsproved powerful enough that in a
few years computer display quality not only equaled that of television, but computers were
also used in making television images. The modern computer graphics system has taken a
further step beyond and attempts to build a real (or real-looking) three-dimensional reality.
The development of computer graphics is best described as accumulation rather than evolution. Each new system builds upon the older designs, retaining full backward compatibility. Even the latest 3D graphics systems retain the ability to work with the first
rudimentary block graphics. Just as you share genes with some of the lowest forms of life,
such as bacteria, planaria, and politicians, your sleek new computer comes complete with
state-of-the-art 1981 graphics technology.
Block Graphics
You dont need a lot of computer power and an advanced operating system to put graphics
on your screen, which is good because in the early years computers didnt have a lot of
power or decent operating systems. In fact, even teletypes that are able only to smash
numbers and letters on paper can print primitive graphic images. By proper selection of
characters, standing far from printouts, and squinting, you could imagine you saw pictures
in some printouts (a triangle of text might vaguely resemble a Christmas tree, for example).
When computers operate like teletypes, their graphic output faces the same limitations as
printoutscharacters can only approximate real-world images. To try to improve matters,
the designers of the original computer took advantage of the extra potential of storing
characters as byte values. Because one byte can encode 256 different characters, and the
alphabet and other symbols total far short of that number, the first computers designers
assigned special characters to some of the higher-numbered bytes in its character set.
Beyond dingbats and foreign language symbols, a few of the extra characters were
reserved for drawing graphic images from discrete shapes and patterned blocks that partly
or entirely fill in the character matrix.
When your computer is operating in text mode, such as in command mode, you can still
create rough graphic images by strategically locating these character blocks on the screen
so that they form larger shapes. Other extra characters comprise a number of single and
double lines as well as corners and intersections of them to draw borders around text
areas. The characters are building blocks of the graphic images, and consequently this
705
706
form of graphics is termed block graphics. Figure 19.2 shows the block graphic characters in
the standard computer character set.
FIGURE 19.2
Standard computer
block graphic
characters.
To a computer display system, block graphics are considered text and are handled exactly
like ordinary text characters. All the text attributes are available to every character of
block graphics, including all the available text colors, highlighting, and reverse video
characteristics. The characters are also pushed onto the screen in text mode, which gives
them high-speed potential, but they are available only in text mode or the Windows DOS
box. Because they use the high-order ASCII charactersforeign territory for most sevenbit email systemsyou cannot ordinarily use them for images in ordinary email.
Bitmapped Graphics
Windows marked the transition of the primary operating mode of computer display systems. From character-based displays, Windows ushered in the age of the bitmapped display. Bitmapped graphics improve the poor quality of block graphics by making the blocks
smaller. The smaller the blocks making an image, the finer grain that can be shown, and
therefore more detail. Physical aspects of the display system impose a distinct and
unbreakable limit on how small each block can bethe size of the individual dots that
make up the image on the video screen. The sharpest and highest-quality image that
could be shown by any display system would individually control every dot or pixel on the
screen.
The most straightforward way of handling the information to be displayed on such a
screen is to assign some part of memory to each pixel, just as two bytes are given over to
each character of a character-mapped display. In the computer scheme of things, because
the data controlling each pixel is stored as one or more memory bits, this kind of display
system is often called bitmapped graphics. Alternately, because each pixel or point on the
video screen can be separately addressed through memory, this method of controlling the
video display is often called all points addressable graphics or an APA display.
In the bitmapped graphics system, display memory stores an exact electronic representation of the onscreen image. Its actually a time-slice of what you seethe software running on your computer is constantly sending new data into display memory to update the
screen image. The memory temporarily stores or buffers the changes frame until they are
read out as a complete image frame dozens of times per second. Because of this function,
graphics display memory is often called the frame buffer.
Output
As with character-mapped memory, programmers have their choice of methods to write

bitmapped graphics to your monitor screen. The BIOS can do it, but Windows substitutes
its own drivers for the purpose. Programs can also write directly to memory, but modern
video systems make that chancy because each display system uses its own memory locations, display modes, and resolutions. Video drivers bridge these differences.
Bitmapped graphics hold the potential for being much sharper than block graphics. More
pixels mean more detail. The number of dots on a screen and the ultimate number of pixels are many times the number of characters that are displayed on that same screen, from
64 to 126 times greater. However, bitmapped graphics imposes its own, interrelated penaltiesmemory usage and speed.
The amount of memory required by a graphics display system depends on two factors: the
sharpness of the display image and the number of colors (or gray levels) to be displayed.
Each increase in sharpness and number of colors means that your computer is putting
more detail (more information) on its screen and storing more information in its display
buffer. As a result, it must move around more information (more bytes), which means
more work. And the harder your computer works, the longer it takes to complete its job.
Vector Graphics
Bitmapped graphics are sometimes termed raster graphics because the technology organizes the screen into a series of lines called a raster thats continually scanned dozens of
times a second. Although raster graphics are the basis of all computer displaysas well as
todays television and video systemsthis is not the only way to put a computer image on
a monitor. A completely different technique does not regularly scan the screen at all.
Instead, it precisely controls the circuitry operating the horizontal and vertical deflection
yokes. It doesnt trace scan lines but instead draws figures the same way you would as a
series of strokes of a paintbrush. To keep the screen lit, it constantly retraces the figures.
Because the signals controlling the monitor drive the electron beam in the CRT as a
series of vectors, this image-making technique is usually termed vector graphics. Alternately,
this kind of display system is sometimes called a stroker because of the kinship to drawing
brushstrokes. Although not used on computers, the term pops up occasionally in the
descriptions of expensive computerized workstations.
Resolution
The number that quantifies the possible sharpness of a video image is called resolution. It
indicates how many individual pixels an image contains that your display system will
spread across the width and height of the screen. Because your computer generates the
image as an electrical signal completely independent from your computer monitor (it
would make the same image even if your monitor wasnt connected to the computer at
all), physical properties of the monitor (such as its physical dimensions) play no part in
resolution measurements. In other words, the number of pixels in an image does not vary
707
708
with the size of the screen that it is displayed upon. Resolution is expressed without reference to units of linear measurementresolution is described in pixels or dots rather than
dots per inch. For example, the computer minimal resolution mode displaycalled VGA
(for the circuitry of the first display system to use it, the Video Graphics Array)has a resolution of 640 pixels horizontally by 480 pixels vertically in its native graphics mode.
Todays typical display has at least 1024 pixels horizontally and 768 pixels vertically.
Dots per inch is a measure of actual onscreen sharpness, and it depends both on resolution
and the size of the resulting image. At the same resolution, a larger screen has less sharpness than a smaller screenall else being equal.
The higher the resolution of an image, the more pixels it will contain. The more pixels,
the more memory needed to store them.
Graphic Attributes
How much memory is required for a given resolution depends on a second factor in addition to the number of pixelsthe number of bits assigned to each pixel. At minimum,
each pixel requires a single bit of storage. That bit can be used to code either of two conditionswhether the pixel is illuminated on the screen or invisibly dark. In the simplest
bit-image graphics system, one bit of memory would then be used to map the condition of
each pixel in the video display.
Whats lacking from this primitive mapping system is contrast and color. All bits are
treated the same and their associated pixels look about the same, either on or off. The
result is a single-hued picture with no variation or shading, essentially the same sort of an
image as a line drawing. Although that may be sufficient for some purposes (for instance,
the display of a chart or graph that mimics the monochrome look of ink on paper), color
and contrast can add impact.
The way to add color to bitmapped images is much the same as adding color to characterbased displaysadding attribute information. Additional memory is devoted to storing the
attribute of each bit. The bitmapped system works somewhat differently from the character-based mode, however. All the memory devoted to a pixel is used to describe it. No
information needs to be devoted to identifying a character or pattern for each picture element because each one is essentially a featureless dot.
Color Planes
A single bit per pixel results in what graphics folk call a two-color system, because it puts
everything in black and whiteeach pixel is either on or off. Putting more color in the
image requires encoding more informationmore bits and more memory. Adding a second bit per pixel doubles the number of possible displayable colors. (Shadesdegrees of
darkness or lightare considered different colors in the terminology of computer graphics.) Every additional bit assigned to each pixel likewise doubles the number of possible
colors. Hence, with n bits, 2n colors are possible.
Output
In computer graphics, the number of bits assigned to coding color information is sometimes described as the number of color planes. This term relates to the organization of display memory. The memory map of the graphic image can be visualized much like a
Mercator projection of the world, with latitude and longitude lines corresponding to the
different positions of the bits corresponding to pixels in the image. Additional bits per
each pixel add a third dimension, much like layers of maps stacked atop one another, a
series of flat planes containing the color information.
With more colors used in an image, the better the apparent image-quality and the more
life-like its appearance. For this reason, the temptation is to increase the bit-depth of each
pixel as high as possible. However, the more colors or color planes, the more storage is
needed for encoding each pixel. Moreover, much to the dismay of purveyors of video
memory, the human eye is limited in its ability to resolve individual colorsmost people
can distinguish only a few million distinct colors. Color monitors are even more limited in
the number of colors they can display. Most monitors top out at about 262,144 colors,
corresponding to the capabilities of an 18-bit display system. Once these limits are
reached and enough memory is assigned each pixel, further improvements do not
improve appearances.
The practical limit on color is a bit-depth of 24 bits, which allows a system to store and
theoretically display any of 16,777,216 hues. Display systems with this bit-depth are
termed 24-bit color or True Color systems, because they can store sufficient information to
encode more colors than anyone could possibly seethey hold a truly accurate representation of any color.
Although some of the capabilities of True Color display systems are superfluous because
they exceed the human ability to distinguish colors, True Color is a convenient system for
designers because it assigns one byte of storage for each of the three additive primary
colors (red, green, and blue) to each pixel. Before memory became cheap and processors
raced beyond a gigahertz, this three-byte-per-pixel memory requirement was a burden on
computer systems, straining both speed and storage.
Some newer display systems have a 32-bit color mode. Instead of allocating the additional
byte of storage to color information, however, most of these 32-bit systems put the extra
bits to work as an alpha channel. The bits in the alpha channel hold control rather than
color information. In effect, the alpha channel provides a storage place for special-effects
information. The bits in the alpha channel normally are not tallied in counting color
planes.
The newest scanners may have 48-bit color capabilities. The extra bits are for colors you
cannot see but could. When you want to adjust an image (for example, to brighten a dark
scene), the extra bits ensure that even dark shadows have color differences. Brighten the
image, and the shadows will have detail instead of being big, black blobs.
709
710
The math for finding the amount of memory required to display a color graphics screen is
straightforward. Simply multiply the number of pixels on the screenthat is, the resolutionby the bit-depth of each pixel. Then divide by eight to translate bits into bytes. For
example, an XGA screen with 1024 by 768 pixels comprises exactly 786,432 pixels. If you
want True Color (one byte of storage per color, or a total of 24-bits per pixel), you need
18,874,368 bits to store the image datathats 2,359,296 bytes. A display adapter with
only 2MB wont handle that combination of colors and resolution, but (rounding up to
the next generally available increment) a board with 4MB will do nicely.
Table 19.3 summarizes the memory required for common resolutions at various bit-depths
for standard business graphics.
TA B L E 1 9 . 3
Display Memory Required for Given Resolutions and Color Depths
Resolution
Mono
16 Colors
256 Colors
High Color
True Color
Bits per pixel
16
24
Bytes per pixel
0.125
0.5
640480
38,400
153,600
307,200
614,400
921,600
800600
60,000
240,000
480,000
960,000
1,440,000
1024768
98,304
393,216
786,432
1,572,864
2,359,296
1152864
124,416
497,664
995,328
1,990,636
2,985,984
12801024
163,840
655,360
1,310,720
2,621,440
3,932,160
16001200
240,000
960,000
1,920,000
3,840,000
5,760,000
19201340
321,600
1,286,400
2,572,800
5,145,600
7,718,400
20481536
393,216
1,572,864
3,145,728
6,291,456
9,437,184
The values shown in this table refer only to two-dimensional display systems. The threedimensional systems that are becoming popular gorge themselves on memory. Most have
two frame buffers to take advantage of the double-buffering and use additional memory in
rendering the imagefor example, to serve as a z-buffer. (Double-buffering and z-buffering are both discussed in the section later in this chapter titled Three-Dimensional
Graphics.) More memory is required for describing objects and even stuff you never see.
Color Coding
The best and worst display systems assign the same number of bits to each of the three
primary colorsa bit or an entire byte. For intermediary color depths, however, the base2 digital nature of the computer and the three-fold nature of color vision come into direct
collision. For example, if you want to assign a single byte to store the colors of each pixel,
Output
how can you evenly allocate eight bits among three colors? With two bytes per pixel, how
do you divide 16 by 3 evenly?
You dont. But you dont have to. You dont even have to code colors as a mix of red,
green, and blue.
Because the human eye is most sensitive to green and its shadings (probably something to
do with primitive humans living in an environment lush with chlorophyll-green plants),
some color-coding systems split their bit assignments evenly and assign the odd bit to
green. For example, when the first computer engineers designed a 16-bit VGA color system, they assigned five bits to red and blue and gave six to green.
Color Spaces
In mixing colors to produce a full spectrum, the number of hues any system can produce
is limited by the range of its medium. Because most systems of colors involve three signals, such as the three primary colors of light, mapping them requires three dimensions,
which in turn defines a volume or space. The range of colors that a specific system can
handle is consequently called the color space. In the RGB system used by computer monitors, the range of red, green, and blue signals define the three dimensions of the color
space, but the RGB system is only one possible color space.
Other image systems encode colors in an entirely different manner. One of the most common is to encode colors by brightness (technically called luminance and abbreviated as Y)
and two color (or chromaticity) values (abbreviated C1 and C2) that essentially correspond to coordinates on a map of colors. In more general form, the signals used are
termed Y, Cr, and Cb. In effect, Y is the luminance or overall image brightness, derived
by totaling together all three of the RGB signals. Cr represents the difference between the
red and luminance signal. Cb represents the difference between the blue and the luminance signal. Subtract Cr from Y and you get the red signal. Subtract Cb from Y and you
get blue. Subtract both the resulting red and blue signals from Y and you get green.
A related system used by composite video signals transforms the YCrCb signals into
intensity, hue, and saturation. In the intensity/hue/saturation color space, intensity
describes one axis of the color space. Saturation is defined as the distance from the intensity axis, and hue is the direction from the axis. Together these three measurestwo distances and an angledefine the three dimensions of the space. In effect, hue is
represented as an angle in the phase relationship between two signals.
In conventional video signals, as used by televisions, VCRs, and other video equipment,
the intensity information is called the luminance signal, and it takes the form of a standard
black-and-white television signal. The saturation information is represented as the corresponding amplitude of an added signal called the color burst. The phase relationship
between the luminance signal and the color burst defines the hue. In this way, the three
dimensions of the video color space are cleverly encoded in two signals.
711
712
Commercial printers use a different color scheme to start, the so-called process colors of
cyan, magenta, yellow, and black (often abbreviated as CMYK). These colors correspond
to the ink colors used in the four-color printing process. The black adds depth to the
printed colors.
These different color spaces and coding methods are useful to particular output devices
CMYK colors for storing images that eventually will be printed and published; luminance
and chrominance coding for images that will eventually be used in broadcast-style (as
opposed to computer) video systems. To be displayed by normal computer monitors, they
must be translated from their native format to the RBG signals used by computer monitors.
Color-Mapping
Another method of encoding colors in memory that requires translation has found greater
use in computers, particularly when a manufacturer wants to scrimp on video memory. A
technique called color-mapping stores only code numbers for colors, allowing each code
number to stand for almost any color. One code number gets stored in video memory for
each onscreen pixel. The display system matches the stored numeric code values to a
Color Look-Up Table (CLUT) that tells which color corresponds to each number; then
that color is sent along to the monitor. Because of the nature of colors in the real world,
color-mapping can lead to substantial economies in display memory.
When the values stored in screen memory directly indicate what color appears on the
screen, as they do in the preceding example, the colors are said to be direct mapped.
Direct mapping allows any pixel to be any color, but most images are made from far fewer
colors. By mapping the colors, you can trim the amount of memory used to store an
image. Although memory is not so much an issue with computers as it used to be, file size
remains a consideration when shipping images across the Internet. Consequently, most
images use some kind of mapping when they are stored. Most image-compression
schemes use some form of color-mapping. The GIF (which stands for Graphic Interchange
Format) files widely used on the Web are color-mapped to use only eight bits of storage
for each pixel.
Graphic Commands
The secret weapon of achieving high speed in any graphics system is the high-level
graphic command. By combining all the tiny steps of complex screen operations into a
single routine, some of the most intricate onscreen images can be encoded as a few commands. Not only do high-level graphic commands make programs more compact, they
allow your computers microprocessor to offload the work of building images in the frame
buffer.
Your microprocessor uses the high-level graphic commands to send instructions to a
graphics subsystem. In todays computers, that means a graphics accelerator chip or a 3D
Output
accelerator. The accelerator executes the commands to move pixels around in the frame
buffer.
The range of these graphic commands is large. Each accelerator chip has it own repertory
of them. The most common among them are detailed in the following subsections.
Bit-Block Transfers
Bit-block transfers are instructions that tell the graphics chip to move data from one place
to another in display memory. Instead of moving each byte of screen data through memory, the microprocessor only needs to tell the graphics chip what block to move (the
source of the data) and where to put it (the destination). The graphics chip then carries
out the entire data-transfer operation on its own.
Often shortened to BitBlt, bit-block transfers are most commonly used for scrolling an
image up the screen. You can easily see the effect the command makes in video performance. When you scroll a bit-image up the screen using a graphics chip, the top part of
the image often snaps into its new position, leaving a black band at the bottom of the
screen that slowly fills with the remainder of the image. The initial quick move of the top
of the image is made entirely in display memory using BitBlts. The rest of the image must
be brought into display memory through the I/O bus or local bus, thus resulting in delays.
Drawing Commands
Drawing commands tell the graphics chip how to construct part of an image on the
screendrawing a line, rectangle, or arc, or filling a closed figure with a solid color or
pattern. Often called graphic primitives, these commands break the image into its constituent parts that can be coded digitally to build a shape on the screen.
Before your computers microprocessor puts a line on the screen, it first has to compute
where each bit of the line will appear. It must compute the coordinates of each pixel to
appear on the screen and then transfer the change into display memory across the bus.
With a graphics chip, the microprocessor only needs to indicate the starting and ending
points of a line to the chip. The graphics chip then computes the pixels and puts the
appropriate values in display memory.
Sprites
Sprites are small images that move around the screen as a unit, much like an onscreen
mouse pointer. General-purpose microprocessors have no provisions for handling sprites,
so they must compute each bit of the sprite image anew every time the sprite moves across
the screen. Many graphic chips have built-in capabilities to handle sprites. They store the
bit-pattern of the sprite in memory and only need instructions telling them where to
locate the sprite on the screen. Instead of redrawing the sprite, the graphics chip need
only change the coordinates assigned its onscreen image, essentially only remapping its
location.
713
714
Windowing
Windowing is one of the most common features of todays graphic operating systems. Each
task is given an area of the screen dedicated to its own operations and images. Keeping
straight all the windows used by every task is a challenge for a general-purpose microprocessor. Graphics chips and 3D accelerator chips, however, are usually designed to manage windows using simple commands. Once an onscreen window is defined, it can be
manipulated as a single block rather than moving individual bytes around. The windowing
operations can be strictly software manipulations, or the graphics chip may include special
hardware provisions for streamlining the control of the windows.
In a conventional windowing system, software controls the display of each window. The
layout of the screen is calculated, and the proper values for each pixel are plugged into the
appropriate locations in the memory map. The image is generated by reading each memory location in sequence and using the information it contains to control the intensity of
the electron beam in the display as it sweeps down the screen. Every memory location is
scanned sequentially in a rigid order.
Hardware windowing works by slicing up the frame buffer. Although each dot on the
screen has one or more bits of memory assigned to it, the map no longer needs to be an
exact coordinate-for-coordinate representation of the screen. The video chip no longer
scans each memory location in exact sequential order as the video beam traces down the
screen. Instead, the memory scanned to control the beam is indicated by pointers, which
guide the scan between different memory areas. Each memory area pointed to represents
an onscreen window.
Each window can be individually manipulated. The memory used by a window can even
be mapped into the address range of the system microprocessor while the rest of the
screen is handled separately. As a consequence, most of the calculating normally required
to change a window is eliminated. Therefore, screen updates speed up substantially.
Panning
Hardware panning takes advantage of the some of the memory in the video system thats
not needed as a frame buffer. For example, your video board may have 2MB of memory
but uses only 1.5MB to store a full 1024-by-768-pixel image with 16-bit color. The extra
half megabyte of memory can hold an image thats bigger than that displayed on the monitor screenthe monitor image essentially becomes a window into display memory.
Instead of stretching out for 1024-by-768 pixels, for example, the extra display memory
might allow the filling of an 1152-by-864-pixel map. To pan the onscreen image one way
or another on the screen, the display circuits only need to change the address of the area
routed through the output of the board.
Changing addresses is much faster than moving blocks of bytes with BitBlt instructions, so
hardware panning takes place very quicklyas long as the video boards memory holds the
entirety of the image to be displayed. This technique is most useful when you have an
Output
older monitor thats not quite up to todays high scan ratesyou put as much image on
your screen as your monitor can tolerate while organizing your work across a larger desktop.
Depending on the hardware and drivers youve installed in your computer, Windows
allows you to adjust the size of your desktop and onscreen display independently through
the Settings tab of Display Properties.
Three-Dimensional Graphics
All the latest display adapters claim to have the capacity to display three-dimensional
images. In other words, they claim to be 3D boards. In terms of a display adapter, 3D
means more and less than it seems. 3D does not mean what it did in the 1950s when your
parents or grandparents put on blue-and-red glasses with cardboard frames to watch The
Creature from the Black Lagoon jump out of the movie screen at them. In computer terms,
that kind of 3D is a stereoscopic display.
In computer terms, 3D means adding simulated depth to the flat images on a monitor
screen. But a 3D graphics adapter does more than add the appearance of a third dimension. It also gives 3D motion to your images to add excitement to games and give attention-grabbing effects to your presentations.
Indeed, the motion part of the 3D boards job is its toughest assignment. Not only must it
make 3D images, but it also has to make them fast enough (at least 15 frames per second)
that you think they are moving rather than just flashing at you.
Generating those images takes more processing power than was available in a computer a
few years ago. The 3D boards have to move megabytes, even hundreds of them, every
second. They have to cope with mathematical functions you probably never knew existed
in high schooland frightened you into an Arts degree once in college. Yet they now
encapsulate all the power they need in todays graphics accelerators. Better still, most of
the math is gone. When developers write programs, they only need to call the advanced
functions built in to programming interfaces such as Microsofts DirectX.
No matter what you use your 3D board for, it needs to handle a few basic functions to
generate its illusion of 3D reality. These functions can take place in hardware or softwarethat is, your graphics accelerator can take care of the heavy-duty computing itself
or rely on your computers microprocessor to do the hardware and merely pass along conventional two-dimensional functions to your 3D board. Obviously, a 3D board is supposed
to do all the 3D work itself, but DirectX will take over if your board doesnt have the
hardware functions it needs.
Tessellation
Computers face a problem in creating and manipulating three-dimensional objects.
Because a computer works only with numbers, objects must be mathematically described
715
716
for the computer to have any idea of how to deal with them. Finding a set of equations to
describe a complex object is a daunting taskso daunting that programmers dont dare
face it. Instead, they break complex objects into easily describable pieces and use the computer to manipulate those pieces. What they do is make their images the same way a
mosaic artist makes a picture from small tiles.
The process of breaking the image into tile-like pieces is termed tessellation. The term
comes from the name given to one of the mosaic artists tiles, tessera (the plural form is
tesserae).
In 3D graphics, the computer manipulates each tessera individually. When its transformation is complete, the computer recombines all the tesserae to create the original object. To
eliminate the seams between the individual tessera and other artifacts of the manipulation,
the graphics system filters the resulting image combination.
In practical 3D display systems, the tesserae are polygons, usually triangles. The performance of 3D processors is often expressed in the number of polygons (or triangles) that
can be rendered in a second. The number has to be prodigious. A complex object may
require 20,000 or more polygons. Rotate it with 15 updates per second, and your graphic
chip has to render 300,000 polygons per second.
The polygons per second description of a product really doesnt reveal a lot. The rendering time for any polygon depends on its shading, texture, and other attributes.
Chipmakers rate their products with their theoretical capabilities, and you can guess
whether they would use best-case (monochrome, smooth surface triangles) or worst-case
(True Color, polygons with complex textures, filtering, and antialiasing) conditions.
Texture-Mapping
At this level, the individual tesserae of the image would have the appearance of armor
plates or a stealth fighter. Each one would be a single, solid, unrelieved color. Outside of
jousts and midnight bombings, reality rarely looks so armor plated. Real objects have
shading and texture that makes them look, well, realistic.
To add realism to the 3D objects your computer generates, the graphics system adds texture. Through a process called texture-mapping, a two-dimensional texture is applied to
the surface of each tessera, making up the image. In effect, the texture is glued to each
surface of the image.
In the texture-mapping process, the computer starts out with an image of the texture. The
complete texture is broken into small constituent parts, similar to the pixels in a video
image. Each of these pieces is termed a texel. The computer manipulates the texture by
rotating and scaling it and applying perspective techniques to make it correspond to the
manipulations performed on the tessera to which it is applied. The computer then maps
the texels to pixels.
Output
Because of the manipulations of the tesserae and texture, one pixel does not always correspond to one texel. Typically, the position of a given pixel appears between texels. Simply
applying the color of the closest texel to the pixel results in an unrealistic image. The
image may get blocky as multiple pixels take on the same color, or it may shimmer with
slight movements as pixel values shift dramatically between texel values. To prevent these
effects, the graphics system may apply bilinear filtering during the texture-mapping
process. The pixel color takes on a value intermediate between the colors of the four adjacent texels. The result is a more realistic appearance of the texture color, particularly as
the image is moved or rotated.
Most 3D systems store their textures in several different levels of detail for situations
requiring different 3D depths and resolutions. Typically, each level is designed for half the
resolution of the preceding one, so a collection would include textures at resolutions of
100, 50, 25, and 12.5 percent. Each texture level is prefiltered and tuned to best appearance at its resolution level. Taken together, the stored collection of a single texture at various levels of detail is termed a mipmap.
Objects can appear at any depth in the image, but the mipmap stores only discrete levels
of detail. At some depths the texture applied to a polygon will require a level of detail
between those in the mipmap. The 3D processor interpolates an intermediate texture
using a technique termed trilinear filtering. The processor first performs bilevel interpolation (bilevel filtering) and a further linear interpolation between the detail levels applicable on either side of the polygons depth.
Texture-mapping is one of the most complex of the operations required in three-dimensional graphics systems. Making the necessary calculations for mapping and filtering may
take up half or more of the processing time for the image. Consequently, texture-mapping
benefits greatly from hardware acceleration.
Depth Effects
What separates ordinary graphics displays from 3D displays is the appearance of depth to
the image. In a 3D image, some objects appear closer than others. The difference in depth
is, of course, an illusion because the face of the picture tube or LCD remains flat. The 3D
display system must fool the eye into believing it perceives depth.
The problem is not new to 3D display systems. Ever since the first cavemen scratched
walls with charcoal and ochre, artists have attempted to add depth to their paintings and
drawings. Over the years they discovered (and often rediscovered) the techniques of perspective and other means that take advantage on the depth cues the human eye uses to put
distant objects in their places.
Recession
One of the primary depth cues to the human eye is image size. The smaller an object of a
known size appears to be, the farther away it seems. Three-dimensional graphics systems
717
718
use a technique termed perspective divide to simulate this effect. To make a distant object
smaller, the graphics system scales distances in the x and y coordinates of the image by a
factor that is proportional to the z coordinate. In simplest form, distances in the x/y coordinate system are divided by the z valuethe larger the z value, the smaller the object
becomes.
The effect is the same as one-point perspective in the art world. Parallel lines, such as railroad tracks, converge at a single point in the distanceas z approaches infinity, the values
of x and y approach zero and the coordinates collapse into a point.
Atmospheric Perspective
Artists have long used the technique of atmospheric perspective to add depth to paintings.
Because of the effects of looking through a long reach of atmosphere, which isnt perfectly
clear because of the haze of dust and water vapor suspended in it, distant objects appear
paler and bluer than nearer objects. Artists capitalize on this visual effect and mix white
and a trace of blue with the color of objects that are supposed to appear in the distance in
their paintings.
The corresponding technique in three-dimensional computer graphics is termed fogging.
To make an object appear more distant, the graphic system adds a fixed color called the
fog color to the hue of the object. The amount of fog color added in increases with the
apparent distance of the object. Because the fogging technique corresponds to the appearance of natural haze, the technique is sometime called by that name, haze. This technique
is also called depth cueing.
Lighting Effects
Photographers and artists quickly learn that lighting is the key to making their twodimensional images appear to represent three dimensions. The pattern and depth of
bright areas and shadows allow the human eye to determine depth and the relative locations of objects. When an object sits between another object and the light source, the first
object is brightly lit and the second object is in shadow. The technique of rendering
images using the effects of light and shadow is called chiaroscuro, from the Italian for
clear-dark. Chiaroscuro rendering is one of the great challenges facing the 3D graphics
system.
Ray-Tracing
The most powerful and compelling way to render the lighting of a three-dimensional
scene uses the technique of ray-tracing. The computer follows, or traces, the path of every
light ray that impinges on the scene. The process is complex. The computer determines
which object each ray strikes based on the direction of the origin of the ray. After the ray
strikes the object, the computer determines how much of the ray illuminates the object
and is reflected. Based on the angle of the object surface to the light beam, the computer
determines a new path for the reduced beam and plots it to the next object, continuing
Output
until the beam is so diminished it has no further effect. Once all the rays have been traced,
the computer sums up the amount of light that has struck each surface. Those surfaces
with the most rays are the brightest; those with few rays are in shadow.
Of course, to make the math tractable, the computer deals with only a reduced set of rays
(the more, the better the rendering). The math is so complex that computers cannot perform ray-tracing on reasonably sized images in real time. The technique works extremely
well for static images (if youre patient) but requires each frame of an animation to be
individually rendered, taking perhaps a minute per frame. The ray-tracing technique consequently is not suitable to generating compelling 3D animations in real time.
Shading
Computer graphics systems use a simplified means of creating the chiaroscuro effect
called shading. In the simplest form, the computer determines the angle of a surface to the
light source and, using the reflectivity of the surface, computes how bright the surface
should be rendered. Gouraud shading, also known as smooth shading, takes an additional
step. It interpolates lighting values across the face of a surface to give gradual color transitions from a bright edge to a dim edge.
Z-Buffering
One way to eliminate the display of hidden surfaces is to track the depth of picture elements. The depthor the distance away from you the element is supposed to appearis
assigned a value. This value corresponds to the position of the element on the z-axis of
the three-dimensional coordinate system. In the z-buffering technique, this depth value
gets stored in a special z-buffer. As the graphic chip updates the image, it compares the z
value of each pixel to that stored in the z-buffer. When the z value in the buffer is less
than that of the newly rendered pixel, the old value is nearer and would obscure the new
value, so the new value is discarded. If the new value is lower and would thus appear in
front of the old value, the old value gets discarded, replaced by the new value.
Transparency
Objects in the real world are not always opaque; otherwise, seeing through the windshield
of your car would be a much greater challenge. To account for the effects of a transparent
object sitting in front of another object, many 3D imaging systems store a transparency
attribute for colors in addition to the normal red, green, and blue values. The storage for
this additional attribute data is termed the alpha channel. Commonly the alpha channel is
an extra eight bits added to the 24 bits used to store True Color pixel data, resulting in a
32-bit storage system (or four bytes per pixel). Most systems assign higher values to
greater opacity, so 255 would be totally opaque and 0 totally transparent.
When calculating the appearance of a given pixel, the graphics processor uses the alpha
channel values to determine its resulting color through a process termed alpha-blending.
The processor adds an fraction of the color of the transparent object set by the alpha
719
720
channel value to the color of the background object. Mixing the colors together gives the
appearance that the background color is seen through a layer of the transparent object.
By itself, of course, alpha-blending only changes the color of pixels. You need more to
give the true illusion of transparency. The viewers eye also takes cues from the shape of
the transparent object (defined by the color change made by the alpha-blending) as well as
the difference between the background color seen with and without the transparent mask.
The programmer must take all of these factors into account to produce a compelling
three-dimensional image.
Double-Buffering
To create smooth animation, 3D display adapters use the technique of double-buffering. As
the name implies, this technology puts two frame buffers in control of the graphics chip.
The front buffer corresponds to the traditional display buffer, connected to the rasterization circuitry that reads it sequentially and sends the video data to the screen. While the
front buffer is scanned, the graphics chip draws in the back buffer, where the rendering
operation is hidden and access is not limited by the timing of the rasterization process.
Once the graphics chip completes its drawing, the two buffers are switchedthe back
buffer becomes the front buffer and the image drawn in it gets sent to the display. The
graphics chip can then begin drawing in the new back buffer.
By cycling between the two buffers, double-buffering allows a fast but limited form of animation. It also achieves the same end as double-ported video RAM without using expensive, specialized memory chips. At the same time, it eliminates the appearance of the
drawing process on the screen.
Image Creation
The various hardware technologies are only tools used by your computers display system
in creating a bitmap of the screen image. Your softwarethat is, your computer game
first must generate the scene that will be displayed and make a new one as often as 30
times each second.
Generating the 3D images used in todays games is the most computationally intense task
your computer faces. In fact, the work involved in creating and updating all the information in a 3D graphic scene in real time is beyond the capabilities of even todays most
powerful microprocessors and display systems. Thats not to say the images you see on
your screen today are impossibleits just that they are not all they seem to be.
Of the many lessons the designers of 3D graphics have learned from Hollywood, the art
of building sets is probably the most important. Watch any movie epic, and youll see lavish scenes of grand structures, huge crowds, and endless panoramas, none of which ever
really existed. The panoramas may be painted, projected, or matted into the scene. The
crowds are photographed to make a few people seem like a mass, or the same
Output
supernumeraries are shot two or four times in every scene. And the huge sets and extravagant buildings are mere facades, a pretty face with nothing behind but bare two-by-fours
propping everything up. The important lesson is that only what shows counts.
Thats the secret that makes todays 3D graphics look as good as they do. When confronted with an overwhelming calculating chore, your graphics hardware does exactly
what any high-school student would do in the situation. It ignores most of it. But far from
blaming what it doesnt do on the errant ways of the family canine, it takes the approach
of the Hollywood set builder. It assesses exactly what will show in the final image and
concentrates its work on that. As far as the behind-the-scenes stuff goes, the software only
does whats necessary so that the whole set doesnt collapse around the stars. In other
words, writing good 3D graphics software is the art of knowing what not to do. Engineers
call this strategy culling, which means picking the best or most important.
Spaces
Unlike two-dimensional display software, which creates each scene exactly as you see it on
the screen, a 3D display system images a scene as objects. An object is anything that can
be manipulatedmoved, rotated, stretched, or squashedas a single entity. The software
describes each object numerically and transforms it from frame to frame using mathematics. Each object defines its own model space with its own coordinate system for these
manipulations.
The objects are mapped into the coordinate system of the world in which they act, corresponding to the Hollywood sound stage or location set. All objects in this world space use
the same coordinate system that defines exactly where they appear in the scene.
Once the software has put all the objects into place, it can generate a cameras-eye view of
the scene, like dropping the tripod legs of a Panavision camera to film the scene. The
camera sees the scene from its own coordinates, creating what engineers call the camera
space or view space. At this point, the software can trace the path of light rays from the
camera to distant parts of the scene. This view lets the software begin cheating. When a
ray hits a solid (as opposed to transparent) object, it can go no further. The software
knows that it can ignore anything from that point on, so it does not have to calculate
other objects or background textures that lie beyond that point.
This view space is actually encompassing a solid volume, a pyramid with the camera at its
apex, called the view frustum. The software transforms this view into clip space by distorting
the frustum into a cube. That is, it mathematically spreads out the pyramid. This process
distorts the objects in the scene, making those closer to the camera larger, which makes
the more distant objects appear smaller. This transformation yields perspective and a distance effect.
Finally, the software squashes the clip space cube into a two-dimensional view of the scene
that corresponds to what you would see on your video display, the screen space. The object
721
722
descriptions that survive culling and have been transformed into screen space get loaded
into the frame buffer and passed along to your computer display for your viewing.
3D Pipeline
Most 3D display systems process scenes in a 3D pipeline, much as a microprocessor executes programs. The pipeline allows the 3D accelerator to work on several steps of display
processing at the same time to speed up its operation. It starts out in software, typically
the game you are playing. The software defines the objects and whatever transformations
they require from one frame to the next. The software describes an object as a set of
points representing the vertices of triangles (triangles, because the three points making up
the vertices uniquely describe a plane surface). At this point, the object is nothing more
than a few points describing a wireframe made up of triangles. Even so, the software can
start the culling process, identifying the triangles on the back of the object, those that
wont appear in the final view.
The software can put the objects at this stage into their world space and construct a view
space, again culling the areas that will be in view. It will also define the lighting of the
scene and where shadows will fall.
At this point, the software is ready to develop the screen space. It now fills in the rectangles with the proper shades and textures. To add aerial perspective, it may also add fog to
distant objects. It may also smooth out jagged outlines. The software then moves the rendered image into the frame buffer (alternately, it may render the image directly in the
frame buffer). At that point, the image need only be rasterized by the video controller and
sent on its way to the monitor and then to your eyes.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
20
Keyboards
The primary input device for most computer systems is the keyboard,
and until voice-recognition systems are perfected to the point that they
can recognize continuous speech, the dominance of the keyboard is not
likely to change. Even then the keyboard will probably remain unapproachable for speed and accuracy for years to come. The keyboard also
is more suited to data entry in open offices, airplanes, and anywhere your
privacy is not ensured and your sanity not beyond reproach.
When youre buying a new computer, the keyboard is the least of your
worries. After all, the manufacturer takes care of it for you. Nearly every
desktop computer comes completely equipped with a keyboard (otherwise the computer wouldnt be much good for anything). Notebook
computers have their keyboards completely built in. Moreover, keyboards are pretty much all the same these daysor at least they all look
the same. Over the last decade, the key layout has become almost completely standardized, the one true keyboard design that might have been
ordained by God. With a desktop machine, you get 101 keys on a surfboard-size panel that monopolizes your desktop or overflows your lap.
With a notebook computer, youre stuck with whatever the computermaker thought best for you and the market value of its executive stock
options.
The default keyboard that comes with your computer is more variable
than you might think, however. Underneath all those keys you might
find one or another exotic technology that would seem best left to the
realm of engineers. But the technology used by your keyboard determines not only how it works but also how long it works and how much
726
you will enjoy working with it. It may even influence whether keyboarding is a time of
pleasure or pain. The underlying differences are enough to make you consider casting
aside the (usually) cheap, default keyboard the maker of your desktop computer packed in
the box and getting something more suitable to your fingers and your work. When you
consider a notebook computer, a difference in keyboards might be enough to make you
favor one machine over another, particularly when the rest of the two systems are well
matched.
Technologies
The keyboard concepta letter for every pushbuttonis almost ancient, dating back to
the days of the first typewriter. The design seems fixed in stone. Actually, its the product
of something even more immovablehuman inertia. The basic layout and function of the
keyboard has changed little since the last half of the 19th century, and the mold for the
computer refinement was first cast in 1987. No matter what computer you buy today,
youre almost certain to get a keyboard that follows the now industry-standard design.
Thats good because you can confront any computer and start working in seconds. But
thats also bad when those seconds stretch to hours of uninterrupted typing.
All keyboards have the same function: detecting the keys pressed down by your fingers
and relaying that information to your computer. Even though two keyboards may look
identical, they may differ considerably in the manner in which they detect the motion of
your fingers. The technology used for this processhow the keyboard works electricallycan affect the sturdiness and longevity of the keyboard. Although all operate in
effect as switches by altering the flow of electricity in some way, the way those changes are
detected has evolved into an elaborate mechanism. The way that technology is put into
action affects how pleasant your typing experience is.
Nearly every technology for detecting the change in flow of electricity has been adapted
to keyboards at one time or another. The engineers goal has been to find a sensing mechanism that combines accuracydetecting only the desired keystroke and ignoring errant
electrical signalswith long life (you dont want a keyboard that works for six words) and
with the right feel, the personal touch. In past years, keyboard designers found promise
in complex and exotic technologies such as Hall-effect switches, special semiconductors
that react to magnetic field changes. The lure was the wonder of magnetismnothing
needs to touch to make the detection. A lack of contact promised a freedom from wear, a
keyboard with endless life.
In the long run, however, the quest for the immortal keyboard proved misguided. Keyboards rated for tens of millions of keypresses met premature ends with a splash from a
cup of coffee. In the end, manufacturers opted for the simplest and least expensive, hardcontact technology, as close to a plain switch as you can get without a wall plate. The
CHAPTER 20 KEYBOARDS
Technologies
chief alternative, capacitive technology, led the pack at the start. It was more reliable,
longer lived, more complicatedand for a long while, more popular. It was also more
expensive. In todays dollars, the keyboard of the first computer cost more than an entire
current computer system. Something had to give.
Contact
The direct approach in keyboards involves using switches to alter the flow of electricity.
The switches in the keyboard do exactly what all switches are supposed to doopen and
close an electrical circuit to stop or start the flow of electricity. Using switches requires
simpler (although not trivial) circuitry to detect each keystroke, although most switchbased computer keyboards still incorporate a microprocessor to assign scan codes and serialize the data for transmission to the system unit.
Design simplicity and corresponding low cost have made switch-based keyboards todays
top choice for computers. These keyboards either use novel technology to solve the major
problem of switchesa short lifeor just ignore it. Cost has become the dominant factor
in the design and manufacture of keyboards. In the tradeoff between price and life, the
switch-based design is the winner.
Three switch-based keyboard designs have been used in computers: mechanical switches,
rubber domes, and membrane switches.
Mechanical Switches
Mechanical switches use the traditional switch mechanism, precious metal contacts forced
together. In the discrete switch design, the switch under each keyboard station is an independent unit that can be individually replaced. Alternatively, the entire keyboard can be
fabricated as one assembly. Although the former might lend itself to easier repair, the
minimum labor charge for computer repair often is higher than the cost of a replacement
keyboard.
The contact in a mechanical switch keyboard can do double-duty, chaperoning the electrical flow and positioning the keycaps. Keyboard contacts can operate as springs to push the
keycaps back up after they have been pressed. Although this design is compelling because
it minimizes the parts needed to make a keyboard, it is not suited to computer-quality
keyboards. The return force is difficult to control, and the contact material is apt to suffer
from fatigue and break. Consequently, most mechanical switch keyboards incorporate
springs to push the keycaps back into place as well as other parts to give the keyboard the
right feel and sound.
Although several manufacturers have built keyboards with mechanical switches, the design
has fallen from favor. Probably less than two percent of all keyboards use discrete
switches.
727
728
Rubber Dome
Rubber dome keyboards combine the contact and positioning mechanisms into a single
piece. A puckered sheet of elastomera stretchy, rubber-like syntheticis molded to put
a dimple or dome under each keycap, the dome bulging upward. Pressing on the key
pushes the dome down. Inside the dome is a tab of carbon or other conductive material
that serves as one of the keyboard contacts. When the dome goes down, the tab presses
against another contact and completes the circuit. Release the key, and the elastomer
dome pops back to its original position, pushing the keycap back with it.
The rubber dome design initially won favor on notebook computers, where its resistance
to environmental stress (a euphemism for spilled coffee) was a major strength. The
design also is inexpensive to manufacture. The switches for the entire keyboard can be
molded as one piece about as easily as making a waffle. The design was readily adapted to
desktop keyboards and provides the foundation for many inexpensive, lightweight models.
Properly designed, a rubber dome keyboard has an excellent feelthe give of the individual domes can be tailored to enable you to sense exactly when the switch makes contact. A
poor design, however, makes each keypress feel rubbery and uncertain. Moreover, some
elastomers have a tendency to become stiff with age, and as a result, some keys can
become recalcitrant.
Membrane Keyboards
Membrane keyboards are similar to rubber domes except they use thin plastic sheetsthe
membraneprinted with conductive traces rather than elastomer sheets. Most designs use
a three-sheet sandwichtop and bottom sheets with printed contacts and a central insulating sheet with holes at each key position to hold the contacts apart. The top sheet is
dimpled, and pressing down on the dimple pushes the top contact down to meet the bottom contact. The dimple snaps from one position to another, giving distinct tactile feedback that indicates when youve pressed a key.
The membrane design often is used for keypads to control calculators and printers
because of its low cost and trouble-free life. The materials making contact can be sealed
inside the plastic, impervious to harsh environments.
By itself, the membrane design makes a poor computer keyboard because its contacts
require only slight travel to actuate. However, an auxiliary key mechanism can tailor the
feel (and key travel) of a membrane keyboard and make typing on it indistinguishable
from working with a keyboard based on another technology.
Non-Contact
The biggest problem with all contact keyboards is the contacts themselves. Although
switches work well for room lights, they are fickle when it comes to the minuscule
Touch
voltages and currents used by computer systems. At a microscopic (and microtemporal)

level, making a contact is an entire series of eventstiny currents start and stop flowing
until finally good contact is made and the juice really gets going. The brief little pulses
that precede making final contact arent apparent when you throw a light switch, but a
computer can detect each little pulse as a separate keystroke. Keyboards have special circuits called debouncers that make your keypresses unambiguous. However, as contacts age,
they oxidize. Oxidation puts a less-conductive layer of oxide on the contacts and makes
your keypresses even less reliableto the point they cannot be unambiguously detected.
Early in the history of the computer, manufacturers scorned contact keyboards because
they knew traditional contact materials were fated to short lives, not the tens of millions
of keystrokes they expect from keyboards. Modern keyboards use ceramics and exotic
metals that, resistant to oxidation, can achieve the required life.
The most popular of the non-contact designs is the capacitive keyboard. Capacitance is
essentially a stored charge of static electricity. Capacitors store electricity as opposite static
charges in one or more pairs of conductive plates separated by a nonconductive material.
The opposite charges create an attractive field between one another, and the insulating
gap prevents the charges from coming together and canceling out one another. The closer
the two charged plates are, the stronger the field and the more energy that can be stored.
Moving the plates in relation to one another changes their capacity for storing a charge,
which in turn can generate a flow of electricity to fill up the increased capacity or drain off
the excess charge as the capacity decreases. These minute electrical flows are detected by
the circuitry of a capacitive keyboard. The small, somewhat gradual changes of capacity
are amplified and altered so that they resemble the quick flick of a switch.
The first computer keyboards and some of the more robust designs of the past all used
capacitive technology. The best of these designs could work reliably for 100 million keystrokes compared to 10 to 15 million for most contact designs. Although that extended
life may be commendable, most people never typed even 10 million keystrokes during the
lifetime of their computers. They usually move on to another computer well before the
keyboard wears out. The added cost of the capacitive design consequently yields no additional useful life, and the design has fallen from favor. The last capacitive keyboard from a
major manufacturer rolled out of the Key Tronic Corporation factory in early 1999.
Touch
Today the principal dividing line between keyboards is not technology but touchwhat
typing actually feels like. A keyboard must be responsive to the touch of your fingers
when you press down, the keys actually have to go down. More than that, however, you
must feel like you are typing. You need tactile feedback, sensing through your fingers
when you have activated a key.
729
730
The most primitive form of tactile feedback is the hard stopthe key bottoms out and
stops moving at the point of actuation. No matter how much harder you press, the key is
unyielding, and that is the problem. To assure yourself that you are actuating the key, you
end up pressing harder than necessary. The extra force tires you out more quickly.
One alternative is to make the key actuate before the end of key travel. Because the key is
still moving when you realize that it registered your keystroke, you can release your finger
pressure before the key bottoms out. You dont have to expend as much effort, and your
fingers dont get as tired.
The linear travel or linear touch keyboard requires that you simply press harder to push a
key down. In other words, the relationship between the displacement of the key and the
pressure you must apply is linear throughout the travel of the key. The chief shortcoming
of the linear touch keyboard is that your fingers have no sure way of knowing when they
have pressed down far enough. Audible feedback, a click indicating that the key has been
actuated, can help, as does the appearance onscreen of the character you typed. Both slow
you down, however, because you are calling more of your mind into play to register a simple keystroke. If your fingers could sense the actuation of the keys themselves, your fingers could know when to stop reflexively.
Better keyboards provide this kind of tactile feedback by requiring you to increase pressure on the keyboard keys until they actuate and then dramatically lower the force you
need to press down farther until you reach the limit of travel. Your fingers detect the
change in effort as an over-center feel. Keyboards that provide this positive over-center
feel are generally considered to be the best for quick touch-typing.
A spring mechanism, carefully tailored to abruptly yield upon actuation of each key, is the
classic means of achieving a tactile feel and can be adapted to provide an audible click
with every keypress. The spring mechanism also returns the key to the top of its travel at
the end of each keystroke. The very first computer keyboards were elaborate constructions that used a separate spring assembly for each key. Modern keyboards use a single
overall spring assembly or, more likely, an elastic rubber dome that pops between positions. Dome-based keyboards give satisfying tactile feedback, but individual keys may sporadically require increased force for a stroke or two, thus subverting their smooth
operation. Nevertheless, the low cost and good reliability make dome technology popular
among keyboard-makers.
Soft-touch keyboards use a compressible foam to work as the spring mechanism as well as
to cushion the end of each keystroke. Soft-touch keyboards give a more linear feel but are
preferred by some people for exactly the same reason others dislike themtheir lack of
snap and quiet operation.
Another influence on the feel of a keyboard is the force required to actuate a key. Some
keyboards require you to press harder than others. In general, however, most keyboards
Legends
require between 1.9 and 2.4 ounces of pressure to actuate the key. Stiff keyboards can
require as much as three ounces.
On March 22, 1999, Key Tronic Corporation introduced the first varied key-feel keyboard, one that required differing amounts of pressure to activate its keys. Called
ErgoForce by its developer, the varied key pressures are tailored to the fingers expected to
activate them. For example, the A key, which is usually operated by the left little finger,
requires only 1.25 ounces (35 grams) of force to activate, whereas the spacebar requires
2.3 ounces (80 grams). Depending on the location of each key, it requires one of five levels
of effort to activate35, 45, 55, 65, or 80 grams. According to the manufacturer, the tailored effort makes typing easier and improves the level of typing comfort for people using
a keyboard for long periods.
Keyboards also differ in how far you must press down on a key to actuate it. Full-travel
keyboards require your fingers to move down between 0.14 and 0.18 of an inch to actuate
a key. Studies show that the full-travel design helps typists achieve high speeds and lower
error rates. In laptop and notebook computers, where every fraction of an inch counts,
however, keyboards sometimes are designed with less than full travel. A short-travel keyboard actuates with less than about 0.10 inch of key travel. Whether you can live withor
even prefera short-travel keyboard is a personal issue.
Legends
Each key on a keyboard bears a legend identifying the letter, symbol, or function it activates. These legends are usually applied using one of three technologies:
Double-shot keycaps have the legend molded in. Making each keycap is a twostep process. Black plastic is molded into the inner shape of the keycap that attaches
to the switch mechanism. The legend protrudes from the top of this inner cap.
Then light plastic is molded over and around the inner cap with the legend showing
through the top.
Using the double-shot technique, each keycap for each key is a separate and different part that must be individually installed in the correct position on the keyboard as
the final step in the production process. Assembly is complex, and stock-keeping of
parts and completed keyboards can be costly. Although once popular, this technology has fallen from favor.
Pad-printed keycaps have their legends stamped into place. A machine inks on
the legends on all the keys on a fully manufactured keyboard at one swoop. Then a
transparent protective layer is applied over the ink and is baked to a permanent finish. Notebook computers and keyboards with light legends over dark caps often have
their keys pad-printed. In the case of notebooks, the assembled keyboard gets baked
before the keyboard is installed on the computer.
731
732
Laser-marked keycaps have their legends burned into place with a laser. The
process is akin to a laser printer that operates on keycaps rather than plastic. The
complete keyboard is put into a big machine with a laser and mirrors, and the laser
burns the legends into all the keycaps in one operation. The advantage of the
process is that the legends only need to be applied at the last moment, right before
the keyboard is shipped out with a new computer, and the language and layout of
the keyboard need not be set until then. The process is fast, the legends are long
lived, and stock-keeping is simplified. Most new keyboards are now laser-marked.
Layouts
On a full-size keyboard, the spacing between the keycaps for individual character keys is
0.75 inch (19 millimeters), center to center. The keycaps themselves are about 0.5 inch
(12.5 millimeters) across at the top, dished to help you place your fingers. The shape of
this curve is somewhat arbitrary. Most American designs put a concave cylindrical curve
(its curved only around the longitudinal axis) at the top of the keys; some European
designs use a concave spherical curve.
QWERTY
The one unvarying aspect of keyboards also seems the most oddthe nonalphabetical
arrangement of the alphabet keys. Anyone new to typing will be amazed and perplexed at the
seemingly nonsensical arrangement of letters on the keys of the typical computer keyboard.
Even the name given to this esoteric layout has the ring of some kind of black magic or odd
cabalaQWERTY. Simply a list of the first six characters of the top row of the nominal
arrangement, the absurdity harks back to the keyboard of the first practical typewriter.
There is no doubt that the standard arrangement is not the only possible ordering of the
alphabetin fact, there are 26! (or 26 factorial, exactly 403,291,461,126,605,635,584,000,000)
different possible arrangements of letters alone, not to mention the further complications of
using rows of different lengths and nonalphabetic keys. QWERTY is not the only possible
layout, and its probably not the best. Nor is it the worst. But it is the standard that millions
of people have spent years mastering.
A legend surrounds the QWERTY key arrangement. The typewriter was invented in
1867 by Christopher Sholes, and his very first keyboard had its letter keys arranged alphabetically. Within a year of his invention, however, Sholes discovered what he viewed as a
superior arrangement, QWERTY.
According to the common myth, Sholes created QWERTY because typists pounded on
keys faster than the simple mechanisms of the first typewriters could handle their strokes.
The keys jammed. The odd QWERTY arrangement slowed down the typists and prevented the jams.
Layouts
Sholes left no record of how he came upon the QWERTY arrangement, but it certainly
was not to slow down speedy typists. High typing rates imply modern-day touch typing,
ten fingers flying across the keyboard. This style of typing did not arise until about ten
years after Sholes had settled on the QWERTY arrangement. Typewriter development
was indeed slowthe Shift key wasnt added to the basic design to permit lowercase characters until 1878!
Other hypotheses about the QWERTY placement also lead to dead-ends. For example,
breaking a strict alphabetic order to separate the keys and prevent the type bars (the levers
that swing up to strike letters on paper) from jamming doesnt make sense, because the
arrangement of the type bars has no direct relationship to the arrangement of keys.
Dvorak-Dealey
The most familiar challenger to QWERTY, one that crawls in a distant second in popularity and use, is the Dvorak-Dealey letter arrangement, named for its developers, August
Dvorak and William L. Dealey. The name is often shortened to Dvorak. Figure 20.1
shows the Dvorak layout applied to a typical keyboard.
FIGURE 20.1
Dvorak-Dealey key
layout.
The Dvorak-Dealey design incorporates several ideas that should lead to faster typing. A
basic goal is to foster the alteration of hands in typing. After you strike one letter with a
key under a finger of your left hand, the next key youll want to press likely is under a
right-hand finger. This hand alteration is a faster typing strategy. To make hand alteration
more likely, the Dvorak-Dealey arrangement places all vowels in the home row under the
left hands fingertips and the consonants used most often in the right hands home row.
Note that the Dvorak-Dealey arrangement was developed for speed and does nothing to
make the keyboard more alphabetic or easier to learn to use.
The first publication of the Dvorak-Dealey keyboard was in the 1936 book Typewriting
Behavior, authored by the developers of the new letter arrangement. To back up the philosophic and theoretical advantages attributed to the Dvorak-Dealey arrangement, tests were
conducted in the 1930s on mechanical typewriters, amounting to typing races between the
QWERTY and Dvorak-Dealey key arrangements. Dvorak and Dealey ran the tests, and
not surprisinglythey came out the winner by factors as large as 30 percent.
Dvorak believed in both his keyboard and his test results and wrote papers promoting his
ideas. Alas, the more he wrote, the greater his claims became. Articles such as There Is a
733
734
Better Typewriter Keyboard in the December 1943 issue of National Business Education
Quarterly has been called by some experts full of factual errors. Tests run by the United
States Navy and the General Accounting Office reported much more modest results for
Dvorak.
Notwithstanding the exaggerated claims, the Dvorak layout does offer some potential
advantages in typing speed, at least after you become skilled in its use. The penalty for its
increased typing throughput is increased difficulty in typing when confronted with a
QWERTY keyboard.
The design of the computer makes converting to Dvorak relatively easy. Whereas typewriters have to be redesigned for the new key arrangement, you can just plug a new
keyboard into your computer. Commercial Dvorak keyboards often are available by
special order.
In fact, if you dont mind your keytop legend bearing no likeness to the characters that
actually appear on your screen (and in your files), you can simply reprogram your computer to think that it has a Dvorak keyboard by intercepting the signals sent by the keyboard to your computer and converting them on the fly.
Windows makes specific provisions for the Dvorak key arrangement. The standard keyboard driver allows you to select the Dvorak option. From Windows Control Panel, select
the Keyboard icon. Next, select the Language tab, on which youll find a Properties button. Clicking this button will reveal a screen like that shown in Figure 20.2, one that
allows you to choose the keyboard language. The Dvorak layout is available as an option
in the United States layout, as shown in the figure.
FIGURE 20.2
Selecting the Dvorak
layout in Windows.
Layouts
Control and Function Keys

The classic key layouts fail when it comes to computer keyboards. The array of keys they
offer simply doesnt match the needs of a computer control system. Certainly the alphabet
hasnt changedat least over the 125-year history of the QWERTY arrangement. But
computers have needs not addressed by the letters of the alphabet or basic numerals. They
require navigation keys and buttons to control system functions. Even thats not enough
to provide an easy-to-use control system for applications and operating systems. In addition (as well as subtraction and other math operations), a dedicated numeric keypad can
help immensely when your work involves entering a lot of numbers.
Each additional key adds another option in laying out the keyboard. Another word for it is
potential chaos. Although computer manufacturers have exploited some of the opportunities to craft their own, peculiar keyboardsparticularly in notebook computersthe basic
design has gravitated toward a single standard.
A common design for all keyboards seems an obvious idea, but it didnt arise spontaneously. After a short, wrenching evolution between 1981 and 1987 involving three major
design changes, the layout has remained subject to tinkering. The last significant change
came when Microsoft added three keys tied specifically to functions in the Windows operating system. Although most keyboard-makers stick to this arrangement, they sometimes
alter the size and shapes of some keysjust enough to frustrate you during your first
attempts at using an unfamiliar computer.
As with QWERTY, the arrangement of auxiliary keys is quite arbitrary, to the point of
defying explanation. As personal computers were emerging as a technology, the Video
Data Terminal with a detached keyboard was still a relatively new concept, as were function keys. The first personal computer keyboard appeared to be designed by programmers
rather than typistsit had a small Enter/Return key but placed the function keys handily
left of the alphabetic keys, just a pinkie-touch away. Experienced typistsmany of the
early adopters of personal computers, and the most vociferous of them were writers
complained, and the engineers complied, enlarging the Enter key to make the 84-key AT
keyboard (named after the 1984 IBM Personal Computer AT, which originated it).
The next wave of new users, more interested in mastering software and entering numbers,
complained that the left-hand function keys didnt correspond to the help legends that
sometimes appeared in a row at the bottom of the screen. As a result, in its next generation of personal computers, the Personal Systems/2 of 1987, IBM created a third design.
It was one of those attempts to satisfy everyone that leaves everyone dissatisfied. To add
extra features, IBMs engineers added keys, making a total of 101. Termed the Advanced
Keyboard by IBM, it is also commonly called the Enhanced Keyboard. Figure 20.3 illustrates
the United States layout of the Advanced Keyboard.
735
736
FIGURE 20.3
The IBM Advanced
Keyboard layout.
The key additions were several. A new, dedicated cursor control pad was provided separate from the combined numeric and cursor pad, and several other control keys were
duplicated in another small pad. Two new function keys (F11 and F12) were added, and
the whole dozen were moved to a top row, above and slightly separated from the alphanumeric area. Duplicate Ctrl and Alt keys were provided at either side of the spacebar, and
Caps Lock was moved to the former location of the Ctrl key.
A few functions of Windows dont fall readily under the caps of 101-key keyboards. To
better match the operation of Windows, many keyboard-makers now add three additional
keystwo Windows keys and a pop-up menu keyin the otherwise vacant area around
the spacebar, which has also shrunk to provide more room. The two Windows keys, identified by the zooming window logo on their caps, serve as attention keys to pop you into
the Windows Task Manager. One is located on the left of the space bar between the Ctrl
and Alt keys. The other fits on the right side, just right of the right Alt key. The third key
serves to select the item at which the mouse cursor points. It is located to the right of the
right Windows key.
Most makers of Windows keyboards also modify other keys for easier typing. In particular, they enlarge the Enter key to the size used by the lamented 84-key design. To provide
the extra key area, the backslash shrinks and moves upward, sandwiched between a shortened Backspace key and the plus/equals key. Figure 20.4 illustrates the most common
form of the 104-key Windows keyboard.
FIGURE 20.4
Layout of the 104-key
Windows keyboard.
To distinguish their products and make computers easier to use, some keyboard-makers
have added application pushbuttons to their keyboards that are keyed to particular software
features. In most cases, these added controls take the form of round, rubberized pushbuttons in a line above the function keys. Popular Microsoft keyboards, for example, provide
pushbuttons for commonly used Internet functions, such as triggering your browser forward and back as well as entering your email system.
Layouts
The functions of these added buttons are defined by the keyboard driver software. The
keyboard simply passes a code to your computer indicating youve pressed one of the buttons. The driver software then checks which button youve pressed and passes a predefined command along to your software. If your software differs from what Microsoft
expects, the keys might not work properly. If you have installed the keyboard properly,
you can adjust its settings through the Keyboard Properties menu in Windows Control
Panel.
Ergonomic Designs
No one likes to type, except possibly masochists and men from Mars. For most people
typing is a pain. In fact, extended typing without taking a break can cause pain and worse,
occupational ailments such as Carpal Tunnel Syndrome. The best way to avoid such problems is to avoid typingyet another good reason to quit early each day. When you cant
avoid pounding on the keyboard all day long, the next best choice is to find a keyboard
that minimizes the pain.
Human factors specialists who deal in the science of how people relate to the workplace
and machines, called ergonomics, point out that the keyboard was not designed with human
beings in mind. The original layout of keys in four (or five or six) rows was a mix of the
arbitrary and the needs of the original lever-operated typewriter mechanism. As good as
straight rows are for pounding on paper, they are an anathema for most human hands.
Normal typing forces you to splay your hands apart, bending them at the wrists and keeping them in that position while you type away. The bend stresses the ligaments in your
wrist and squeezes the nerves that lead into your hand through the carpal tunnel (essentially a hole in the bones of your wrist). Over time the stress adds up and may eventually
impair the functioning of your hands.
Computer keyboards do not suffer the same constraints as the keyboards of mechanical
typewriters. Designers can put the key switches in any position they choose. Some of the
more imaginative designs result in ergonomic keyboards. These split the alphanumeric keypad in two and angle the two halves in respect to one another. The result is an oddlooking keyboardat least to the eyes of those long used to the mechanical typewriter
designthats supposed to make typing less painful.
The theory underlying the ergonomic design is that the typing position of your hands on
the split keypad lets you keep your wrists straight. Your hands follow more of the natural
position they would assume when hanging down from your sides (hence the name of
Microsofts Natural Keyboard). Because there is less stress, prolonged typing should be less
painful on a split-pad keyboard.
With old-fashioned keyboards, and especially those of mechanical typewriters, you must
bend your wrists in another direction, vertically, for normal typing. When your arm lies
737
738
flat on your desk, you must angle your hands upward to reach the keys of a keyboard lying
atop your desk. When you use the little feet at the back of your keyboard to angle your
keyboard further, you must bend your wrists even more.
As with the horizontal bending of your wrists, these vertical bends can also cause stress
and, possibly, carpal tunnel syndrome or other occupational ailments. Consequently, many
keyboards have extended wrist rests that force you to minimize the bending of your wrists.
Unlike changes to the layout of key caps, these ergonomic designs require a brief (if any)
training period in which to accustom yourself to them. They might seem odd at first, but
you can quickly start using them effectively. The choice of a regular or ergonomic design
is simply a matter of your own preference.
Electrical Function
In the macroscopic world in which we live, switches operate positively without hesitation
or doubt. Switch on a light, and it comes on like, well, a light. In the realm of microelectronics, however, switches are not so certain in their first steps toward changing state. As a
switch makes contact, it hesitates, letting tiny currents flow and then halting them; then
letting more flow. These initial jitters are called switch bounce, and they result from a number of causes. The contact materials of most switches are far from perfect. They may
become coated with oxidation or other impurities, and they often dont mesh together
perfectly. The contacts touch, bounce, cut through the crud, and finally mate firmly
together. In the process, many tiny pulses of electricity can slip through. These pulses
arent enough to affect a light bulb but can be more than sufficient to confuse a semiconductor circuit.
For use in computer circuits, switches require special electronics to remove the jitters, to
debounce their contacts. Such debouncing circuits monitor the switch contacts and
change the hesitating initial stabs at changing state into a sharp, certain switch.
Unfortunately, each switch contact requires its own debouncer. This is not much of a concern with a single switch, but it can be a nightmare for the designer who must debounce
the 101 switches in a typical keyboard.
Instead of individual debouncing, computer keyboards use a different process for eliminating the hesitation from keyboard switches. The keyboard electronics do not detect when
each switch changes but rather check periodically to see whether a switch has changed.
When the electronics note a change in the state of a switch corresponding to the press
of a key, they generate an indication of the switch switching that is sent along to your
computer.
This process has its own shortcomings. The go-and-look method of detection adds a brief
wait to signaling each keypress. Moreover, each and every key must be individually
Electrical Function
checked. But the high speed of todays computer circuits makes this system entirely workable. The process is reduced to a routinethe keyboard electronics scan each key at rates
approaching a million times every second, looking for a change in the state of any key. In
general, a microprocessor (an 8048-series device in the most popular keyboards) scans the
keyboard for current changes every few microseconds, and the minute current flow caused
by a keystroke can be detected. Because a slight chance exists that random noisethe stuff
that must be debounced awaycould cause a current pulse similar to that generated by a
keystroke, keyboards may require that the increased current flow be detected during two
or more consecutive scans of the keyboard.
In todays computers, the keyboard can use either of two technologies to relay the effects
of your keypresses to your computer. In the traditional design, used in systems since the
first computers, the keyboard sends data to your computer in the form of scan codes. The
personal computer industry is making an effort to shift keyboard communications from
scan codes to formatted data sent through the USB port.
Keyboard Port
In the traditional (legacy) keyboard design, the keyboard connects to your computer and
sends its codes, called scan codes, through a specialized serial port. A microprocessor in the
keyboard assigns a code of pulses to your every keypress, and a second processor called a
keyboard controller converts the scan code pulses into bus-compatible data using a hexadecimal code.
When it receives a scan code, the keyboard controller chip issues an interrupt to notify your
computers microprocessor that youve pressed a key and it had better do something about it.
To service the interrupt, your computer sorts through the scan codes and figures out which
keys are pressed and in which combination. The program code for doing this is part of your
systems BIOS. The computer remembers the condition of the locking shift keys by changing
special memory locations, called status bytes, to reflect each change made in their condition.
Each press of a key generates two different scan codesone when the key is pushed down
and another when it pops back up. The two-code technique allows your computer system
unit to tell when a key is pressed and held downfor example, when you hold down the
Alt key while pressing a function key.
Rather than a single standard for scan codes, there are three. These comprise three operating modes and correspond to the three generations of keyboard development. Mode 1 is
the system used by the original 83-key keyboard of the first computers. Mode 2 is the system used by the 84-key keyboard of the computer AT. Mode 3 is the system introduced
with the 101-key keyboard design thats now ubiquitous among computers. In general,
more modern keyboards can emulate older models by shifting their operating modes
based on commands sent to the keyboard by your computer.
739
740
In any given operating mode, each key generates a unique scan code. Even if the same
legend appears on two keys, such as the duplicate number keys in the alphanumeric and
numeric/cursor keypads, the individual keys generate the same codes. The code for a
given key is determined by its position and is not affected when the Caps Lock or other
shift key is in effect. In Mode 1, all scan codes are single bytes, and the make and break
codes of individual keys are different. In Modes 2 and 3, scan codes may be one or more
bytes. In general, the make code is a single byte and the break code is two bytes, the byte
F0(hex) followed by the make code.
Although scan codes were originally fixed to key positions rather than characters, when
manufacturers alter the arrangements of their auxiliary keys they maintain the same scan
codes for the characters as in the basic design. Otherwise, every key layout would generate
different scan codes and confuse both your computer and you as you type. Foreign languages have their own scan code mappings. Each different keyboard design for a specific
language puts different characters on punctuation keys, so the scan codes elicited by these
characters will vary in languages other than English.
Modern 101-key and 104-key keyboards have two Enter keys, one on the alphanumeric
keypad, the scan code for which is listed with the rest of the keys in this pad, and a second
one in the numeric/cursor keypad. This second Enter keyas well as the duplicate number and math function keys on the numeric/cursor keypadhas its own scan code. As a
result, your computer can distinguish which Enter or number key (or whatever) you press,
just as it can distinguish between left and right Shift keys. Although for most operations
the two keys work identically, in some cases they do not. For example, hotkey sequences
often make use of one or the other of the Shift keys but not both.
All computer keyboards have dedicated function keys that have no predefined role. What
they do often changes with each application you run, although some programmers strive
for a bit of consistency (such as making F1 elicit help and F10 exit). The function keys
may be arrayed in two vertical rows of five on an 83-key or 84-key keyboard or as a single
horizontal row of 12 on 101-key and 104-key keyboards. In either arrangement, the scan
codes of these keys are the same, although the scan codes of F11 and F12 wont be available from keyboards with only ten function keys.
Each key of the dedicated cursor keypad thats interposed between the alphanumeric and
numeric/cursor keypads on 101-key and 104-key keyboards also generates a scan code distinct from the keys with duplicate functions located elsewhere.
Normally you do not have to deal with scan codes. The computer makes the translation to
numbers and letters automatically and invisibly. The converted information is used in
generating the information that appears on your monitor screen, and it is also made available to the applications you run and even the programs you write. Sometimes, however,
when you write your own programs, it is useful to detect every key change. You may, for
Electrical Function
example, want to cause something to happen when a certain key combination is pressed.
Your program need only read the keyboard input port and compare what it finds there to
a scan code table.
Your computer receives these scan codes at a special I/O port that operates much like a
serial port without explicitly following the RS-232 standard. Instead of an array of data
and handshaking signals, the keyboard uses only two. By activating or deactivating these
two signals in combinations, the system manages communications both from the keyboard
and, in all but the initial computer keyboard design, from the host system. Although the
very first computers used a one-way flow of data from the keyboard to the computer, current keyboards use a bidirectional interface with its own command protocol.
Protocol
Keyboards use different serial codes for bytes depending on operating mode. In Mode 1,
the standard protocol uses nine bits per byte. The first bit is a start bit, which must be a
logical one (1). The eight data bits follow in sequence, ordered from least significant to
most. In Modes 2 and 3, keyboards use an 11-bit protocol. Each byte begins with a start
bit, but a logical 0. Next, the eight data bits follow in sequence, least to most significant.
Next comes a parity bit. All keyboards use odd parity, so this bit is set to a logical one
or zero so that an odd number of logical ones appears in the combination of eight data
bits and one parity bit. The protocol ends with a stop bit, which is always a logical one.
Table 20.1 lists the details of this protocol.
TA B L E 2 0 . 1
Standard Keyboard Serial Byte Protocol
Bit Number
Function
Mode 1
Modes 2 and 3
Start bit
Data bit 0
Least significant bit
Least significant bit
Data bit 1
Data bit 2
Data bit 3
Data bit 4
Data bit 5
Data bit 6
Data bit 7
Most significant bit
Most significant bit
10
Parity
Not used
Depends on data
11
Stop bit
Not used
741
742
Handshaking
The signaling and handshaking system is quite elaborate. The keyboard uses the clock line
to send out signals that synchronize the bits on the data line with the receiver in your
computer. In addition, the computer uses the clock line to throttle the flow of data from
the keyboard. The clock line uses tri-state logic, which allows both ends of the connection
to alter the signal. Normally the keyboard supplies a voltage to the clock line and interrupts it to provide the synchronizing signal. Your computer can also pull down the voltageessentially shorting it out without harm to the keyboardas a means to signal its
data needs. The data line also uses tri-state logic.
The keyboard monitors the status of both the data and clock lines to determine when it
can send information to your computer. Only when both the clock and data lines are high,
not pulled low by your computer, will the keyboard send character data to your computer.
When the clock line is low, the keyboard holds any characters it wants to send in its
buffer. If the clock line is high but the data line is held low by your computer, the keyboard waits to receive a command sent from your computer. The keyboard monitors the
clock line throughout the transmission of each byte, and if your computer pulls the line
low before your keyboard sends off the parity bit, the keyboard stops its transmission and
holds the character in its buffer, waiting until the clock line going high signals that it can
retry its transmission.
To send a command byte to the keyboard, your computer first checks to see whether
any data is coming its way. To immediately stop the data flow, it pulls the clock line
low. Then, by letting the clock line go high again and pulling the data line low, it signals the keyboard that it will send data. The keyboard then counts the bits on the data
line. If it receives the correct number, it sends an acknowledgment byte, FA(hex), back
to the computer. If the bit count is wrong, it asks for a retransmission with the FE(hex)
command.
Software Commands
Your computer has a modest repertory of commands that it can issue to the keyboard to
adjust its internal operation. It can alter the operating mode of the keyboard (and hence,
which scan codes it sends out) with the command F0(hex) followed by the mode number.
It can also alter the typematic rate using the command F3(hex) or alter the status of the
locking key indicators with the command ED(hex). Table 20.2 summarizes the commands that your computer can send your modern keyboard. Very old keyboards, such as
the 83-key models accompanying the first computers, cannot receive these instructions
because of their unidirectional interface. When in Mode 1, keyboards only respond to
the Reset command FF(hex), but they then can switch to another mode to act upon other
instructions.
Electrical Function
TA B L E 2 0 . 2
Commands Sent to the Keyboard from the Host PC
Byte Value
Command Definition
ED
Change keyboard indicators (bitmapped).

Bit 0, Scroll Lock.
Bit 1, Num Lock.
Bit 2, Caps Lock.
EE
Echo.
EF
Invalid command.
F0
Set keyboard mode to byte following (1, 2, or 3).

00 = Keyboard sends out current setting.
01 = Switch to Mode 1.
F1
Invalid command.
F2
Send keyboard I.D.
F3
Set repeat delay and rate to bit-coded value.

Format 0ddbbaaa.
Delay = (dd+1) 8 250 msec.
Rate = (8 + aaa) * 2 ^bb * 4 msec.
F4
Clear buffer.
F5
Restore default settings and wait for enable.
F6
Restore default settings.
F7
Set all keys typematic.
F8
Set all keys make/break.
F9
Set all keys make.
FA
Set all keys typematic/make/break.
FB
Set key type typematic.
FC
Set key type make/break.
FD
Set key type make.
FE
Error; requests retransmission.
FF
Reset keyboard.
743
744
In addition to its scan codes, the keyboard has a modest range of commands and status
information it can relay back to your computer. For example, it can tell your computer
that it has too many characters in its in-board buffer, whether it has properly received a
command from the computer, and the status of its internal self-test. Table 20.3 lists these
commands and status signals.
TA B L E 2 0 . 3
Commands Sent from the Keyboard to the Host PC
Byte Value
Command Definition
00
Buffer overflow (Modes 2 and 3).
AA
Self-test passed.
F0
Byte that follows is break code.
FA
Acknowledge last command.
FD
Self-test failed.
FC
Self-test failed.
FE
Last command in error; resend.
FF
Buffer overflow (Mode 1).
E0
Byte that follows is scan/release code Mode 2.
To start communications, the keyboard starts by sending a series of the self-test passed
characterAA(hex)to your computer using the wrong parity to indicate that it has not
been initialized. When your computer has gone through the boot process sufficiently to
be ready to accept data, it sends an error acknowledgment to the keyboardFE(hex). The
keyboard responds by correcting the parity and sending the self-test passed character, providing of course it passed its internal self-test. At this point, the keyboard and computer
are ready for normal operation.
Host Interface
The keyboard interface inside your computer is a complete subsystem with its own dedicated microprocessor. The original design used an Intel 8042 microprocessor for handling
keyboard communications, but in modern computers, circuitry equivalent to such a
microprocessor is typically built in to the chipset. This microprocessor monitors the data
and clock lines linking it to the computer, deciphers the serial code, and repackages it in
parallel form for passing to your computer.
Electrical Function
Your computer receives data from the keyboard interface and sends data to it through a
pair of I/O ports. When the ports receive a byte and have it ready for processing by your
computer, they generate a hardware interrupt.
In modern computers, the location of keyboard ports and the interrupt generated by
the controller may vary with the hardware design and BIOS of your computer. If you
are curious, you can check the ports used from the Windows Control Panel by selecting
the Keyboard icon and then the Resources tab. Youll see a display like that shown in
Figure 20.5.
FIGURE 20.5
A Windows display
of the resources used
by the keyboard subsystem.
At this point, the keyboard controller passes a scan code sent from the keyboard to your
system, and the keyboard BIOS routine determines the character to be sent to your
operating system. The standard BIOS interruptprocessing routine generates a second
interrupt that permits a subroutine to process the scan code before it gets processed
into a character. Under Windows, these BIOS routines are replaced by keyboard drivers
that perform the same functions. By default, current consumer versions of Windows
install two keyboard driversa real-mode driver (keyboard.drv) and a protected-mode
driver (vkd.vxd). You can see the drivers installed in your system or change the drivers
using the Drivers tab of the Keyboard Properties dialog box in Windows, as shown in
Figure 20.6.
745
746
FIGURE 20.6
The Windows display
of installed keyboard
drivers.
Compatibility
The vast majority of todays keyboards use the same interfaces, protocols, and scan codes
and are essentially interchangeable at the electrical and logical level. Because key layouts
often differ, however, your fingers might find some compatibility problems, especially
when confronted with the layout of the keys on a new notebook computer.
You may encounter compatibility problems in two cases, however. Very old computers and
keyboards used a one-way, send-only keyboard design thats different from current
designs. Most of these predated the current 101-key design. In the transition period
between the acceptance of the two designs, many aftermarket keyboards were switchable
between two operating modes. The old, one-way mode was designated XT (or rarely, PC)
mode. The modern mode is distinguished as AT (or rarely, PS/2) mode. Very old computers require XT mode keyboards.
The connectors used by very old keyboards also differ from the current design, using a
large five-pin plug, as discussed later. Note that the connector is not a reliable guide to
the operating mode of the keyboard. Keyboards using the large, old, five-pin connector
may operate as either XT or AT keyboards. All keyboards using the modern, miniaturized
connector operate in the AT mode (although some old models may have switches allowing
support of XT mode).
Another compatibility issue affects keys beyond the 101 defined for the PS/2 keyboard
design. Your computer needs a software driver to recognize these additional keys.
Windows generally can identify your keyboard when it scans your hardware and set itself
up accordingly with the correct driver (or it will ask you for a disk).
Electrical Function
If you have reason to believe that its choice was incorrect or if you want to install a keyboard manually into Windows for any reason, Windows gives you that option. You access
the keyboard selection in the General tab of the Keyboard Properties dialog box. To open
the Keyboard Properties dialog box, select the Keyboard icon in Control Panel.
If the keyboard Windows displays is not the one you want, click the Change button.
Windows will respond with a list of keyboards it inherently knows how to manage, as
FIGURE 20.7
Selecting a keyboard
in Windows.
As with installing any hardware on your computer, Windows allows you to install a keyboard without built-in Windows support using a driver supplied by the keyboard-maker.
If the keyboard you want to use is not listed as one of those directly supported, click the
Have Disk button and slide the floppy disk or CD supplied by your keyboard vendor into
the appropriate drive slot when Windows instructs you to. You specify a disk location and
complete the installation process as you would with any other hardware device.
Connections
The scan-code system and serial signaling simplify the connection scheme used by computer
keyboards. Scan codes are sent from the keyboard to the computer serially so that only one
wire conductor is needed to convey the keyboard data information. A second conductor is
required to serve as a return path for the data signal; as a ground, it serves as a common
return for all other circuits in the keyboard cable. To synchronize the logic in the keyboard
with that in the computer, a separate wire is used for a keyboard clock signal. A fourth and
final wire is used to supply the keyboard with the five-volt direct current power it needs to
operate. These four conductors are all that is necessary to link keyboard to computer.
The physical embodiment of those signals is the keyboard connector. Nearly all keyboards
now use a miniature six-pin DIN connector. This design is usually called PS/2-style
because IBM introduced it with its PS/2 computers in 1987, and it has served every since.
747
748
The six pins are arrayed in a circle around a rectangular plastic tab that, along with three
guides in the shield, keys the connector against improper insertion. As with the five-pin
connector, the pin numbers follow an irregular pattern. Figure 20.8 shows the six-pin
miniature DIN keyboard connector and its signal layout.
FIGURE 20.8
The six-pin miniature DIN keyboard
connector.
5. Keyboard clock
3. Ground
1. Keyboard data
6. Reserved
4. +5VDC
2. Reserved
Only four pins are significant to keyboard use: Pin 1 is assigned keyboard data; pin 3, ground;
pin 4, five volts; pin 5, keyboard clock. Pins 2 and 6 are reserved, and the shield is attached as
a chassis ground. Table 20.4 lists the signal assignments on the PS/2-style keyboard connector.
TA B L E 2 0 . 4
Six-Pin Miniature DIN Keyboard Connector Pinout
Pin
Description
Direction
Data
In
Reserved
N/A
Ground
N/A
+5 V
Out
Clock
In
Reserved
N/A
Shield
Ground
N/A
Older keyboards use a five-pin pluga standard, as opposed to miniature, DIN connector. Because the five-pin and six-pin keyboard connectors use the same signals with only a
slight rearrangement, a simple adapter will convert one style of connector to another.
These keyboard adapters are readily available on the computer parts market. If youre so
inclined, you can make your own using the interconnection guide given in Table 20.5.
TA B L E 2 0 . 5
Keyboard Connector Adapter Wiring Scheme
PC Connector
Five-Pin DIN
Six-Pin Mini-DIN
PS/2 Connector
Signal Function
5 Clock
1 Data
Electrical Function
PC Connector
Five-Pin DIN
Six-Pin Mini-DIN
PS/2 Connector
Signal Function
2 Reserved
3 Ground
4 +5 V
NC
6 Reserved
In some computers, the keyboard cable can be disconnected from the keyboard itself. This
detachable design makes the cable easy to service (by replacing it) and a single keyboard
adaptable between five-pin and six-pin cabling standards. The keyboard-to-cable connection uses a modular (AMP) jack on the rear of the keyboard with a matching plug on the
cable, as shown in Figure 20.9.
FIGURE 20.9
The modular keyboard connector and
signal assignments.
A. Reserved
B. Keyboard data
C. Ground
F. Reserved
E. +5VDC
D. Clock
As shown in the figure, this modular connector has the following signal assignments: A,
reserved; B, keyboard data; C, ground; D, keyboard clock; E, five volts; F, reserved. When
looking at the gold contacts of the connector, the contacts are labeled in reverse alphabetical order from left to right. Table 20.6 summarizes these signal assignments.
TA B L E 2 0 . 6
Modular Keyboard Connector Pinout
Pin
Description
Direction
Reserved
N/A
Data
Out
Ground
N/A
Clock
Out
+5 V
In
Reserved
N/A
Shield
Ground
N/A
749
750
USB Port
In the long run, computer-makers hope to make the keyboard port as obsolete as steampowered personal computers. The alternative is the Universal Serial Bus (USB), which
computer-makers hope will replace all the other ports on the back of your computer.
As computers shift to the use of the Universal Serial Bus, keyboard connections will move
to this high-speed interface. USB-based keyboards were displayed as early as February,
1996, but the early introduction proved premature. Even in 2003, computer-makers prefer
the conventional keyboard interface because it is cheap and reliable, despite admonitions
from Intel and Microsoft to move to USB. As the interface matures, however, USB keyboards are destined to become more popular.
Because USB is designed and designated a universal interface, one that will link devices
other than keyboards, it is treated in a separate chapter devoted to ports, Chapter 10.
The inner workings of USB keyboards are immaterial to your computer. It just looks for
data packets received through the USB port. Any packets containing keyboard data get
routed by the USB driver to the keyboard driver of the operating system. Then the operating system pulls the data from the packets to see what youve typed. If your computer
wants to send a command to the keyboard, it notifies the keyboard driver, which generates
the code of the command and passes it along to the USB driver, which routes it out the
port and to your keyboard as another packet on the USB circuit.
The flaw in this design is that it doesnt work until the operating system and its USB drivers load. Before then, the keyboard driver has no route through which to get keystroke
codes. Thats all right, because if the operating system hasnt loaded, the keyboard driver
wont have loaded, either. With neither the USB nor keyboard driver running, your computer faces a problem if it needs to read the keyboard while booting upfor example, to
step through the setup menu.
Your computers BIOS makes up for this flaw. It patches together the old and new keyboard systems.
Wireless
As computers merge into entertainment systems, youre apt to sit farther and farther away
from the display screen. A home entertainment computer, for example, fits into your living room more like a television set, at one end of the room with you at the other. Just as a
wireless remote control has become de rigueur for your television, a wireless keyboard
gives you the same flexibility when the computer takes over.
Wireless
Wireless keyboards are two-piece systems. Each involves the keyboard itself, which functions as a transmitter, and a base station, which acts as the receiver. (Keyboard communications are actually a two-way street, but the transmitter/receiver distinction is valid as a
view of the data you generate by typing.) A single base station can serve both a wireless
keyboard and wireless mouse (or other pointing device).
The first wireless keyboards used infrared technology (similar to, but predating, IrDA),
with the interface (a little red optical sensor) built in to the host computer. Today, however, wireless systems are add-ons that plug into conventional interfaceseither the keyboard port or USB port. The connection between the base station and your computer is
completely conventional. Your computer receives exactly the same scan codes or USB
packets it would expect from a plug-in keyboard.
All the magic is in the wireless link. And wireless links might as well be magic because
they need to follow no standards other than the Federal Communications Commissions
limits on unlicensed radiation. That is, as long as they dont use too much powerdefinitely less than what is apt to cause interferenceand stick to the authorized frequency
bands, wireless keyboard-makers are free to do whatever they want with their signals.
The range of these wireless systems stretch from about 5 feet to more than 30 feet.
Unfortunately, the only guidance available as to the range of a particular keyboard is the
claim of its manufacturer.
751
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
21
Pointing Devices
Your mom probably told you that its impolite to point, but your mom
probably wasnt trying to move a mouse cursor across her computer
screen at the time. Trying to navigate through Windows with nothing
but keyboard commands is like threading a needle in boxing gloves, only
more frustrating. (It is Windows, after all.) You could point at the screen
and show Windows where to put the [expletive deleted] cursor, but you
could shout four-letter words until your face was as red as the nose of a
career politician to no avail. (Maybe thats why they call it a cursor.) You
need a method to point out to Windows where to move the cursor
namely, a pointing device.
As with any problem, technology has brought us not one, but several,
solutions. When you sit at your desk, mice and trackballs fall naturally to
hand for pointing your intentions to Windows. Go mobile, and you
might use a TouchPad or TrackPoint to indicate cursor moves. Although
each achieves the same result, each uses a different set of technologies.
Mice
The computer mouse was the first computer pointing device, developed
concurrently with the graphic environmentthe problem and solution
created together. Through its near 40 years of evolution, the mouse has
changed surprisingly little. Todays mice take the same basic form as the
originalmade to fit the handwith the same control layout of buttons
at your fingertips. Over so long a period, however, species naturally
diverge, and so have mice. Where once there was only one breed of
computer mouse, a mechanical creation wearing a single switch, now
there are several.
754
Mice can be distinguished by five chief differences: the technology they use to detect
motion (mechanical or optical), the number of buttons they have, the presence of a scroll
wheel, the manner in which they connect with their computer hosts, and the protocol or
language they use to encode the information they send to your computer. The last two of
these differences also apply to other pointing devices, so well hold back on our discussion
of them until weve looked at mouse alternatives.
As a mechanical transducer, a mouse somehow has to be able to detect its motion across
your work surface. Although engineers have a wide range of position-sensing technologies
to choose fromthey could even use the satellite-based Global Positioning System to
determine the absolute location of the mouse and then deduce changes you make in its
positiononly two techniques have actually been used in practice. These are mechanical
sensing and optical sensing.
Mechanical Mice
The first mouse was a mechanical design based on a small ball that protruded through its
bottom and rotated as the mouse was pushed along a surface. Switches inside the mouse
detected the movement and relayed the direction of the balls rotation to the host computer.
Although the ball is free to rotate in any direction, only four directions are detected, corresponding to two axes of a two-dimensional coordinate system. The movement in each of
the four directions is quantified (in hundredths of an inch) and sent to the host as a discrete signal for each discrete increment of movement.
The mechanical mouse works on just about any surface. In general, the rotating ball has a
coarse texture and is made from a rubbery compound that even gets a grip on smooth surfaces. In fact, you can even turn a mechanical mouse upside down and spin the ball with
your finger (although youll then have difficulty fingering the pushbuttons!).
On the other hand, the mechanical mouse requires that you move it across a surface of
some kind, and all too many desks do not have enough free space to give the mouse a
good run. (Of course, if all else fails, you can run a mechanical mouse across your pant leg
or skirt, but youre likely to get some odd looks.) In addition, mechanical parts can break.
A mechanical mouse tends to pick up dirt and lint, which can impede its proper operation.
You should therefore regularly clean your mechanical mouse even if you think your desktop is spotless. Cleaning is easy, if bothersome. Twist the ring on the bottom of the mouse
that retains the ball counterclockwise to release the ball. Remove both ring and ball. With
a Q-tip dampened with rubbing alcohol, scrub off the residue you find inside, particularly
that on the small rubber rollers (you may need to use your fingernail to pry loose some of
the more tenacious grime). After the alcohol evaporatesusually after a few dozen
secondsreassemble your mouse.
CHAPTER 21 POINTING DEVICES
Mice
Inside the mechanical mouse, the ball rolls against two perpendicular sensors. These sensors generate electrical pulses as they rotate, and the mouse sends the pulses to your computer. By counting the number of pulses in each direction, your computer can calculate
the movement of the mouse.
The sensors can use various technologies. For example, the wheel may be optically
encoded with an alternating pattern of black and white (or clear and opaque), and a
photo-detector senses the changes. Alternatively, the sensor may use electromagnetic
technology, sensing as a pattern of magnetic material passes a detector.
Optical Mice
The alternative to the mechanical mouse is the optical mouse. Instead of a rotating ball,
the optical mouse uses a light beam to detect movement across your desk surface. Because
no moving parts are involved, theres less inside an optical mouse to get dirty and break.
Original Design
The first optical mouse required special patterned mouse pads to operate properly and
used two pairs of LEDs and photodetectors on its bottom, one pair oriented at right
angles to the other. Its matching mouse pad was coated with an overlapped pattern of blue
and yellow grids. Each pair of LEDs and photodetectors detected motion in either direction across one axis of the grid. A felt-like covering on the bottom of the mouse made it
easy to slide across the plastic-coated mouse pad.
The big disadvantage of the optical mouse was that it required that you use its special
mouse pad and put the pad somewhere. The pad itself could get dirty, be damaged, or
simply get lost. On humid days, the plastic coating of the pad might stick to your bare
forearm and lift off in sheets. For these and other reasons, this form of optical technology
has fallen from favor in mouse design despite its simplicity and low cost.
Current Design
Todays optical mice bear little resemblance to the original optical designs. Comparing the
two is like comparing an abacus to a computer. The original optical mouse only had to be
smart enough to count lines as they rippled past a sensor. The current design uses a pattern-recognition system in a digital processor that in itself is more powerful than a complete computer of only a few years ago. The sophisticated pattern-processing capabilities
of the modern optical mouse frees the critter from the tyranny of the mouse pad. Instead
of following the preprinted pattern on the pad, todays optical mice detect the minuscule
variations in the brightness of whatever surface you run them across. They rely on the
texture thats present on most surfaces as the pattern they require to sense motion.
The modern optical system has several advantages. As with older optical mice, no moving
parts are required to sense motion, so the mice require no maintenance. They dont get
755
756
dirty, and they do not wear out. Moreover, optical mice can work on surfaces that cause
mechanical mice to falter, such as soft fabric or curved hard surfaces. On the other hand,
current optical technology has a few shortcomings. The optical sensors do not work well
on surfaces such as glass that lack visible detail or are reflective. In addition, surfaces can
be too patterned. A highly repetitive pattern such as the screens applied to photos to print
them in newspapers and magazines can confuse modern optical sensors.
A current optical mouse has three functional parts: an illumination system, a sensor thats
essentially a small video camera, and a digital signal processor for pattern recognition.
The optical mouse includes its own illumination for the surface over which it operates,
generally an inexpensive red light emitting diode (LED). The self-contained light source
is more than a means to operate the mouse when you turn the lights low. Its key to creating a pattern on the work surface that the mouse can identify. When the light source is
kept at an oblique angle, close to the surface and sharply focused, it can create a pattern of
bright areas and shadows even on seemingly smooth surfaces.
The sensor detects the light-and-shadow pattern created by the illumination system on
the work surface. Like a miniature video camera, the sensor comprises a lens and photodetector array. The lens focuses the pattern on the photodetector array, a matrix of
light-sensitive CMOS elements. In todays optical mice, this camera system is small and
accurate enough to resolve details measuring about 800 to the inch.
The digital signal processor (DSP) samples the output of the sensor about several thousand times each second. It compares each image to the previous one, looking for changes
in the light-and-shadow pattern. The processor can detect both the direction and magnitude in the shift in the pattern, which directly correspond to any movement of the mouse.
Because the sampling occurs so often, sequential images overlap and changes are apt to be
small, which allow relatively simple algorithms to suffice for detecting changes.
The choice of algorithm is the choice of the mouses designer. DSPs make brute-force
methods easy. The DSP samples the brightness at every pixel of the sensor of the previous
sample to the current sample. It then shifts the current sample one pixel in an arbitrary
direction and checks the match. It tries every direction in turn, then shifts another pixel
and tries each direction. Once it has measured the difference between all of these, it
knows the one showing the least difference is the one that reflects the most likely movement of the mouse. Designers can simplify and quicken this procedure by adding more
intelligence. For example, when the DSP finds an exact match, theres no reason for it to
keep checking. Or rather than choosing an arbitrary direction for its first test, it could
start in the same direction as the last movement.
The sampling rate determines the maximum speed at which you can move your mouse
and expect it to reliably indicate the motion. The slowest-sampling mice capture about
1500 images per second; the fastest, about 6000 images per second. According to
Mice
Microsoft, a 2500-samples-per-second rate is sufficient to handle a mouse that moves at a

physical rate of 11 inches per second. At 6000 samples per second, a mouse can track
movement made at speeds up to 37 inches per second. Even the most rapid mouse movements rarely exceed 30 inches per second.
After the processor has detected the motion of the mouse, a modern optical mouse functions much like a mechanical mouse, relaying position information to the host computer
through a serial or USB port.
Buttons
In its purest form, as envisioned by the inventor of the mouse, Douglas Engelbart, a
mouse has exactly one pushbutton. Movement of the mouse determines the position of
the onscreen cursor, but a selection is made only when that button is pressed, preventing
any menu selections that the mouse has inadvertently dragged across from being chosen.
One button is the least confusing arrangement and the minimum necessary to carry out
mouse functions. Operating the computer is reduced to nothing more than pressing the
button. Carefully tailored menu selections allow the single button to suffice in controlling
all computer functions. The Apple Macintosh uses this kind of mouse with one button.
As Microsoft developed Windows, however, its engineers opted instead for two buttons
next to each other in place of one. Two buttons allow designers more flexibilityfor
example, one can be given a Do function, and a second, an Undo function. In a drawing program, one might lower the pen analog that traces lines across the screen while
the other button lifts the pen.
Of course, having three buttons would be even better. The programmer would have still
more flexibility. Even a fourth button could be addedbut as the number of mouse buttons rises, the mouse becomes increasingly like a keyboard. It becomes a more formidable
device with a more rigorous learning curve. A profusion of mouse buttons is counterproductive.
Three buttons would appear to be the practical limit, because three positions are available
for the index, middle, and ring fingers while the thumb and pinkie grab the sides of the
mouse. Designers have, however, created mice that are little more than miniaturized keyboards with dozens of buttons. Five-button mice are in current production.
In general, having three or more buttons is superfluous. Most Windows applications are
designed to recognize only the two standard mouse buttons. Special applications may take
advantage of additional buttons. Moreover, the driver software for most multibutton mice
lets you decide the function to assign to each button. The software makes the button
functions programmable, allowing you to make each button mimic your press of a series or
combination of keystrokes on your keyboard.
757
758
The technology behind the buttons is simple, no matter the number of buttons. Each button presses down on an electrical switch or contact. The circuitry inside the mouse detects
your press as the closure of an electrical circuit and then relays a digital code to your computer to indicate the button has been pressed. When you release the button, the circuitry
responds with a different code that indicates youve let go. Each of the buttons has its own
code value for its press and release.
Wheels
In 1997, Microsoft added a new feature to the design of its first Intellimouse, a rubber
wheel nestled between its two pushbuttons. This scroll wheel allows you to scroll the screen
without clicking on scroll bars.
Actually, the mouses wheel puts only about one-sixth of the circumference of a rubber tire
under your fingers control, the rest is hidden inside the mouse. You recognize it as a
wheel only by its feel. The tire, about three-eighths inch across and made from textured
rubber to give your fingers a better grip, fits a small plastic wheel-and-axle thats molded
as a single piece.
Inside the mouse, the wheel is connected to a rotary motion sensor. In fact, the wheel may
be part of the sensor itself. As with the motion sensors of mechanical mice, the one used
by the wheel may be based on optical or magnetic technology. The optical sensor detects
the movement of a black-and-white pattern on the wheel, counting the stripes to detect
how many degrees youve turned the wheel. Magnetic sensors detect a grooved or
notched texture to the magnetic material on the wheel to count the degrees you spin the
wheel. The electronics of the mouse send out codes corresponding to the wheels motion
that are recognized by the mouses driver software.
The wheel should smoothly spin but doesnt by design. To prevent you from inadvertently
scrolling with every brush of the wheel, mechanical resistance is added to the wheel. Most
often it takes the form of detentsthe mouse notches from one position to the next as it
rotates. A spring-loaded stop lodges its circular base into a matching notch or pit in the
wheel. As you rotate the wheel, you force the stop upward, and it pushes itself back down
into the next notch. The action is purely mechanical and independent of the motion
sensing.
Resolution
Mice are sometimes rated by their resolutionthe number of counts per inch (CPI) that
they can detect. When a mouse is moved, it sends out a signal indicating each increment
of motion it makes as a single count. The number of these increments in an inch of movement equals the mouses CPI rating.
Trackballs
The higher the CPI, the finer the detail in the movement the mouse can detect. Unless
the mouse driver compensates, higher resolution has an odd effect. More resolution translates into faster movement of the mouse pointer onscreen because the screen pointer is
controlled by the number of counts received from the mouse, not the actual distance the
mouse is moved. Consequently, a higher-resolution mouse is a faster mouse, not a more
precise mouse.
The true limit on how precisely you can position your mouse cursor is your own hand. If
you want to be more accurate with your mouse, you can compensate for your human limitations by opting for less resolution from your mouse. Because you have to move your
mouse physically farther for each onscreen change, a lower mouse resolution helps you
put the cursor where you want it.
Trackballs
Just like its namesake in the Rodentia, the computer mouse needs room to roam and ends
up almost anywhere. A mouse does not stay put because moving around is in its nature
and thats how you use it.
The problem is that many folks do not have room for a roaming rodent. Their desks are
just too cluttered or they are traveling with a laptop and neglected to carry a desk along
with them into the coach-class cabin. More insidious issues also involve mice. Pushing a
plastic rodent requires clumsy, wasteful, and tiring whole-arm movements. Mice are inefficient and exhausting.
The leading mouse alternative, the trackball, eliminates these problems. Essentially a
mouse turned upside down, the trackball is much like it soundsan often big ball that,
when rotated, causes the screen pointer (mouse cursor) to track its movements. The trackball spins in place and requires no more desk space than its base, a few square inches.
Switches
As with mice, trackballs also require switches so that you can indicate when the cursor is
pointing to what you want. Most trackballs have two or three pushbuttons that duplicate
the selection functions of mouse buttons. Although some trackballs boast four buttons, the
foursome typically functions as two duplicate pairsmirror imagesso that one trackball
can serve either a right or left hand.
No standard exists for switch placement on trackballs because no consensus exists on how
you are supposed to operate a trackball. Some are designed so that you spin the ball with
your fingers. Others prefer that your thumb do the job. Which is better depends on
whom you believe and how you prefer to work.
759
760
According to one theory, your fingers are more agile than your thumb and therefore more
precise at spinning the ball, so you should get a trackball that you operate with your fingers. A competing theory holds that the thumb has more muscle control than the fingers,
so a thumb-operated trackball makes more sense. Some trackball-makers wisely avoid such
issues and make trackballs that can be used equally adeptly by the posed or opposed digit.
In truth, which design and how you use a trackball is a matter of preference rather than
hard science.
Ball Size
Another trackball design choice is the size of the ball itself and how it is retained inside
the mechanism. Various products range in dimensions from the size of a shooter marble
to those equaling cue balls. Bigger once was thought to be better, but the trackballs built
into laptop and notebook computers are making smaller sizes popular.
Because no definitive study has shown the superiority of any particular size for a trackball,
the best advice is to select one that feels best to youor the one that comes already
attached to your computer.
As with the balls in mechanical mice, trackballs naturally attract dirt. Although the trackball doesnt pick up dirt from rolling around, dust does fall upon the ball and the oils from
your fingers collect there, too. A readily removable ball can be quickly and easily cleaned.
This sort of serviceability is absent from many trackball designssomething to consider if
you plan on using your trackball for a long time.
Optical trackballs are growing in popularity. Most work like old-fashioned optical mice,
only the pattern used for recognition is part of the ball itself. If you look closely at the big
red ball that these optical trackballs use, you can see the faint lines that sensors inside the
trackball mechanism detect. Optical trackballs require less maintenance than the mechanical kind, although if they get really dirty youll need to swab off their optical sensors,
which are readily visible once you remove the ball. Use a Q-tip dampened with plain
water. Resort to window cleaner only if using water alone is not successful.
Handedness
Most mice are symmetrical. Although two- and three-button mice define different functions for their left- and right-side buttons, most enable you to flip the functions of the
buttons to suit right- or left-hand operation if what button falls under a given finger is
important to you.
Trackballs, however, are sometimes asymmetrical. In itself, that can be good. An asymmetrical trackball can better fit the hand that it is designed for, right or left. But this handedness among trackballs poses a problem when you make your purchaseyou have to
determine whether you want a right- or left-handed trackball. Which you need does not
TouchPads
necessarily correspond to the way you write. Some left-handed people prefer right-handed
trackballs (and some righties like lefties). Consequently, a right-handed trackball isnt
always the best choice for a right-handed person. Before you buy a one-handed trackball,
make sure you know which hand you will favor in using it. If you switch hands when you
tire from using one hand for spinning your trackball all day long, you may not want a
product with definite handedness at all.
Resolution
Somewhere buried in the specification of a trackball, you may find a resolution rating,
either as the number of counts per inch (CPI) of movement or as DPI (as with mice).
These figures are rarely published anymore because todays driver software makes them
irrelevant. Trackball resolution indicates how much you have to spin a ball to move the
cursor on the screenhigher resolution requires more spin to move the cursor. Software
drivers allow trackball manufacturers to give you several choices for the effective resolution of their products so that you can tailor their actions and reactions to match the way
you work. In addition, most trackball-makers offer ballistic operation in which the translation of ball movement to onscreen cursor change varies with the speed of the balls spin.
This yields fast positioning without loss of precision. The only problem is getting used to
such nonlinear control in which the speed at which you spin the trackball has as much
(and sometimes more) effect as how far you spin it.
Unlike most other peripherals, no one trackball is objectively better than the rest.
Operating any of them is an acquired talent like brain surgery, piano-playing, or haircombing. Any judgment must be subjective, and which is better suited a particular user
depends most on personal preference and familiarity.
TouchPads
In their developmental years, notebook computers used a variety of pointing devices.
Several tried integrated trackballs. The designers whims determined placementsometimes at the corner of the screen, sometimes below the keypad. In any case, the balls were
small (they had to be to fit in a portable system), and the perfectly located trackball was
rare, indeed.
Hewlett-Packard developed a pop-out mouse tethered to the right side of the system. The
system detected mouse movement through the thin, flat tether, so you could hold the
mouse in the air and its movements would nevertheless register. The downside was the
need for space on the right, either air or desktop, a requirement not easily met in crowded
aircraft.
The notebook industry needed a better alternative for a pointing device and found it in
the TouchPad (the name is a trademark of Synaptics Corporation). Almost by default the
761
762
TouchPad became the most popular pointing device for notebook systems. It wasnt that
the TouchPad necessarily was so good. All the alternatives were worse. Figure 21.1 shows
a TouchPad.
FIGURE 21.1
A TouchPad on a
notebook computer.
The TouchPad detects the location of your finger on the pad by detecting the electrical
capacitance of your finger. Your finger attracts a minute static electrical charge, which
causes a small current flow in the circuitry of the TouchPad. The electronics associated
with the TouchPad detects this minute current from two adjacent edges of the pad, so it
can precisely locate your finger in two dimensions. By following changes in the current, it
can determine the motion of your finger across the pad.
Mechanically, the foundation of the TouchPad is a printed circuit board. The top of the
board holds a pattern of conductive sensor lines etched into place. The bottom of the
board holds the electronics that make the pad work. A layer of Mylar covers the top of
the board to protect it and gives your finger a smooth surface to trace across. Current
designs are capable of resolving 1000 points per inch (40 points/mm).
A TouchPad is inherently an absolute position detector. The electronics of the pad converts the absolute location information it determines into mouse-like relative motion that
can be sent to your computer in standard mouse protocol.
The TouchPad can be completely sealed from any kind of contamination, so it is an excellent choice as a pointing device for a machine that has to operate in a hostile environment. On the other hand, useful TouchPads must be large enough to conveniently
register your finger movements. That means they steal space from the top of your notebook computer. Most notebook computermakers take advantage of this space requirement to give you a hand rest below the keyboardand to give the manufacturer more
room for the components inside the computer.
TrackPoint
Computer-makers can choose from several sizes of TouchPads for their products.
Typically the active pad surface measures 62 by 46.5 millimeters, although larger (90.6 by
72.1 millimeters) and smaller (55.9 by 32.9 millimeters or less) versions are also readily
available.
Most manufacturers locate the TouchPad just below the space bar on their keyboards.
This location allows you to use your thumbs to move the mouse so you dont have to take
your hands from the home row when typing, an important consideration if you touchtype on your computer. Notebook computers differ in the height and sensitivity of their
TouchPads. Although the differences are subtle, they affect usability. If the TouchPad
height is wrong for your typing system, youre likely to inadvertently move the mouse, or
even activate it. Youll want to try out the keyboard and the TouchPad of any portable
computer that has one before you commit to buying it.
TrackPoint
Invented by IBM, which has trademarked the name and patented the technology, the
TrackPoint system was developed by Ted Selker and Joseph D. Rutledge of IBMs
Thomas J. Watson Research Center and first used in IBM notebook computers. In principle, the pointing stick is a miniature joystick thats stuck between the G and H keys of a
conventional keyboard. The pointing stick protrudes just two millimeters above the normal typing surface. Its position enables you to maneuver it with either index finger while
the rest of your fingers remain in the home row. Because in normal touch-typing your fingers should never cross the G/H boundary, it does not interfere with normal typing. The
selection function of mouse buttons is given over to bar keys at the lower edge of the keyboard, adjacent to the spacebar. Figure 21.2 shows the placement of the TrackPoint.
FIGURE 21.2
A TrackPoint device
in its native
environment.
763
764
The nubwhich typically is removable in case you wear it smoothmechanically connects to a pair of solid-state pressure sensors mounted at right angles to one another.
When you press against one side of the TrackPoint device, it senses the pressure you
apply even though the nub itself does not move. The TrackPoint electronics and software
driver convert this pressure data into an indication of relative motion. The harder you
press, the greater the signal the pressure sensor generates and the faster it tells your computer to move the mouse pointer. The paired sensors give you two axes of control corresponding to moving the mouse along its X and Y axes.
The TrackPoint system has several advantages in addition to its favorable typing location.
It is entirely solid state and sealed so it, like a TouchPad, is environmentally rugged. It has
no moving parts to wear out, except for the pointing nub, which you can readily replace.
The disadvantage is that many people find it unnatural to use until they have acquired
experience using it.
Game Controllers
When you drive your car, you dont use a keyboard and mouseat least not yet. When
youre racing along Santa Monica Boulevard in a round of Grand Theft Auto 3, you probably dont want to type in commands to mash the accelerator through the floor and hang a
flying Louie. You want the experience and the control system to be as real as possible
without live ammunition. You want a real steering wheel, real pedals, and the real aroma
of searing rubber and SAE30 motor oil flaming out your exhaust pipe.
So far electronic devices arent very adept at generating odors (although bad software
really does stink), but some special hardware can help make you more adept in playing
gamesmechanisms that give you better control. Computer people have a clever name
for the devices that help you control games. They call them game controllers.
The history of game controllers goes back to the first personal computers and beyond. As
soon as computers graduated from the workbench to the television set, people played
games on them. For control, they used joysticks. Even the business-oriented IBM Personal
Computer made provisions for joysticks and a two-player option called game paddles using
a special interface called the game port. But such primitive control systems are as pass as
Pong, the first real video game.
Todays elaborate wheel-and-pedal combinations are possible thanks to the architectural
changes made by Windows. The game port hardware interface defined the control
afforded by joystickstwo position sensors and a single pushbutton. Thanks to an additional layer of interface abstraction (driver software) and a better connection system
(USB), designers are free to create any kind of control system they want when working
with Windows.
Interfaces
The essential part of any game controller is one or more sensors or transducers. A sensor
samples a physical measurementbe it brightness, distance, direction, or weight. A transducer converts or translates a physical property or change into an electrical signal. (The
line between them is hard to drawtransducers are sensors, but not all sensors are transducers.) A pressure sensor, for example, is a transducer that varies its output with how
hard you press on it (say, by hammering the accelerator). The magnetic or optic motion
sensors in mice are not generally considered transducers.
The difference isnt what matters. Rather, the game controller depends on being able to
sense what you dopush a pedal, spin a wheel, pull back on a throttle or stickand quantify it. The controller then takes the result of that quantification, translates it into a digital
code, packages the information in a data packet, and whisks it out to your computer
through its interface. The driver software in your computer receives the packet from the
interface and translates its content into a form compatible with your game (or other program) and passes it along through the programs Application Program Interface (API).
Your game then decides what to do with the control data.
Theres no limit to the number of controls possible using this system. You could have a
game controller as complex as the bridge of the Starship Enterprise. (Actually, thats easy
because the bridge is just a set, and none of the controls really work.) The controller
designer is free to create anything imaginable (but technically and economically feasible,
of course). Moreover, the system works both ways, so designers can put feedback mechanismsfor example, controllers that shake you up so you know when youre really applying powerto make games more realistic. The only essential element is that the game
software understand the information sent by the controller. Controller designers must
match the requirements of game APIs. The software driver and interface are simply the
glue that holds everything together.
Interfaces
To communicate its codes to your computer, any pointing device needs to be connected in
some way. Notebook computers have it easysignals can take the direct route because
they are built in. Desktop systems require some kind of port to which you can connect
your mouse or other pointing device. Most systems today have a dedicated mouse port.
But the industry-wide push to eliminate legacy ports of any kind, as well as the need for
alternative pointing devices, are making the USB interface popular (at least among computer-makers). A few mice using the legacy RS-232C serial port are also available.
Mouse Port
The most popular way to connect a mouse to a desktop computer is via the mouse port.
Physically the mouse port resembles a keyboard jack. Both use a six-pin miniature DIN
765
766
connector. The design owes its heritage to the first IBM Personal Systems/2 (PS/2) and
consequently is sometimes termed a PS/2 mouse port. The only lasting change affected by
the computer industry on this port design has been color-coding. To distinguish a mouse
port from a keyboard port, the mouse port jack and connector are colored aqua, a bluish
green. (Keyboard ports are purple.)
Electrically the mouse port is also similar to a keyboard port, using low-speed serial technology to send signals from mouse to computer using its own interrupt to avoid conflicts
with other devices. This design limits the use of the mouse port because it is not fast
enough to handle the needs of more sophisticated pointing devices.
USB Port
True to the universal in its name, the USB port accommodates mice and any other kind
of pointing device. The port, even in its early Version 1.0 form, offers enough speed for
the most elaborate pointing devices. Moreover, because pointing devices are not used during bootup before the operating system and its drivers load, theres no need for special
BIOS provisions for handling USB mouse control. The hardware installation of a USBbased pointing device requires only plugging it in, which is why this connection system is
preferred.
Serial Mice
The serial port was the first connection system used by mice, and its shortcomings provided the incentive to develop the dedicated mouse port. In general, mice make no onerous demands on the serial port. They operate at a low communication rate (1200 bits per
second) and adapt to any available port thanks to driver software. But putting a mouse on
a serial port is hardly trouble-free. Because every mouse movement generates a serial-port
interrupt, if your system has more than two serial ports, you can easily generate interrupt
conflicts with a mouse. Because serial ports 1 and 4 (that is, the ports that DOS calls
COM1 and COM4) share interrupt 4, and serial ports 2 and 3 (COM2 and COM3) share
interrupt 3, a mouse can conflict with another device connected to the other port sharing
its interrupt. If you have a serial mouse, its always best to plug your mouse into the port
that does not have to share its interrupt (for example, COM1 if your computer has three
serial ports) to avoid surprisesthe kind that can crash your computer. Better still, use a
mouse port or USB mouse.
Wireless
The most irksome trait of the mouse is the tail that inspired its name. The cord dangling
from the far end of the mouse is a desktop hazard, liable to snag whatever youre working
on. Thank goodness that inkwells are about as likely as chimera on modern desks; otherwise, you could count on your mouse making your work a uniform ink-stained blue at
Protocols and Drivers
least once a day. Mice consequently benefit from wireless technology even more than keyboards.
Making a mouse wireless is much the same as cutting the keyboard cord. The wireless link
operates between the mouse and a base station, which connects to your computer through
one of the aforementioned standard mouse interfaces. The wireless link is proprietary,
allowing the manufacturer to chose whatever technology and design best fits the purpose
and price of the product.
Infrared (IR) technology, although chosen by some manufacturers, does not mate with the
typical cluttered desktop environment. The IR sensor in the base station must be able to
see the mouseliterallyand any object in the way can create errors of omission, undetected mouse movements. If you confine your mouse to a pad, you can put the base station adjacent to the pad without risk of losing communication. But with todays optical
mouse technologies, the temptation is to move your mouse on any suitable desktop bare
spot, which may not be near the base station.
Radio technology is a better match for the lifestyle of the modern mouse, and most manufacturers have shifted to it. Because radio waves penetrate most desktop objects, you can
put the base station almost anywhere and still expect your mouse to work.
If youre thinking of going wireless, a better choice is to buy a wireless keyboard and
mouse as a package so you need only one base station for both.
Protocols and Drivers

Mice convert the motions they detect into a digital code that can be processed or analyzed
by your computer. The only loose end is what code the mouse uses. A standard mouse
code would help software writers craft their products to better take advantage of mice. A
standard mouse code would be so useful, in fact, that the industry has come up with four
distinct standards, called mouse protocols, all of which were in use at one time. These standards were developed by four of the major forces in the mouse industry, and each bore its
originators name. These include Microsoft, Mouse Systems Corporation (for a period
known as MSC Corporation), Logitech, and IBM Corporation. The first three were
designed for individual mouse products created by the respective companies. The IBM
protocol was introduced with the PS/2 series of computers, which came equipped with a
built-in jack that accepted a mouse. Other pointing devices use their own protocols, complicating matters further.
In truth, you dont need to know the details of any of these protocols. Thats the challenge
handled by the software driver used by your pointing device. And thats why you must
match the right driver software to your pointing device.
767
768
Today, the Microsoft mouse protocol is the most prevalent. Most generic mice will work
as a Microsoft mouse if you dont have a specific driver for your mouse. But the generic
Microsoft mouse protocol does not take advantage of the specific features of the products
of other manufacturers.
In normal circumstances, Windows will find and recognize your mouse or other pointing
device and install the proper driver software for it. Because Windows has a large repertory
of built-in drivers, you may not need a specific driver for your device. You can also install
a new mouse manually. Windows gives you at least two ways of accomplishing the installation: through the Add New Hardware icon in Control Panel or the Mouse icon in
Control Panel (using the General tab and clicking the Change button).
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
22
Scanners
The scanner is a science-fiction aficionados dream, a device that shifts
between dimensions. It takes three-dimensional objectsyour artwork,
pages from books and magazines, even small objects such as coinsand
reduces them to two-dimensional images by looking at them one dimension at a time. Without a trace of the alien technology or the supernatural, the scanner converts images you see in the real world and hold in
your hand into the flatland of graphics, ready to adorn your documents
or edit into arresting artwork.
The essence of any scanner is elementary. As you would expect, it scans.
That is, it uses a long array of light sensors to scan across whatever you
put into its range of sight. Each light sensor detects the differences in
the brightness of reflections off a tiny spot of the object or image you
scan.
In the typical scanner, the light sensors comprise a linear (hence, onedimensional) array of charge-coupled devices (CCDs), squeezed together
hundreds per inch in a narrow strip that stretches across the full width of
the largest image that can be scanned. The width of each scanning element determines the smallest area that can be individually judged, thus
the finest resolution the scanner can detect within a single line. The narrower each scanning element and the closer they are all packed together,
the higher the resolution and the finer the detail the scanner can capture.
Circuitry inside the scanner reads each sensing element in the line, one
by one, and in order. From that data, it creates a string of serial data representing the brightness of each point in each individual scan line. Once
the scanner has collected and arranged the data from each dot on the
line, it advances the sensing element to read the next line.
772
Types
How the view of the scanning sensor moves to that following line is the fundamental
design difference between scanners. Somehow the long line of sensing elements must shift
their attention with extreme precision over the entire surface of the image to be captured.
Nearly all scanners require a mechanical sweep of the sensors across the image, although a
few low-resolution scanners use video technology to sweep their view electronically.
To make a sweep in a mechanical scanner, engineers have devised two primary strategies.
One requires the image sensor to move across a fixed original, like you examining a statue
in a museum by walking around it. The other moves the original in front of a fixed scanner the same way you might examine an apple for intruders that have bored inside, by
holding it in your hand and turning it around in front of your eyes. With a video scanner,
nothing moves except an electron beam.
Drum Scanners
The very first scanners helped newspapers and wire services send images across the country with the ease of telegraphing messages. Someone in one newspaper office wrapped a
photo around a metal cylinder or drum, and the scanner spun the drum around while a
single light sensor checked the brightness of the photo at a single spotwhich became a
chain of observations as the drum continually spun the image under the watchful photoeye. With every spin of the drum, the light detector moved slightly down the photo until
it got to see the entire image. A matching spinning drum covered with light-sensitive
paper (like photographic film) at the other end of the connection created the image by
scanning it with a light beam.
Engineers adapted this same moving-cylinder approach into the drum scanner. Instead of a
single photo-eye, the drum scanner uses a linear array so that a single spin of the drum
covers the entire image.
Operationally, the drum scanner works like a printing press in reverse. You feed a piece of
paper that bears the image you want to capture into the scanner, and the paper wraps
around a rotating drum that spins the image past a sensor string thats fixed in place inside
the machine.
The drum design lends itself to document processing. The mechanism puts the paper
being scanned in motion, so adding a page feeder is a relatively simple addition. Because
of their orientation to scanning printed pages, drum scanners are sometimes termed page
scanners.
Today, most consumer-model drum scanners lack the drum that gave them their name.
Instead of wrapping each sheet of paper around a drum, most scan each sheet as it slides
through their mechanisms flat. The image sensor peers through a narrow slit across the
paper path, recording line after line as the paper rushes through. Expensive, precision
CHAPTER 22 SCANNERS
Types
scanners used in the graphic arts industry still cling to the classic drum design because of
its precision and simplicity.
The drum scanner mechanism imposes a stiff penaltyit allows only thin, flexible images
to be scanned. In general, a drum scanner accepts only sheets of normal paper. Books (at
least while intact) and solid objects are off limits. Moreover, most drum scanners accept
only a few sizes of paper, typically the 8.5-by-11-inch sheets of business documents.
Consequently, drum-scanning technology today is restricted to high-volume (and expensive) document processing systems used in big businesses.
Flatbed Scanners
The flatbed scanner takes the opposite tack. Instead of moving the paper to scan it, the
flatbed scanner moves its line-up of sensors down the sheet. It earns its flatbed name
from the flat glass surface, the bed upon which you must place the item to be scanned,
face down. In most flatbeds, the linear array of scanning sensors is mounted on a bar that
moves under the glass, automatically sweeping across the image. The clear glass lets the
sensors see up to the image. In addition, the glass protects the sensors and gives them a
target fixed in place at a preset distance from the scanner, which keeps things in focus.
Flatbed scanners have precision mechanisms that step the sensors or image a small increment at a time, each increment representing a single scan line. The movement of the
mechanism, which is carefully controlled by the electronics of the scanner, determines the
width of each line (and thus the resolution of the scanner in that direction).
Flatbed scanners are like copying machines in that anything you can lay flat on their glass
faces can be scannedbooks, magazines, sections of poster, even posteriors and other
parts of your anatomy if you get imaginative, bored, or drunk. Of course, the scanned
image can be no larger than the scanner bed.
In the past, the chief drawback of the flatbed scanner has been price. But manufacturers
have refined flatbed technology to the point flatbed scanners are sometimes given away
free to entice you to buy a particular computer. Although top-quality flatbed scanners for
graphic arts professionals still demand hefty prices, you can buy an entirely satisfactory
flatbed scanner for little more than $50. Those prices have made scanners that use other
technologies scarce.
Hand Scanners
Hand scanners are a variation on the flatbed design (believe it or not!) that make you the
motive force that propels the sensor over the image. You hold the T-shape hand scanner
in the palm of your hand and drag it across the image you want to scan. A string of sensors peers through a plastic window in the bottom of the hand scanner to register the
image.
773
774
Hand scanners must cope with the vagaries of the sweep of your all-too-human hand. If
you move your hand at a speed other than that at which the scanner expects, lines will be
scanned as too wide or too narrow, resulting in image distortionat best the aspect ratio
may be off, at worse the scanned image will look as wavy as the Atlantic under the influence of an errant typhoon. To avoid such disasters, the hand scanner uses a feedback
mechanism that tracks the position of the image. Most have a roller that presses down
against the image youre scanning to sense how fast you drag the scanner along. The rate
at which the roller spins gives the scanners electronics the feedback it needs about scanning speed. From this information, the software that controls the hand scanner can give
each scanned dot its proper place.
At one time, hand scanners were a low-cost alternative to flatbed designs. Because they
omitted the most expensive parts of most scannersthe precision mechanism for moving
the paper or sensorthey had an automatic edge in price. With the plummet in prices of
flatbed scanners, however, hand scanners were hard-pressed to keep up. A low-cost flatbed
is now likely to be less expensive than a hand scanner. Consequently, few hand scanners
are left on the market.
Those remaining have survived because of the chief remaining advantage of the hand
scannerportability. Hand scanners are compact and easy to carry. You could plug one
into your notebook computer and carry the complete system to the neighborhood library
to scan from books in its collection.
In addition, using a hand scanner can be quicker than using a flatbed because you can
make fast sweeps of small images instead of waiting for the lumbering mechanism of a
flatbed to cover a whole sheet. Hand scanners may also adapt to some nonflat surfaces and
three-dimensional objects. For example, most will easily cope with the pages of an open
atlas or gothic novelalthough few can do a good job on a globe or watermelon.
On the downside, the small size of the hand scanner means a single pass of the scanner
will cover an image no more than about four inches wide. Although thats enough for a
column of text (and most scanners offer a means of pasting together parallel scans of
larger drawings and photos), the narrow strips of scan make dealing with large images
inconvenient. On the other hand (and in the other direction), because a hand scanner is
not limited by a scanning mechanism, it can allow you to make absurdly long scans, typically limited only by the scanning software you use.
Note that hand-scanning is like typingits a learned skill. To use a hand scanner effectively, youll have to practice until you learn to move the scanner smoothly and at the
proper speed, which means very slowly at high resolutions.
CHAPTER 22 SCANNERS
Types
Video Scanners
A video scanner is the electronic equivalent of a photographic copy stand. That is, the scanner operates like a camera, taking in a view of the entire image in a single look. That
makes a video scanner fastcapturing an image takes a fraction of a second. Youll spend
more time setting up the image or object to be scanned than the scanner needs to scan.
Typically a video scanner uses a conventional video camera to capture an image. Most
video scanners permanently mount the camera on a stand and give you a stage on which
you put the item to be scanned. The stage may have a backlight to allow you to scan photographic slides or negatives, or it may be a large bed for sheets of paper or even threedimensional objects.
Video scanners avoid all the problems and inaccuracies imposed by mechanical scans.
They have the potential for the greatest precision. Typically, however, video scanners yield
the lowest quality. Like a video camera, video scanners require a CCD element for every
pixel they scan, and affordable two-dimensional CCD arrays have only a few hundred
thousand pixels. Because video scanners use the same CCD arrays as video cameras, they
have the same resolution as video cameras, not measured in dot per inch but in pixels
across the entire image. They are suited to snapshots and catalog illustration but not highquality scans.
Photo Scanners
As their name implies, photo scanners are special-purpose devices aimed at capturing digital
images from photographic prints. Most use an adaptation of drum-scanner technology.
They move the original rather than the sensor, sliding the photo past the image sensor,
flat.
Photo scanners can be quicker to use because you only need to slide your snapshots in like
youre feeding dollar bills into a vending machine. Because their mechanisms are inherently less complex than those of flatbed scanners, at one time dedicated photo scanners
had a price edge. With the current generation of low-cost flatbed scanners, however, that
advantage has vanished. A flatbed can do everything a photo scanner canand morebut
if all you need to scan is photos, youll have a quicker and easier time with the dedicated
device.
Slide Scanners
The slide scanner is not a special technology but rather a special implementation of flatbed
or video scanner technology. A slide scanner is a transmissive scanner rather than reflective. That is, it registers the light that is transmitted through an image rather than the
light reflected from the image. The source of illumination is on one side of the image, and
the image sensor is on the other. The image must be on a transparent medium.
775
776
A flatbed-style slide scanner is optimized for the higher-resolution needs of scanning small
photographic transparencies or negatives but relies on a modified flatbed scanner mechanism. It needs only more precision in its control because of the smaller size of the scanned
area of negatives and slides (and the correspondingly high resolution required). A video
slide scanner is subject to the same limitations as any video scannerchiefly, low resolutionbut gives you an inexpensive means of capturing limited-resolution snapshot-quality
images from slides and negatives.
Features
Beside their basic mechanism, scanners are distinguished by their features. Among these
are whether the scanner can produce color images, its scanning speed, the dynamic range
it can handle, its resolution, and whether it can recognize text characters and translate
them into character rather image data. The availability of options such as transparency
adapters, sheet feeders, and optical character recognition software also makes one model
more suited to some applications than others.
Scanning Speed
Early color scanners give monochrome models a hefty edge in performance. The earliest
color scanners were three-pass machines. That is, they required three passes to make a
complete image, one pass for each of the primary colors. These ancient scanners used
three separate light sources of different colors and took a full scan in each color. Nearly all
modern scanners use one-pass designs. They have a single light source and rely on filtering
in their photodetectors to sort out the colors. One-pass color scanners can operate just as
quickly as monochrome models, although transferring a large color image measuring
dozens of megabytes still takes longer than moving a monochrome image one-third its
size.
The speed at which the scanning CCD moves across the image area is only one factor in
the total time required to make a scan. Most scans require at least two separate passes.
First, you command the scanner to make a pre-scan, which is a relatively quick, low-resolution pass across the image that helps you establish its brightness range and also lets you
target a specific area for scanning. Then you make the actual scan at the resolution you
want.
In addition, the interface used by a scanner influences the speed of scans, as noted later in
the chapter. The high-resolution bit-images produced by the scanner represent a huge
amount of datamegabytesand a slow interface constricts this datas flow.
If the scan you want to make is large, you also have to wait for image processing, both in
the scanning software and in the host application. Very large scans can add minutes or
CHAPTER 22 SCANNERS
Features
more to the total scan time if you exceed the memory capabilities of your computer.
Although Windows can take advantage of virtual memory to let you capture images of
nearly any size, this technology uses your disk drive for extra storage space, which adds
the seeking, writing, and reading times to the total time of your scan. If you plan to regularly make large scans, youll speed things up more by adding memory to your computerin the dozens of megabytesrather than looking for a faster scanner.
Dynamic Range
At one time, the base-level distinction between scanners was like that of television sets
color or monochrome. And, as with televisions, the monochrome variety is almost extinct.
But color scanners arent all equally colorful. Some recognize more hues than others.
The compass of colors a scanner can discern is termed the scanners dynamic range. The
most common means of expressing dynamic range is bit-depth, the number of bits needed
to digitally encode the total color capacity. Most common scanners can distinguish 256 (8bit), 1024 (10-bit), or 4096 (12-bit) brightness levels in each primary color. Just to make
their scanners seem more capable, scanner manufacturers add up the total number of
colors within the scanner repertory, so youll see 24-bit, 30-bit, and 36-bit color scanners.
The actual dynamic range and the bit-depth of a scanner are not necessarily the same. A
high-quality scanner will be able to resolve the number of brightness levels its bit-depth
implies. The bit-depth actually specifies the range of the analog-to-digital converters that
convert the level detected by the scanners CCD sensors into digital signals. Taking advantage of that bit-depth requires that the scanned image be properly focused on the CCD
sensor under optimal illumination. If the focus of the scanners optics is off, pixels will
blur into one another, which lowers image contrast and the dynamic range of the scanner.
Similarly, if the illumination provided for the image during the scan is uneven, the variations will wipe out some of the available brightness levels of the dynamic range.
Consequently, two scanners with the same number of bits quoted for their dynamic range
may, in fact, have different actual dynamic ranges.
Most computers can, of course, display from 256 to 16.7 million different hues (that is, 8bit to 24-bit color). When that is more than you or your software wants to manage, the
scanners palette can easily be scaled back, either through hardware controls or through
software, to an even smaller bit-depth. With even a minimal 24-bit scanner capable of giving your software more than enough color, the extra bits of higher-cost scanners might
seem superfluous.
Those extra bits are very useful, however, when the scanner preprocesses image data
before passing it along to your computer. A scanner with 36-bit dynamic range can capture all the shades and hues of an image and let you process them down into 24-bit color
for your computer to use. You get to choose how to handle the conversion, compressing
777
778
the dynamic range or cutting off colors you dont want to use. The extra bits ensure that
your scanner can capture all the detail in the darkest shadows and brightest highlights.
When you use a transparency adapter, greater dynamic range helps you compensate for
thinner or denser originals, potentially yielding workable scans from transparencies you
can barely see through.
The trend among high-end scanners is toward a dynamic range of 48 bits, giving you a
greater range of manipulation. Note, however, that most existing software for image
manipulation cannot yet handle 48-bit images (although the current release of Adobe
Photoshop can).
Many scanners have automatic modes through which they determine the proper brightness and contrast ratios to take best advantage of the translation of the scanners dynamic
range into the 24-bit level of color (or other level of color) used by your computer. The
most common means of making this optimization is to pre-scan the image. The scanner
then checks the brightest and darkest points of the scanned area. Using these values to
establish the actual range of brightness and color in the image, the scanner can adjust its
transformation to yield the image with the greatest tonal range to your applications.
D-max
With slide scanners, another factor influences your ability to work with marginal slides
and negativesthe maximum image density, usually abbreviated as D-max, that the scanner can handle. A more dense image is darkertheres more dye or silver in the image.
Because slide scanners work by shooting light through the slide or negative, a very dense
image may prevent any light from getting through at all. The D-max indicates how dense
an image can be before the scanner can no longer distinguish the light shining through it.
Any part of an image thats more dense than the scanners D-max rating blocks up as solid
black (or, if youre scanning a negative, solid white).
Technically speaking, the density of a photographic negative or slide is the ratio of the
intensity of light shining through the image over the intensity of light that actually gets
through, expressed as a logarithm. The scientific formula for density is as follows:
Density = log (incident light/transmitted light)
As a practical matter, the best of todays slide scanners cope with a D-max of about 4.2.
That means they can detect light thats diminished by a factor of more than 10,000. A
scanner with a D-max of less than 3 will have difficulty dealing with the full range of
image brightnesses, even on properly exposed film.
CHAPTER 22 SCANNERS
Features
Resolution
Scanners differ in the resolution at which they can capture images. All scanners have a
maximum mechanical limit on their resolution. Its equal to the smallest step that their
sensor can be advanced; typically a minimal scanner will start with about 300 dots per inch
and go up from there in regular steps such as 600, 1200, then 2400 dots per inch. Specialpurpose slide scanners achieve resolutions as high as 10,000 dots per inch. Because it represents the limit of the quality the scanner hardware is able to resolve, this measurement is
often termed the hardware resolution of the scanner. Another term for the same value is
optical resolution.
Beyond the mechanical resolution of a given scanner, the control software accompanying
the scanner often pushes the claimed resolution even higher, to 4800 or even 9600 dots
per inch, even for an inexpensive scanner. To achieve these higher-resolution figures, the
control software interpolates dots. That is, the software computes additional dots in
between those that are actually scanned.
This artificial enhancement results in a higher resolution value quoted for some printers
as interpolated resolution. Although interpolating higher resolution adds no more information to a scanwhich means it cannot add to the detailit can make the scan look more
pleasing. The greater number of dots reduces the jaggedness or stair-stepping in the scan
and makes lines look smoother.
The new dots created by interpolation add to the size of the resulting scanned file, possibly making a large file cumbersome indeed. In that interpolation adds no new information, it need not be done at the time of scanning. You can store a file made at the
mechanical resolution limit of your scanner, then later increase its apparent resolution
through interpolation without wasting disk space storing imaginary dots.
As with colors and shades of gray, a scanner can easily be programmed to produce resolution lower than its maximum. Lower resolution is useful to minimize file size, to match
your output device, or simply to make the scanned image fit on a single screen for convenient viewing. Although early scanners and their control software shifted their resolution
in distinct increments75, 150, and 300 dpi, for examplemodern scanner-plus-software
combinations make resolution continuously variable within wide limits.
The actual hardware resolution of a scanner is fixed across the width of the image by the
number of elements in the CCD sensor that determines the brightness of each pixel. The
hardware resolution along the length of the scan is determined by the number of steps the
CCD sensor takes as it traverses the image area. The size of these steps is also usually
fixed. Scanning software determines lower as well as higher resolution values by interpolating from the hardware scan. Consequently, even when set for 50 dpi, a scanner will
sense at its hardware resolution level, deriving the lower-resolution figure through software from its higher capabilities.
779
780
Transparency Adapters
As with people, scanners are not blessed with the ability to see in the dark. To make a
proper scanthat is, one that doesnt resemble a solar eclipse in a coal minethe scanner
needs a light source. All scanners have their own built-in and usually calibrated light
sources. In drum and flatbed scanners, the light sources are inside the mechanism, typically one or three cold cathode tubes that glow brightly. Handheld scanners often use
light emitting diodes (LEDs) as their illumination source. In any case, in normal operation the light reflects from the material being scanned, and the CCD sensors in the scanner
measure the brightness of the reflected light.
Some source materials fail to reveal their full splendor under reflected light. The most
important of these are transparencies such as photographic slides or presentation foils.
These are designed to have light shine through them (that is, transmitted light).
To properly scan these media, the scanner must put the media between its light source
and its sensor. Slide scanners have the source for transmitted light built in. Most other
desktop scanners have an optional secondary source for transmitted light called a transparency adapter. The secondary light source tracks the CCD sensor as it scans across the
image, but from the opposite side of the original.
Most commonly the transparency adapter takes the form of a thicker cover over the glass
stage on which you lay your originals to be scanned. A few scanners have add-on arms
that scan over the top of the transparencies you lay on the stage. The latter style works
well but does not hold original transparencies as flat against the stage as do the former.
Optical Character Recognition

Scanners dont care what you point them at. They will capture anything with adequate
contrast, drawing or text. However, text captured by a scanner will be in bit-image form,
which makes it useless to word processors, which use ASCII code. You can translate text
in graphic form into ASCII codes in two waysby typing everything into your word
processor or by Optical Character Recognition (OCR). Add character-recognition software to your scanner, and you can quickly convert almost anything you can read on your
screen into word processor, database, or spreadsheet files. Once the realm of mainframe
computers and special hardware costing tens of thousands of dollars, OCR is now within
the reach of most computers and budgets.
Early OCR software used a technique called matrix matching. The computer would compare small parts of each bit-image it scanned to bit-patterns it had stored in a library to
find what character was the most similar to the bit-pattern scanned. For example, a letter
A would be recognized as a pointed tower 40 bits high with a 20-bit wide crossbar.
CHAPTER 22 SCANNERS
Features
Matrix matching suffers a severe handicapit must be tuned to the particular typeface
and type size you scan. For example, an italic letter A has a completely different pattern
signature from a roman letter A, even within the same size and type family. Consequently,
a matrix-matching OCR system must have either an enormous library of bit-patterns
(requiring a time-consuming search for each match) or the system must be limited to
matching a few typestyles and fonts. Even then, you will probably have to tell the character-recognition system what typeface you want to read so it can select the correct pattern
library. Worse, most matrix-matching systems depend on regular spacing between characters to determine the size and shape of the character matrix, so these systems work only
with monospaced printing, such as that generated by a typewriter.
Most of todays OCR systems use feature matching. Feature-matching systems dont just
look and compare; they also analyze each bit-pattern thats scanned. When it sees the letter A, it derives the essential features of the character from the pattern of bitsan upslope, a peak, and a down-slope with a horizontal bar across. In that every letter A has the
same characteristic featuresif they didnt your eyes couldnt recognize each one as an
A, eitherthe feature matching system doesnt need an elaborate library of bit-patterns
to match nearly any font and type size. In fact, feature-matching recognition software
doesnt need to know the size or font of the characters it is to recognize beforehand. Even
typeset text with variable character spacing is no problem. Feature-matching software can
thus race through a scan very quickly while making few errors.
Sheet Feeders
The typical OCR application involves transferring the information content of multiple
pages into electronic form. You must, of course, scan each page separately to derive its
information content. With long documents, the chore is time consuming and usually not
the most productive way to spend your working hours.
A sheet feeder automatically runs each sheet of a multiple-page document through a scanner. Although a sheet feeder is easiest to implement with a drum scanner, because the
scanner has to put the paper in motion anyway, some flatbed scanners have built-in or
optional sheet feeders as well.
Sheet feeders are useful primarily for OCR applications. Graphic scanning usually
involves individual setup of each page or image. Actually loading a page into the scanner is
a trivial part of the graphic scan. Adding a sheet feeder to a scanner used primarily for
graphics is consequently not cost effective.
Sheet feeders require loose sheets. They cannot riffle through the pages of a book or
other bound document. When a job requires high productivity and the information is
more valuable than the printed original, some people cut apart books and similar materials
781
782
for scanning using a sheet feeder. In any case, youll probably find a staple-puller to be a
worthy accessory to your sheet feeder.
Electrical Interfacing
At least six different interfaces designs are or have been used by scanners: Small Computer
System Interface (SCSI), General-Purpose Interface Bus (GPIB), standard serial, parallel,
USB, and proprietary. Almost all current products rely on parallel, SCSI, or USB connections.
Parallel models plug into legacy printer ports. Most have special cables or connectors that
allow you to link your printer to the same port used by the scanner. With modern
port/driver software, parallel-interfaced scanners are the easiest to get runningthey
come to life almost as soon as you plug them in (and install the drivers). The parallel
interface is also inexpensive.
The downside of the parallel connection is performance. It is the slowest of the scanner
links, and it may double the scan time of a typical page.
The SCSI interface, on the other hand, is fastthe fastest scanner connection in use. The
penalty is, of course, the need to tangle with a SCSI connection. This need not be a problem. When the SCSI-based scanner is the only device plugged into a SCSI port, getting
the scanner to work is about as easy as with a parallel port. Adding a scanner to a long
SCSI chain is as fraught with problems as linking any additional SCSI device.
The other problem with SCSI-based scanners is that they require a SCSI port. Most
SCSI-based scanners come with their own SCSI host adapters and cables. (This is one
reason the SCSI interface adds to the cost of a scanner.) Installing the adapter in an
expansion slot complicates the installation process. Worse, if you want to use a SCSI scanner with a notebook computer, youll need to purchase a PC Card SCSI adapter.
USB scanners fit in the middle. Although they require a free USB port, nearly all new
computers (including notebooks) have at least one. Most have at least two. Scanners fit
readily into USBs Plug-and-Play system, but they suffer from the same teething difficulties as other USB products. Although USB scanning is quicker than parallel, it is not as
faster as SCSI scanning.
Application Interfacing
As with other input devices, scanners have their own control and signaling systems that
must link to your software to be used effectively (or at all). Early scanners used their own
proprietary application interfaces to relay commands and data. Consequently, each scanner required its own software or drivers. Oftentimes you could only use the scanner
manufacturers own software to grab images.
CHAPTER 22 SCANNERS
Application Interfacing
Thanks to a concerted effort by the scanner industry, that situation has changed. Now you
can expect any scanner to work with just about any graphics program. Moreover, scanning
is consistent across applications. The same screens that control your scanner in
PhotoShop appear in Corel Photo-Paint.
Central to this standardization is Twain. First released in early 1992, Twain is a scanner
software interface standard developed by a consortium of scanner and software makers
called (in its final form) the Working Group for Twain. The primary companies involved
in forming the working group included Aldus Corporation, Caere Corporation, Eastman
Kodak Company, Hewlett-Packard, and Logitech.
The Twain name requires some explanation. Twain is not an acronym, so only its initial
letter needs to be capitalized. Rendering Twain in all capital letters is a typographic error.
That said, the promoters of the standard usually write it in all capital letters as if it were
an acronym.
Officially, the Twain developers have explained that the name is a reference to the purpose
of the interface. Not an acronym, it derives from making the twain (an archaic word for
two) meet, the two of the twain being applications and scanners. However, a few wags
insist that Twain stands for Technology Without An Interesting Name.
When the Twain interface was being developed, it wore a number of different names. The
most common of these were Direct-Connect and CLASP, the latter of which stands for the
Connecting Link for Applications and Source Peripherals. The developers of Twain considered these and others as the formal names of the interface. After searching through lists
of trademarks in use, however, they found so many conflicts they felt that lawsuits would
be a distinct possibility if any of the developmental names were to be used. Instead they
chose the name Twain to, in the words of one of the developers, describe this interface
which brings together two entities, applications and input devices.
Twain links programs and scanner hardware, giving software writers a standard set of
function calls by which to control the features of any scanner. One set of Twain drivers
will handle any compatible scanning device. Because the Twain connection has two ends
your scanner and your softwareboth need to be Twain compatible for you to take
advantage of the connection.
Twain defines its hardware interface as its Source. The Source is hardware or firmware in a
scanner that controls the information that flows from the scanner into Twain. The scanner-maker designs the Source to match its particular hardware and interface. Your software links to the Twain Source through a Source Manager, which is essentially a set of
program calls.
783
784
Twain takes the form of a software driver. The original driver was written in 16-bit code
and takes the name TWAIN.DLL. The working group has also created a fully 32-bit version
of the driver called TWAIN32.DLL. Both the 16- and 32-bit versions work with all versions
of Windows since 95 (including 32-bit versions such as NT, 2000, and XP).
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
23
Digital Cameras
Forget the yellow or green boxes and snub your nose at the one-hour
photo place. A digital camera doesnt need filmthats the whole point
of itand it doesnt need a film company. No need for the photofinisher
any more, either. The digital camera is a revolution in photography.
Better than Polaroid, you dont have to wait to see what develops or suffer the agonizing wait as colors fill in through the fog. Instantly you
know what went wrong (and with photography, something always goes
wrongthe challenge is finding the least-wrong image) and whether you
need to snap again. You can review a half dozen shots at a time in the
camera, choose the ones you want, and take some more without wasting
an inch of film. When youre done, you download everything into your
computer, where you can digitally edit and enhance your images until
everything is perfect. And, finally, you can put to work that photo-quality
printer that came with your computer, electronically paste snapshots into
your digital documents, or illustrate your Web site with your own photos.
The digital camera is the facilitating device for digital photography, often
called non-silver photography because of its lack of reliance on conventional photographic film. Film depends on the photosensitivity of silver
salts and the chemical reactionusing chemicals that are known carcinogenscalled development. Digital photography is much more environmentally friendly. In digital photography, photons are detected
electrically by an image sensor thats so sensitive theres no need for the
chemical amplification of the film developer. And once you tally up the
cost of the film you dont need and the photofinishing you wont use, its
wallet friendly, too.
788
The digital images created by digital cameras are perfect for posting on the Web, pasting
into newsletters and catalogs, cataloging on CD, and printing in full color on an inexpensive inkjet printer. You can edit the images yourself to eliminate the plague of red-eye in
flash photographs or an ex-spouse in family photographs.
The Big Three

There are three kinds of camera that capture digital images. In theory, you can use any of
the three to make digital images, although some work better for specific purposes.
Still Cameras
When anyone talks about digital cameras, they mean digital still cameras. That is, handheld
cameras designed like the film cameras of old. You peer through a viewfinder, push a button, and take a picture.
The most important characteristic of digital still cameras is high resolution. The images
they capture contain as much image informationdetailas the photographs taken by
classic film cameras. The high-resolution images are suitable for printing as conventional
photographs. Their quality is high enough you can even publish them in print magazines
and books. But you can also take lower-resolution imagesor reduce the resolution of
high-resolution imagesto make pictures for the Web and electronic publications.
Depending on the memory you have available to your digital still camera, you can cram
from a dozen to more than a hundred images into a single session.
To match their high resolution, digital still cameras have top-quality lenses characterized
by their sharpness. Most digital still camera lenses also have optical zooming capabilities,
ranging from 2 zooming in inexpensive cameras to a maximum (currently) of about 10.
Digital still cameras are meant to work like their film-based counterpartshandheld or on
a tripodby using the available light or a flash. But most offer additional features. Some
capture short video clips. Most have video outputs so you can plug them into your computer and capture moving images, just as you would with a video camera. You can also use
this video output as a Web cam (providing your computer has a video input).
Video Cameras
Digital video cameras are made for capturing movies digitally. They are characterized by
their fast capture rate. They have to snap at least 30 frames every second. But they do not
need to have as much resolution as digital still cameras. Most have resolution that matches
that of your computers most basic display mode, VGA. That means resolution of 640-by480 pixels.
CHAPTER 23 DIGITAL CAMERAS
Imaging Sensors
Because of the low resolution of digital video cameras, their lenses need not be as sharp as
those of digital still cameras. On the other hand, to give you more versatility in moviemaking, digital video cameras usually have extended zoom ranges, often higher than 30.
Most digital video cameras also have single-shot modes in which they operate as digital
still camerasbut with a difference. They produce video-like resolution rather than stillcamera resolution when taking snapshots. You can also used a digital video camera for a
Web camera if your computer has the proper input and software.
Web Cameras
Web cameras are the low end of the digital-photo spectrum. They have about the same
resolution capabilities as a digital video camera (sometimes less) but skimp on the lens.
Most have a fixed focus lens without zoomwhich means low-cost, at least to manufacturers. Their low resolution level is tailored to match the needs of Webcasting and video
messaging.
Most Web cameras have USB outputs inside of standard video outputs so you can plug
them into computers that dont have video inputs. With the right software, however, you
can use a Web cam to make low-grade movies or still images. In fact, the quality you get
from a Web cam may be enough for posting on the Web (what did you expect?) or even
for putting images into electronic publications.
Imaging Sensors
The heart of any digital camera is its image sensor, which actually captures the image, registering it in a form that can later be used. It detects photons and translates their energy
into a minute electrical current that computer circuits can sense, amplify, digitize, and
store.
Many people call image sensors CCDs, the name of the technology used by most digital
camera image-sensing devices (charge coupled devices). Although CCDs are used in most
cameras, some companies have adopted a similar technology thats usually termed CMOS.
Yes, its the same name as the memory used for system-configuration data in your computer, because the image sensors and the memory use the same semiconductor technologyComplimentary Metal-Oxide Semiconductor technology. Because of the use of
garden-variety circuit technology, CMOS sensors tend to be less expensive, although they
typically are not as sensitive as CCDs.
Although the term CCD usually is used in the singular, the actual image sensor of a digital
camera requires an array of CCDs. The mere presence of light is not as important as its
pattern, which makes up the actual image. A single CCD (or CMOS sensor) element registers only a single point. Consequently, image sensors are an array of individual CCD
789
790
elements. In video cameras, the CCD elements get arranged as a matrix. Camera circuitry
samples each element in turn to scan an image frame.
Size
Image sensors come in various sizes. Typically they measure one-quarter, one-third, or
one-half inch (diagonally). All else equal, the larger the CCD, the greater the number of
elements that can be packed into the array. The number of elements determines the resolution of the image signal produced by the CCD. This number is related to, but is not the
same as, the number of pixels in an image.
The size of the sensor affects other elements of the camera design. For example, the coverage of the lens usually is tailored to match the sensor size. (Thats one reason why camera-makers cannot simply put CCDs inside ordinary 35 mm camerastodays practical
sensors are substantially smaller than 35 mm film and would only register a fraction of the
image from the lens.) Because most image sensors are smaller than the film used in ordinary cameras, interchangeable lenses often act longer on digital cameras. That is, a 50
mm lens from a standard film-based SLR might give the view of a 65 mm lens when
twisted onto a digital camera. The sensor sees only the central part of the image from the
lens. With many of todays high-end digital cameras with interchangeable lenses, 35 mm
lenses act as if their focal length were 1.3 times longer than on a film camera.
Resolution
The maximum possible resolution is set by the number of pixels that the CCD can sense.
The keyword in digital photography currently is megapixelthat is, a million pixels. At
one time photographers used the million-pixel mark as distinguishing good cameras from
bad. Today the best consumer-model digital cameras have about five megapixels of resolution, which results in images comparable to 35 mm film.
The number of total pixels is directly related to the resolution of the image sensor. If you
know the number of elements in an image sensor (video cameras are sometimes described
by this figure), you can determine its highest possible resolution using these formulae:
Horizontal resolution = 4 * sqr(number of pixels/12)
Vertical resolution = 3 * sqr(number of pixels/12)
Unlike the measure of resolution of other computer peripherals, where each pixel requires
an individual sensor for each color (three sensor elements equal one pixel), digital camera
makers exaggerate their resolutions by using the total number of pixels for all colors to
come up with a resolution figure. Most digital cameras use an array of sensors that alternate green with each of the other two colors. Using interpolation algorithms, the camera
calculates three individual color values at each element position, regardless of its color
sensitivity; in effect, creating three virtual pixels (one of each color) for each image sensor
element. A megapixel camera thus has only one million sensing elements.
Lenses
Sensitivity
Image sensors vary in their sensitivity to light. The best can detect a single photon. Those
in digital cameras require dozens of photons to make a detected signal. It is similar to the
rating of film speed. In fact, to give film photographers a better feel for their digital cameras, many digital cameras allow you to adjust their sensitivities to the equivalent of an
ASA or ISO film speed rating (typically the ASA 100 or ASA 400 rating of the most popular films).
The inherent sensitivity of the image sensor is not, however, directly relevant to practical
photography. Other aspects of the design of the camera overwhelm concerns about the
native sensitivity of image sensors. The most important of these is the lens.
Lenses
The lens determines the quality of the images that are formed inside a camera. The lens
and its quality determine what kind of image your camera can collect and to a great
degree set the ultimate quality of picture you can hope to take.
You can take photographs without a lensthats the principle behind pinhole photography. But nearly all cameras have lenses for one very good reason: A lens collects a lot of
light. It captures more photons. That gives the camera more light to work with and makes
exposures quicker. The lens also affects the view, what the camera sees. It even alters the
aesthetics of the image you make.
Aperture
Literally speaking, an aperture is nothing more than a hole, and thats what it means for
cameras. The aperture is the hole in the lens through which light can get to the image
sensor. In photography, digital or otherwise, the aperture is more important than a mere
hole. Its a variable hole, one you can make larger or smaller to let more or less light reach
the film or sensor. A larger aperture lets more light in; smaller lets in less.
Varying the size of the aperture helps a camera of any kind cope with light conditions.
When light is too bright, it can overwhelm the image sensor; too dim, and the image sensor might not be able to find enough photons to make an image. To prevent these problems, most cameras use wider apertures in dim light and smaller apertures in bright light.
The mechanism for setting the aperture is termed the iris of the camera, and it corresponds to the iris of the human eye. By sliding thin plates called iris bladesin manual
mode typically by rotating a ring around the lens termed, appropriately enough, the aperture ringyou can adjust the size of the hole between the blades and thus the aperture.
The size of the aperture is measured as an f-stop. Most commonly, the f-stop is a number
in the geometric series 1.4, 2.0, 2.8, 4.0, 5.6, 8.0, 11.0, 16.0, 22, 32. The series is designed
791
792
so that the next higher stop cuts the light transmitted through the lens to half the value of
the previous stop. An f-stop setting of 8.0 allows half as much light into the camera as a
setting of 5.6. (The sequence is simpler than it looks. Each f-stop differs from its predecessor by the square root of two, the results rounded.) Table 23.1 lists the ISO standard
(nominal) f-stops, the actual (computed) f-stops, and relative light values.
TA B L E 2 3 . 1
F-Stops
Nominal Stop (ISO)
Actual Stop and Half-Stop
Light Value (Relative f/1.0)
1.00
1.00
1.19
1.41
1.41
2.00
1.68
2.83
2.00
4.00
2.38
5.66
2.83
8.00
3.36
11.31
4.00
16.00
4.76
22.63
5.66
32.00
6.73
45.25
8.00
64.00
9.51
90.51
11.31
128.00
13.45
181.02
16.00
256.00
19.03
362.04
22.63
512.00
26.91
724.08
32.00
1024.00
38.05
1448.15
45.25
2048.00
53.82
2896.31
64.00
4096.00
1.4
2.8
5.6
11
16
22
32
45
64
Lenses
Although the sequence of numbers is now an arbitrary sequence, the value of the f-stop is
scientifically defined. It is the focal length of the lens divided by the apparent aperture of
the lens (not the actual size of the hole in the iris but the size of the hole visible through
the lensthe glass in the lens can magnify the aperture or even make it appear smaller).
For example, a lens with a 50-millimeter focal length set at f-stop 4 would have a visible
aperture of 12.5 millimeters. This relationship leads to the common way of writing f-stop
settings. A setting of four is usually written as f/4. In other words, it is the focal length
divided by four.
Lenses are usually described by the widest aperture at which they can be set (for example
f/1.4 or f/2.8). Sometimes the widest aperture setting falls between the numbers of the
standard f-stop sequence (for example f/2.3), but the value represents the same concept.
Many zoom lenses are marked with two f-stops (for example f/2.84.3). The two values do
not represent the total range of stops available from the lens. Rather, they represent the
range of minimum f-stop values. The formula for determining the f-stop of a lens requires
the size of the aperture to vary with the focal length of the lens at a constant f-stop setting. A longer lens requires a wider aperture for the same f-stop setting.
Zoom lenses are able to vary their focal lengths to change the size of the image they make
in your camera. Nearly all zoom lenses automatically change the aperture as you zoom to
maintain a constant f-stop setting. The physical diameter of the lens limits its maximum
aperturethe hole in the iris cant be bigger than the lens itself. When the focal length is
set shorter, however, the largest possible aperture represents a wider (lower value) f-stop,
which is desirable because it allows more light into the lens so that you can take photographs in dimmer light. Consequently, lens-makers let you take advantage of the wider
f-stop settings at shorter focal lengths, and the minimum f-stop value varies with the focal
length setting of the lens. The highest f-stop number represents the widest setting at the
longest focal length setting of the lens. The lower value represents the widest f-stop setting possible at the most favorable focal length setting.
When using a lens, the aperture or f-stop setting can have a dramatic effect on the final
image beyond setting exposure. The f-stop determines the depth of field (more correctly,
the depth of focus) in the image.
Focal Length
Technically speaking, the focal length of a lens is the distance from its nodal point to the
plane of focus of its image of an object an infinite distance from the lens (that is, where
the image appears sharpest). Although this highly technical concept appears to have no
practical value in judging cameras or lenses, focal length has an important ramification. It
determines the field of view a lens provides or, the corollary, the size of the image. As a
practical matter, a lens with a short focal length provides a wide field of view and makes
793
794
things look smaller and farther away. A long lens provides a narrow field of view and large
images that look closer. There are other aesthetic considerations well discuss in the chapter about using a camera.
Wider, smaller, larger, and closer are all relative concepts. With lenses, such terms relate
to a normal lens. For some reason not readily apparent, the photographic world has
decided a normal lens has a field of view of 46 degrees. That is, the angle between the
camera and the left side of what it sees is 46 degrees from a line drawn to the rightmost
side of the image. On a 35 mm camera, that means a 50 mm lens is normal.
Zoom
Technically speaking, a zoom lens is one that allows you to change its focal length within a
given range. For most people, however, the results of the zooming process are more
important. Zooming lets you change the size of the image in your viewfinder and in your
files. You zoom in to make things bigger, as if you had stepped closer to whatever youre
photographing. Zooming out makes things smaller so you can fit more of your subject
into the frame. Zooming also has aesthetic effects, such as changing the depth of field and
compressing distance.
Optical Zooming
The classic zoom is optical. The lens creates the entire zoom effect by changing the path
light beams take through the lens. The optical principle is actually simple. A zoom lens
has an extra internal element that acts as a magnifier, increasing the size of the image by
apparently increasing the focal length of the original lens. The magnifying effect changes
as the magnifier moves with respect to the other elements of the lens. Therefore, by sliding the magnifier, you can continuously change the focal length of the original lens.
In practice, zoom lenses are not so simple. They are among the most complex optical
devices in common use. They have multiple lens elements to correct for image problems
and distortions added by the zoom processfor example, maintaining that constant f-stop
setting as the focal length of the lens changes.
Maintaining the quality of the image produced by a lens throughout its zoom range has
always been a challenge for optical designers. Thanks to modern computers, however,
they have become adept at creating superb zoom lenses. The designs are complex, however, and that means they can be costly to produce. Thats the chief disadvantage of the
optical zoom.
The qualitymeaning the sharpness, contrast, and color saturationof the image produced by a modern zoom lens does not change as you zoom through the entire range of
the lens. Keep that in mind. Its the chief advantage of the optical zoom.
Zoom
Digital Zooming
Optical zooming works on the image before it is actually formed on the image sensor
inside a digital camera. Digital zooming, in contrast, manipulates the image after it is registered by the sensor. In effect, digital zooming is nothing more than interpolating the
image to a different size. The process is exactly the same as you can do with image-editing
software inside your computer. Its called digital zooming only because it takes place within
the digital camera with the results showing in the cameras viewfinder.
As with any interpolation, quality can suffer. The interpolation process cannot create
image data thats not there to start with. Because the zooming process typically fills the
entire frame with an image taken from a fractional part of the frame and discards the rest,
the zoomed image inevitably is of lower quality than the full-frame image.
Zoom Range
The zooming range of a lens is measured in its own x-factor. The number represents
the longest focal length of the zoom lens divided by its shortest focal length. For example,
a 50 to 150 mm zoom lens is described as a 3 zoom (150 divided by 50). At that, the xfactor describes the power of the zoom to increase the linear dimensions of the images it
produces. With a 3 zoom, for example, the greatest magnification of an image will be
three times taller and three times wider than at its least magnification.
A larger zoom range is better than a small one and, of course, typically more costly. Be
alert that sometimes camera-makers publish the total zoom capabilities of their cameras as
a single figure thats a product of both optical (useful) and digital (doubtful) zoom capabilities. For example, a camera advertised with 6 zoom capabilities often has a 3 optical
zoom and a 2 digital zoom. For most practical purposes, you can consider such a camera
to have merely 3 capabilities.
Supplemental Lenses
Just as modern lenses are made from multiple elements, you can add additional elements
to a lens to change its fundamental characteristics. For example, put a positive (convex)
element in front of a camera lens, and you will change its focusit will focus nearer.
Additional elements that you can attach to the basic lens of your camera are called supplemental lenses. Some digital camera makers call them conversion lenses.
Four types of supplemental lenses are sometimes available. These include the following:
Telephoto converters. These lenses make distant images closer and larger without
affecting the cameras ability to focus a faraway object.
Close-up converters. These lenses allow your camera to fill its frame with the
image of a nearby object. You can photograph things closer to your camera but,
when using the converter, you can no longer focus on faraway objects.
795
796
Wide-angle converters. These lenses allow your camera to take in more panoramic
views. This type of converter will add drama to landscape photos but can be even
more useful indoors when you cant otherwise get far enough from large subjects
(for example, groups of people) to fit them entirely in the frame.
Fisheye converters. These lenses take wide-angle to the extreme, giving you a 180degree view by bending the straight lines of the image to fit the frame.
Which of these four types of converters are available to you depends on what your
camera-maker offers or whether aftermarket products fit the camera you choose. If you
want to experiment with additional telephoto reach or wide-angle panoramas, check to see
which options you can buy before you purchase a new digital camera.
Shutters
The role of the shutter in a camera is to limit the amount of light getting in. Certainly too
much light will overexpose film and overwhelm image sensors, but constraining the quantity of light isnt the only function of the shutter. After all, reducing the aperture will similarly prevent overexposure. The shutter controls exposure by its duration, the period
during which light is allowed into the camera. Images often change over timepeople
blink, horses gallop, racecars race, grass grows, and paint peels. By limiting the time during which light is gathered, the shutter can capture a small slice of the image during
which movement is minimal, even invisible. A fast shutter can simply stop motionmore
correctly, stop the blurring effects of motion on captured images.
With a digital camera, the shutter speed is the period over which the image sensor captures photons. The longer the period, the more photons the sensor will capture.
Shutter Speeds
Over the years, shutter speeds have been standardized in a sequence in which each speed
is one-half the next fastest, effectively cutting the light that passes through by one half.
Consequently, each step in shutter speed alters the amount of light reaching the film or
image sensor by the same amount as the change in one standard f/stop. Heres a list of the
standard shutter speeds:
1 second
1/2 second
1/4 second
1/8th second
1/15th second
Shutters
1/30th second
1/60th second
1/125th second
1/250th second
1/500th second
1/1000th second
1/2000th second
1/4000th second
1/8000th second
Shutter speed settings are typically displayed on conventional cameras without the 1/
indication. Consequently, the notch for a shutter speed of 1/125th second will be identified as simply 125.
Cameras differ in the speeds they make available to you. Only more expensive, professional-grade cameras offer the highest speeds, 1/2000th second and greater.
Choosing a shutter speed is not arbitrary. You (or the program of an automatic camera)
must consider several issues. The two most important are proper exposure and the minimization of the effect of camera shake.
With 35 mm cameras, the oft-stated rule for the slowest shutter speed that will yield sharp
images when the camera is handheld is one over the focal length of the lens (or shorter).
For example, with a standard 50 mm lens the slowest speed recommended for handheld
exposures is 1/60th second. Some people are able to make sharp exposures at speeds a
notch or two slower than the general rule. But when you need to make an exposure longer
than these recommendations, you should mount your camera on a tripod to prevent
shake. A tripod is not a cure-all, however. Even a tripod wont prevent the image blurring
because of subject movement.
Shutter Delay
Mechanical cameras dont take a picture the instant you push the shutter release. All sorts
of cams, gears, and levers swing into action as soon as you press, but light cannot peep
through the shutter for several milliseconds as everything comes up to speed. This slight
wait is called shutter lag. With a digital camera, theres no mechanical inertia to slow the
taking of a picture, but most digital cameras impose their own shutter delay penalties that
are substantially longer than those of mechanical cameras. With some digital cameras, you
may wait more than half a second between the instant you press the button and when the
camera actually makes an exposure. If you dont properly anticipate the action, you may be
as surprised as your images. Between the press and the exposure, smiles of surprise melt
797
798
into shock and anger, high divers disappear into the deep, UFOs accelerate into the ether,
and Bigfoot shuffles back into the wilderness. Strangely, longer delays arise most often
with more expensive digital cameras.
The cause of the lag is twofold: mechanical and electronic. Even though digital cameras
take electronic pictures, they are still dependent on mechanical processes for the optics
and focusing. The auto-focusing systems of many digital cameras, especially higher-end
cameras that do a better job of it, may require several hundred milliseconds to lock on a
subject. You can often eliminate this part of the lag by prefocusing or by focusing manually. The electronic delays are imposed in setting up the electronics to actually capture the
image. The time required often increases with the resolution of the camera because theres
more to do with more pixels to capture. You cant do anything about the electronic lag
except learn to live with it and anticipate action. Newer digital cameras are notably faster
than their forebears, but the lag can still be bothersome. Avoid surprises and check the
shutter lag of any digital camera before you buy it.
Exposure
The lens aperture and shutter can control the light striking the light sensor, but to do the
job properly, they need to be carefully controlled. A sunny day may be 10,000 times
brighter than a dark room, yet the cameras light sensor is called upon to register minute
differences in each. At one time, photographers had to guess at the correct aperture and
shutter speed settings. Light meters that measure the brightness of individual scenes gave
them a tool to objectively evaluate the proper shutter and aperture settings. Automatic
exposure systems directly link the light meter to the exposure-regulating controls of the
camera so you dont have to worry about adjusting the settings at all.
Sensing
Making an automatic exposure system work requires the camera measure the brightness of
the scene you want to photograph. So you need to measure the brightness of the scene.
However, brightness is a property not of each scene but of each part of a scene. As long as
the world isnt a uniform gray, some parts of a scene will be brighter than others. This
inevitable difference in brightness is goodwithout it we wouldnt see anything. But it
poses a difficult question for exposure metering: Where should you measure brightness?
Answering that question is tough because it has no one correct answer. Certainly there are
obvious choices: Measure everything, measure the brightest spot, average the brightest
and darkest areas, measure only the center, and so on. Although almost any strategy
results in an image, often acceptable, none will always produce the best possible image.
Finding a way of evaluating image brightness in the greatest number of situations has
Exposure
challenged camera designers since they first glued selenium cells on the fronts of their
products to make automatic exposure systems.
The advent of the digital camera alters only the mechanics of the situation. Digital cameras (and video cameras, for that matter) can, in theory, measure the brightness at every
pixel in an image. But the camera designer must choose how to combine those signals to
control the aperture and shutter.
The issue is not only which pixels to sample but what importance to assign to the signal
from each onethat is, the weighting to assign to each of potentially a few million brightness measurements. The guiding factor is as much aesthetic as engineering, however.
Through the years, a number of schemes have been tried, increasing in complexity as
technology has allowed. The primary choices are three (in order of introduction): fullscene, spot, and matrix, lumped together as metering patterns.
Full-scene is simply an average of the brightness of the entire scene, which works in a surprisingly large number of cases. But its also fooled into the wrong setting in many cases.
Backlightingwhen a person faces the camera with the sun behind himusually results in
a silhouette, leaving the subjects smile lost in the darkness.
Spot metering looks only at a central spot of the image, usually the most important part of
the image. By measuring only the face in a backlit scene, for example, it ignores the bright
sun and lets you see the expression on your subjects face. But a bright or dark spot in the
center of the image can fool the spot meter. (Most spot-based systems let you compensate
by making a measurement on the most important part of your subject, locking the setting,
and then letting you compose the image and make the exposure.
Matrix metering subdivides the image into several areas of importance, applies a weight to
each, and then integrates the values to come up with an exposure setting. The most
advanced form of matrix metering uses computer-based intelligence to deduce what youre
trying to photographnot quite a canned aesthetic sense but no more than a step away.
Camera-makers have discovered that certain patterns of brightnessfor example, bright
around the edges but dark in the centerare trademarks of common photographic situations. The example would indicate a backlit subject, which the camera can identify and
compensate for. The best digital cameras use matrix metering. Each camera-maker has
developed its own set of patterns and algorithms that it believes best cover most photographic situations.
Control
Metering only creates information. The next issue is what to do with the information that
the metering system develops. The shutter speed and aperture interact, so you have to
choose the best setting of each for your particular photographic situation. You have two
799
800
ways of dealing with the choiceslet the camera deal with them or do it yourself (in other
words, automatic or manual exposure control).
Manual
Manual exposure is the old-fashioned way. You read the meter (which may be inside the
camera and displayed in the viewfinder of a modern digital camera) and then adjust both
the shutter speed and lens aperture to match its recommendations. Old match-needle
metering requires you simply to adjust the camera so that the needle of its light meter
moves to the center of its display; newer cameras may ask you to make adjustments until a
red LED turns green, or some such nonsense.
For the most part, manual exposure systems existed because camera-makers had not yet
learned how to make them automatic. Manual exposure still gives you the utmost in control, however. It allows you to control the depth of field of your photographs, the
amount of blurring from image movement, and similar aesthetic pictorial features. If you
have the least pretense toward making artistic photographs, youll want a camera that
allows you to step backward to manual exposure control.
Automatic
Automatic exposure is not a single miracle, its several. Auto-exposure systems may operate in any of three distinct modes and may even allow you to choose which to use. These
modes include the following:
Aperture priority. This mode asks that you set the lens aperture and automatically
adjusts the shutter speed for proper exposure.
Shutter priority. This mode asks that you set the shutter speed, and the camera
automatically adjusts the aperture for proper exposure.
Programmed. This mode asks only that you point the camera in the general direction of what you want to photograph, and the camera adjusts both shutter speed and
lens aperture according to a stored program.
The programs used by programmed exposure systems vary at the discretion of the
camera manufacturer. Some favor shorter exposures to lessen the chance of blurring
with long lenses. Some favor smaller apertures to yield more depth of field to simplify the job of auto-focus systems. More expensive cameras allow you to select from
different programs to match the photographic situation you face. Some cameras
even let you tailor your own programs or select from more exotics programs stored
on removable memory cards.
In any case, most cameras set off flashing lights, beeps, or sirens to warn when you try to
photograph a scene out of the range of the automatic exposure systems capabilities to
ensure a proper exposure. (A few snotty cameras may even prevent you from taking a
Flash
photo that they think they cannot properly expose.) Usually adjusting the parameter thats
left to manual control (the aperture or shutter that has priority) will resolve these difficulties unless the scene is just too dark.
The compromise between automatic and manual exposure is to handle things automatically but give you a veto. That is, let the camera find its own way but allow you to take
over when things get too challenging.
To allow you to override the automatic exposure system and fine-tune the exposure of
your images, most cameras provide exposure-compensation controls. These are also called
backlight switches or something similar in recognition of the situation most often requiring
compensation: when the source of light is behind the actual subject, making the background substantially brighter than the foreground. The exposure-compensation control
simply tells the camera to overexpose the overall image to bring the actual subject into the
proper exposure range. This overexposure control is also useful for other special situations, such as snowy ski scenes and bright sandy beaches. Although advanced matrix
metering systems supposedly cope with this sort of lighting, a camera with exposure compensation gives you a degree of control without sacrificing general automatic operation.
Flash
No matter how sensitive a light sensor is, none can find an image in absolute darkness.
When its truly dark, there are no photons available to register on it. To allow you to take
photographs even when insufficient light is available, most digital cameras incorporate
their own photon sourcea flash. As the name implies, a flash unit makes a flash of bright
light. Although once based on contained explosions sealed inside small light bulbs, todays
flash units rely on the excitation of xenon gas to produce a brief but bright pulse of light.
The brevity of the flash allows the near-instant illumination to stop motion (the object
illuminated cant move far during the brief pulse during which it is illuminated), much like
a mechanical stroboscope apparently stops motion. Repeating flashes can thus operate as
stroboscopes, too, and consequently some old-time photographers call their xenon-based
flash units strobes. Another common term is speedlight, probably created by anxious marketeers wanting consumers to disregard the lengthy period required by their products to
recharge between flashes.
When one is judging flash units or the flash incorporated into a camera, several issues are
important. These include the guide number, exposure control, red-eye reduction, and the
possibility of external connections.
The guide number of a flash unit allows you to determine the aperture setting of your camera for proper exposure using the flash. Guide numbers depend on film speed, so youll
see them listed as something like Guide number 50 with ASA 100 film. The guide
801
802
numbers for the built-in flashes of digital cameras are calibrated to the sensitivity of the
cameras image sensor, so they dont have to include an ASA number, although some manufacturers note the ASA equivalent. The guide number represents the maximum distance
away a subject can be for a proper flash exposure multiplied by the aperture of the camera
lens you want to use. For example, a guide number of 80 indicates that, at an f-stop setting of f/8, your subject should be no more than ten feet away.
Most of the time you dont have to bother with guide numbers. All digital camera flashes
are automatic. They are designed to provide just the right amount of light to properly
illuminate your subject when taking a picture. But the guide number does tell you the
maximum distance at which the automatic flash exposure system will work. Some manufacturers simply list this maximum distance and forego guide numbers entirely so you
dont have to bother with the math.
Focusing
All lenses must be focused to present a sharp image. At the point of focus, the light rays
emanating from any point on the subject are brought back together again as a single point
in the resulting imageor as close as possible to a single point. Rays of light (photons)
leave the subject in many directions. A few of them strike the lens. The wider the lens
aperture, the more of them the lens can collect. Then the job of the curved surface of the
lens is to nudge the photons in a new direction. The curve of the lens varies the nudging
for the different angles at which the photons left the subject. The problem is that once the
lens nudges the photons, they keep traveling straight. The photons from any given point
will converge on a point at a distance from the lens. At greater or lesser distances than the
point at which they all meet, they may be close together but are still spread apart in the
shape of the aperture. When you focus a lens, you move the lens closer to or farther from
the film or image sensor to make the distance proper. Only when the lens is set properly
will the image be sharp or in focus.
You might think that youd be able to set the focus once and forget it. Light, alas, isnt so
cooperative. Proper focus depends on the distance between subject to lens and lens to
image sensor. When one changes, you have to change the other to keep the image sharp.
You and your camera must compensate for the different distances between the camera and
its subject by focusing the lens. Conventional focusing, the traditional way, is to move the
lens closer to or farther from the image sensorwhen you focus on a more distant image,
the front of the image may extend. A more modern technique is internal focusing, which
alters elements within the lens to change its focusing distance by changing the power of
the lens. Either technique works. Its the lens designers choice.
Focusing
From a more practical standpoint, the means by which the focus gets changed is more
important. Three different types of focusing systems are possible, and all are used on digital cameras. These include fixed focus, manual focus, and auto-focus.
Fixed Focus
Fixed focus actually means no focusing at all. The distance between lens and sensor is
physically fixed. Nothing moves. Thats good because theres nothing for you to adjust.
But its bad, too, because a fixed-focus lens creates a sharp image at only one distance
from its subject. Everything else (and often just plain everything) is a bit bleary, out of
focus.
Although this situation would seem highly undesirable and even unacceptable in a camera,
fixed focus often works and is popular in inexpensive cameras. Fixed focus works best
when its shortcomings are hidden by other shortcomings in the camera system. If a lens
never produces a truly sharp image, youll never be able to tell when it is out of focus. If
an image sensor only produces 320-by-240 pixel resolution, any loss of sharpness may be
lost in the jaggies of the big pixels. Consequently, the limitations of inexpensive lenses
cover up the lack of true focus.
The fixed-focus system takes advantage of another optical property. Wide-angle lenses
make focusing less critical, so most fixed-focus lenses are a bit wider than normal. In
addition, a smaller aperture makes focusing less critical, so fixed focus lens have relatively
small apertures and work only outdoors or with flash.
Manual Focus
A manual focus camera puts you in charge. Typically you twist the lens to move it closer
to or farther from the film to compensate for subjects at different distances. The twist
gives you finer control compared to sliding the lens directly like a trombone. You have to
monitor sharpness by eye, either through the viewfinder or display screen. Its not always
easy to do (especially in dim lighting).
Although most cameras use auto-focusing (discussed next) to relieve you of the chore,
sometimes manual focusing can be preferable. Manual focusing lets you control the look
of the image so you can exploit techniques such as selective focus or intentional blurring.
You can also use manual focus when auto-focusing fails. Although mid-range auto-focusing cameras may not offer you the manual option, youll want to look for the option if
youre serious about photography.
Auto-Focus
A good auto-focus system is more accurate than most people at adjusting the focus of a
camera. Auto-focus systems are generally faster at focusing than are mere human beings.
803
804
And auto-focus systems are more reliable. They do not forget to focus the camera before
that once-in-a-lifetime shot.
Auto-focus systems come in either of two types: passive or active. Passive auto-focus systems look at a scene and attempt to figure out what the correct focus is by making edges
as sharp as possible. Active auto-focus systems work like radar or sonar. They send out a
signal, usually infrared, watch for its reflections, and judge from the time elapsed between
transmission and reception how far the signal has traveled.
Both sorts of auto-focus achieve their intended purpose, at least most of the time. Active
systems can be fooled more easily, however. With sonar-based systems, for example, when
your subject is behind a window, the sound waves from the sonar system bounce off the
glass rather than the subject. As a result, the camera focuses on the window and leaves the
subject a blur. Moreover, active auto-focus systems have a finite range. The sound or light
they send out is effective only for a limited distance. For normal length lenses, that distance is the equivalent of infinity focus, so this shortcoming poses no problem. With long
telephoto lenses, however, the active auto-focus system may run out of range before the
lens.
Passive auto-focus systems work like the human visual system by trying to make lines
sharp. To do so, the passive auto-focus system requires something to focus on, a subject
with a definite outline in high contrast. Without adequate image contrast, the passive
auto-focus system wont work. It is particularly susceptible to darkness because low light
levels usually reduce contrast (less light means less overall range in brightness, which
means less contrast). Passive systems dont work at all in the dark. Active systems have no
problem with darkness.
Most auto-focus systems for cameras dont bring objects into absolute focus. Rather, they
have a stepped focus system that adjusts the lens to accommodate a range of distances for
the subject. Cameras may have as few as three focus ranges or steps. More is better, but
infinity is unnecessary.
White Balance
The white balance control of a camera allows you to adjust it for the varying color temperature of the scene you wish to photograph. (Youll find a full discussion of color temperature in Chapter 24, Video Systems.) Although color temperature is a complex concept,
the effect of white balance is easy to judgescenes taken under incandescent lighting look
abnormally orange. Whereas adjusting for color temperature with chemical cameras
meant toting around a set of color correction filters, most digital cameras have automatic
white balance systems that adjust to the available light. Some digital cameras also let you
make manually settings for sunny, cloudy, incandescent, fluorescent, and flash lighting
conditions.
Image Storage
Image Storage
Film combines two discrete functionsit is both a sensing and storage medium. A digital
camera separates the two functions, leaving you with the need for a storage system. Most
digital cameras use flash memory to store the images they capture.
Inexpensive cameras have only built-in memory, which allows them to hold a finite number of imagestypically about a dozenbefore you need to download them to a computer
or printer.
Better digital cameras use a memory card of some type. These cards are removable so that
you can quickly exchange cards to take more pictures, much like loading a new roll of
film. In fact, some vendors of memory cards call them digital film, which they are not. You
can also slide out a memory card and slide it into a dedicated card reader to transfer the
images into your computer. Chapter 16, Memory, describes the various card types.
To increase the capacity of each card, most digital cameras give you the option of storing
your images at two or more quality levels, which trade good looks for more compact storage. (Almost no one uses native or uncompressed mode to store digital camera images
with only a couple of shots, a memory card is full.) Although some early digital cameras
distinguished their quality levels by their actual resolution, more modern cameras operate
at one or two basic resolutions and trade off higher compression for greater storage capacity. When decoded, the more highly compressed images yield lower qualitythey typically look fuzzierbut nevertheless display at the basic resolution level of the camera.
The compression method of choice is JPEG, which stands for the Joint Photographic
Experts Group, a committee that hammered out standards for the algorithms. JPEG uses
lossy compression, which means that some information gets sacrificed in the effort to
shrink file size. The lost information can never be regained, so once an image is compressed, its original quality can never be restored.
At moderate levels of compression, the losses made by JPEG compression show up as a
loss of image sharpness. However, if you examine a reconstructed JPEG image at high
magnification, you can see patterns that arise from the compression algorithm. As you
increase the JPEG compression ratio, these artifacts become obvious, the image loses definition, and it finally falls apart so that your eye can no longer make sense from it.
The storage or operating modes of most modern digital cameras correspond to the degree
of JPEG compression used. The best quality modeoften called exactly thatuses the
least compression, typically file size reduction by about a factor of four. The next increment down may compress by a factor of 12. Economy or low-quality mode may compress
by a factor of 36 or more. Sometimes a cameras economy mode ratchets resolution down
by one giant notchfor example, to VGA resolution, 640 by 480 pixels.
805
806
Viewfinders
The viewfinder is named after its function: It helps you find the view that you want to
photograph. A plain viewfinder does nothing but give you a view. It can be as simple as
two frames that you line up to locate the image that the camera will picture. Most digital
cameras elaborate on the basic viewfinder design.
Optical viewfinders use a combination of lenses that work like a telescope, giving you a view
at a glance so you dont have to line up frames. The lenses are small and separate from the
lens in the camera used to take the actual picture. The least expensive digital cameras use
simple optical viewfinders.
Many cameras have reflex viewfinders, which use the camera lens as part of the viewfinder.
Through the use of a mirror or prisms, you can see the actual image seen by the lens. A
mirror reflects the lens image up into the viewfinder, the reflection being the cameras
reflex. The most expensive professional digital cameras use true reflex viewfinders (as do a
few lesser expensive models).
A reflex viewfinder allows you to see the effects of zooms. Moreover, the reflex viewfinder
gives you a detailed color image without the need for using a power-hungry (and batterydraining) LCD panel. But because you dont see how the cameras electronics are affecting
the image, you can only guess at how the cameras electronics are dealing with the image.
Digital cameras add video viewfinders or LCD panels that allow you to see exactly the image
seen by the camera lens after its processed by the cameras circuitry. In addition to letting
you preview the image before you capture it, the LCD panel also lets you review the
images youve already captured.
Big-screen or view-screen viewfinders give you a larger LCD display meant to be watched
like a television set instead of a pirates telescope. It gives the best possible view of the
image before and after recording. Using big-screen viewfinders can seem unnatural to
anyone accustomed to the peepshow method of previewing images in other camera formats.
Some digital cameras give you what looks to be a conventional optical viewfinder at the
back of the camerayou peer through it with one eyebut put a small LCD panel at the
other end to create an electronic view finder (EVF). The EVF gives you a preview like an
LCD panel at the back of the camera but is smaller and dimmer (it doesnt have to overpower daylight as does the panel on the back of the camera), so it uses substantially less
power than the big panel on the back.
Some digital cameras switch their EVFs to black-and-white in dimly lit situations to help
you focus more easily. Note, too, that the frame rate of the EVF, often slower than conventional video systems, may distort smooth motion. Its something you have to get
used to.
View Panels
View Panels
In framing a shot, the only image you need to see displayed on the LCD view panel on
the back of your camera is a full-screen look through the lens. For the review of photos
youve previously captured, however, most digital cameras offer additional modes. Some
of these include single-image mode, multiple-image mode, slideshow mode, and zoom
mode.
Single Image
In single-image mode, thats all you seea one-shot view. Most cameras will let you step
through all the images youve exposed so that you can look at them one at a time.
Although you cant compare shots (except in your mind), you will see the greatest possible
level of detail. Single-image mode is the basic viewing mode for digital cameras in review
mode, and usually the only preview view.
Multiple Image
In multiple-image modesometimes called thumbnail modeyour camera breaks its
viewfinder into a matrix of individual images, typically an array of four or nine. Although
each image is a fraction of its individual size (say one-quarter or one-ninth), this mode
allows you to make side-by-side comparisons. You can pick out an image to keep or trash
in one quick glance. This mode is particularly useful when you urgently need more memory and want to eliminate wasted shots to gain memory space.
Slideshow
Long before we had home videos to bore our friends and neighbors through wintry
evenings, God gave us the slideshow. Youd drag out a rolled-up screen that never properly unrolled, plug in the projector, and flash side after slide, one at a time, on the screen.
The slideshow mode of a digital cameras electronic viewfinder operates similarly, a oneafter-another display of the images in its memory. This mode allows you to briefly view all
the photos youve captured before deciding to chuck them all and start over.
The slideshow lets you assess the mass of your work in full detail. At that, it can be quite
useful in making a quick evaluation of your technical prowess. As with the presentation of
single images, it does not allow you to make quick comparisons.
Zoom
The small screen and low pixel count in digital camera LCD displays can make checking
image details difficult. Consequently, many cameras incorporate an electronic zoom mode
in their viewfinder systems. Typically this mode will let the LCD panel display one pixel
807
808
for each pixel in the captured image, which magnifies the image by a factor of two to four.
Although you cannot see the entire image at one time, you can usually pan across its
width and height to check each pixel. If youre critical about the images you take, youll
want zoom mode. For ordinary snapshots, however, you may find it unnecessary and may
not miss the feature in a camera that lacks it.
If you plan only to use your LCD panel to review images, you dont much care about
where it is on your camera. You can just tilt the camera to the viewing angle you prefer. If,
however, you want to use the panel as a viewfinder, the mounting of the LCD is decidedly
important. It must be located so you can view it conveniently while pointing the camera
lens at your subject. Not all digital cameras have LCD panels that let you do this conveniently.
Adjustable panels allow you to swivel the LCD to whatever position is most comfortable
for you to view it. In particular, you can tilt it upward so that it functions as a waist-level
viewfinder, which many professional photographers prefer, particularly for portrait photography.
Connections
Downloading images from camera to computer requires some kind of connection. Among
current digital cameras, four interfaces are popularstandard serial, optical, USB serial,
and FireWire. The technical aspects of each are described in Chapter 11, Ports.
The standard serial connection accommodates the widest range of computers but is also
the slowest connection in general use. A so-called high-speed serial interface in the
computer realm operates at a speed of 115,200 bits per second, a rate that may demand
ten minutes to download the contents of a modest-size camera memory unit. (The Apple
Macintosh fares a bit betterits standard serial interface can accept information about
eight times faster, 921 kilobits per second.) Older digital cameras relied on serial connections, as do some inexpensive, low-resolution models. But serial connections are too slow
to handle todays large image sizes.
Optical connections use the IrDA interface. Although serial in nature, an IrDA connection can handle data speeds up to 4Mbps. The chief shortcoming of optical links is the
need for an IrDA interface at both ends of the connection. Although at one time every
notebook computer had an IrDA interface, the interface is uncommon in desktop computers. Many manufacturers are omitting them from their notebook machines as well.
Although the IrDA interface is fast and convenient, youll want to ensure your computer
can deal with it.
Most newer digital cameras take advantage of the Universal Serial Bus port design. The
latest should use USB Version 2.0. This interface combines high speed with connection
Power
convenience. Even USB 1.0, which operates at about 12 megabits per second, is fast
enough to empty a digital cameras memory card in a minute.
Digital video cameras usually use FireWire connections, which provided the only
consumer-level connection system capable of handling video before the advent of
USB 2.0. The FireWire connection has not caught on among makers of still cameras.
Analog Video Output

Better digital cameras have analog video outputs, which you can use for image preview or
as a video source. For example, you can set your digital camera up in a studio and use a
conventional television monitor for big-screen previews of your shots. Or you can record
images from your digital still camera on a VCR or, with a video input, on your computer.
Note that you cannot get video images at the full resolution of your camera.
Analog video outputs of digital cameras typically follow the standards for the part of the
world in which the cameras are marketed. The standards in Europe, most of Asia, Africa,
and South America differ from those of North America and Japan. Buy a camera meant
for the wrong market, and you may not be able to use the analog video outputs. The analog video outputs of digital cameras meant for North America and Japan usually follow
the National Television Standards Committee (NTSC) video standard, which is fully
described in Chapter 24, Video Systems.
Power
Digital cameras are electronic devices and consequently need a steady supply of electricity.
With todays technologies, their needs border on the prodigious and often sneak to the far
side of the border. That power has to come from somewhere, and the places of choice are
twofrom self-contained batteries or from utility-supplied electricity through an external
transformer, the ubiquitous power brick.
Batteries
Unlike camcorders, the batteries for most digital cameras are standard sizes. Most digital
cameras use standard AA or AAA batteries, typically about four (the number needed to get
enough voltage for conventional logic circuits). Most manufacturers generously include
batteries with their products.
Digital cameras are high-current devices. They cannot use standard or heavy-duty batteries based on carbon-zinc technology. They require alkaline or rechargeable cells. In fact,
some manufacturers warn to use their products only with rechargeable batteries, such as
nickel-cadmium cells or nickel-metal hydride (NiMH) cells. Chapter 31, Power, discusses these battery technologies in detail.
809
810
Batteries may seem to behave strangely under the high-current demands of digital cameras. For example, while alkaline cells may last long enough to capture a dozen images,
rechargeable cells of the same size may work for 100 images or more. Subtle differences in
battery construction used by different brands can have a big effect on battery capacity.
Some manufacturers cells may appear to fail after only a few exposures. Such batteries are
far from dead, but they can no longer generate the high current your digital camera
requires.
The high-current demand also makes battery life unpredictable. Youll always want to
carry spares.
When loading fresh batteries in your camera, never mix battery technologies or even battery brands. Different battery brands may use slightly different physical constructions or
even different chemical mixes with the possible result of the internal resistance of the
brands of cells being different, resulting in some cells working harder (and draining faster)
than others. In other words, make sure the cells you use are identical.
AC Power
Nearly all digital cameras make provisions for plugging in an AC adapter. Many, however,
make the AC adapter an option. If you work primarily in a studio, the AC adapter is a
worthwhile expense because it will save you the cost of batteries and the time needed to
change them.
You can substitute a generic AC adapter for the official one offered by the camera manufacturer. Check your cameras specification sheet to determine the current your camera
requires as well as the voltage and polarity of the AC adapter. Often polarity is described
graphically with a plus or minus sign pointing to a dot inside a circle. The dot corresponds to the central conductor of the power connector. The only important issue is that
the diagram thats shown in your cameras specifications match that of the AC adapter you
choose.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
24
Display Systems
Seeing is believing. If you couldnt see the results of your calculations
or language manipulations, the personal computer would be worthless
as a tool. You need some way of viewing the output of the computer system to know what it has done and why youre wasting your time feeding
it data.
Out of all potential technologies for giving you feedback about what
your computer is doing, the one almost universal choice is the one you
see on your screen. But its what goes on behind the screen that really
counts. The screen just shows you the results of a lot of technologies and
systems working together.
The total entity made from these technologies is your computers display
system. At the hardware level, it comprises two distinct parts. The display
is the screen you see, either as a separate monitor or as a panel thats a
permanent part of your computer. The display adapter, which may be a
separate expansion board or built in to your systems motherboard, converts the logical signals in your computer to the form digestible by the
display.
The first display systems used by personal computers were based on television sets. In fact, many were television sets, adapted to the job with a
modulator that turned video signals from your computer into Channel 3
or 4. Todays display systems are vastly more sophisticated than mere
television sets, often having greater resolution than is promised by even
High Definition Television (HDTV) systems. But at heart, computer
display systems remain video systems, based on the same technologies as
the lowly TV.
814
Video Technology
As far as most of your computer is concerned, its job is done when it writes an image to
the frame buffer. After all, once youve got all the pixels arranged the way you want them
in memory, shipping them off through a cable to your display should be a simple matter.
It isnt. The image must be transformed from its comparatively static position in screen
memory to a signal that can traverse a reasonable interfaceafter all, the cable alone for a
parallel interface capable of moving data for nearly a million pixels would likely be thicker
than the average computer. Consequently, the frame-buffer image must be converted into
a stream of serial data for transmission.
The need for this kind of image transformation became evident when television was
invented. The system developed by Philo T. Farnsworth (one of several folks credited
with inventing television) relied on scanning images that naturally produced a serial data
stream. Although television has improved somewhat since the 1920s when Farnsworth was
developing it (at least technical standards have improvedthe quality of entertainment is
another matter entirely), the transmission system remains essentially the same for todays
analog television signals as well as the connections between computers and their displays.
Scanning
Serializing images for the first television system was inherent in the way the cameras captured the images using the technique of raster scanning. The first television cameras traced
an electron beam across the projection of an image on a special material that changed its
electrical characteristics in response to the bright and dark areas of a scene. By focusing
the electron beam, the camera could detect the brightness of a tiny spot in the scene. By
dividing the scene into dots the size of its sensing spot and rapidly examining each one,
the camera could gather all the data in the image.
Although theres no naturally required order to such an examination, the inventors of television looked to what we humans have been doing for centuriesreading. We read text
one line at a time, progressing from left to right across each line. The various inventors of
television (and there are many competing claims to the honor) followed the same pattern,
breaking the image into lines and scanning the dots of each line from left to right.
To make a scan in a classic television camera, the electron beam sweeps across the image
under the control of a combination of magnetic fields. One field moves the beam horizontally, and another vertically. Circuitry in the camera supplies a steadily increasing voltage
to two sets of deflection coils to control the sweep of the beam. These coils are electromagnets, and the increasing voltage causes the field strength of the coils to increase and
deflect the beam further. At the end of the sweep of a line, the field that controls the horizontal sweep of the electron beam is abruptly switched off, returning the beam to the
starting side of the screen. Likewise, when the beam reaches the bottom of the screen, the
CHAPTER 24 DISPLAY SYSTEMS
Video Technology
field controlling the vertical sweep switches off. The result is that the electron beam follows a tightly packed zigzag path from the top of the screen to the bottom.
The primary difference between the two sweeps is that several hundred horizontal sweeps
take place for each vertical one. The rate at which the horizontal sweeps take place is called
the horizontal frequency, or the line rate, of the display system. The rate at which the vertical
sweeps take place is called the vertical frequency, or frame rate, of the system because one
complete image frame is created every time the beam sweeps fully down the screen.
The television receiver scans the inside of the cathode ray tube (CRT) in exactly the same
fashion. In fact, its electron beam is precisely synchronized to that of the camera. The
one-line-at-a-time, left-to-right scan nicely accomplishes the required dimensional conversion.
The video circuits of your computer have to carry out a similar conversion. The only difference is that the image is laid out in a logical two-dimensional array in memory instead
of a physical two-dimensional array on a photosensitive layer inside the camera tube (or
more likely today, a charge coupled device, or CCD).
To make a scan of the video buffer is a lot easier than sweeping an electron beam. Your
computer need only read off addresses in the video buffer in sequential order, one row at a
time. To carry out this task, your computer uses a special electronic circuit called the video
controller, which scans the memory addresses, reads the data value at each address, and
sends the data out in one serial data stream.
Synchronizing Signals
In television, the biggest complication of the scanning systems is ensuring that the camera
and television set displaying the image both scan at exactly the same position in the image
at exactly the same time. The frequencies used by horizontal and vertical scanning must
exactly match. In addition, the camera and television must use exactly the same starting
position.
To keep the two ends of the system locked together, television systems use synchronizing
signals. They take the form of sharp pulses, which the circuitry inside your monitor converts to the proper scanning signals. The television camera generates one special set of
pulses at the beginning of each line and another at the start of each image frame. The
television knows to start each line when it receives the pulses.
In your computer, the video controller generates similar synchronizing signals for exactly
the same purpose. It sends out one (the horizontal synchronizing signal) before each line
in the image and one (the vertical synchronizing signal) at the beginning of each frame.
The monitor uses the pulses to trigger its sweep of each line and to reset to the top of the
image to start the scan of the next frame.
815
816
The video controller doesnt scan at just any frequency. It uses standard frequencies,
which vary with the geometry of the imageits height and width along with the frame
rate. The monitor is tuned to expect these frequencies, using the synchronizing signals
only to achieve a precise match.
In conventional television, the synchronizing signals were designed to sit invisibly in the
same single data stream that conveyed the picture information. In modern production studios and the connection between your computer and monitor, however, the synchronizing
signals are usually kept separate from the picture information. Actually, there are four
common ways of combining or not combining video data and synchronizing signals.
These include the following:
Composite video. The all-together-now television approach that puts all video data
and the two required synchronizing signals into one package for single-wire or
single-channel transmission systems.
Composite sync. Combines the horizontal and vertical synchronizing signals
together and puts them on one wire. Another, separate wire carries the image data.
Separate sync. Gives a separate wire and connection to the image data, the horizontal, and the vertical synchronizing signals.
Sync-on-green. Combines the vertical and horizontal synchronizing signals
together and then combines them with the data for the green data channel.
In any of these four systems, the relative timing of the synchronizing and data signals is
the same. The chief difference is in the wiring. A composite video system requires only
one wire. The other systems use three wires for data (one for each primary color). Syncon-green therefore requires only three connections, composite sync requires four (three
colors, one sync), and separate sync requires five (three colors, two sync). The standard
video system in most computers uses composite syncthe signal monitor cable has four
separate connections for image data and synchronizing signals.
Retrace
In a television signal, the data corresponding to the dots on the screen doesnt fill a video
signal wall-to-wall. The physics of the first television systems saw to that. To make the
image you see, the electron beam in a conventional television picture tube traces a nearly
horizontal line across the face of the screen then, in an instant, flies back to the side of the
screen from which it started but lower by the width of the line it already traced out. This
quick zipping back is termed horizontal retrace, and although quick, it cannot take place
instantly because of the inertia inherent in electrical circuits. Consequently, the smooth
flow of bytes must be interrupted briefly at the end of each displayed line (otherwise the
video information would vanish in the retrace). The video controller must take each
retrace into account as it serializes the image.
Video Technology
In addition, another variety of retrace must occur when the electron beam reaches the
bottom of the screen when it has finished painting a screen-filling image: vertical retrace.
The beam must travel as quickly as possible back up to its starting place, and the video
controller must halt the flow of data while it does so.
Blanking
During retrace, if the electron beam from the gun in the tube were on, it would paint a
bright line diagonally across the screen as the beam returns to its proper position. To prevent the appearance of this distracting line, the beam is forcibly switched off not only during retrace but also during a short interval on either side to give the beam time to
stabilize. The interval in which the beam is forced off and cannot be turned on by any
degree of programming is called blanking because the electron beam can draw nothing but
a blank on the screen.
The classic television signal cleverly combines synchronization, retrace, and blanking
together. The horizontal synchronizing signal is a strong pulse of the opposite polarity of
the image data that lasts for the retrace period. The negative nature of the signal effectively switches off the electron beam, and the frequency of the signal effectively synchronizes the image.
Front and Back Porches

Most computer monitors dont fill their entire screens with data. They center (or try to)
the image within darkened borders to minimize the image distortions that sneak in near
the edges of the screen. To produce these darkened, protected areas, the electron beam is
held at the level that produces a black image for a short while before and after the data of
each image line is displayed. The short interval before the data of a line begins is termed
the front porch of the signal. The interval after the end of the data but before the synchronizing signal is called the back porch. If you examined the signal, youd see that it dips
down for blanking and pops up to an intermediate height (called black level by broadcasters) to create the porches between blanking and data. Use your imagination and the blacklevel signals look like shelvesor porches.
Vertical Interval
The period during which the screen is blanked during the vertical retrace is called, appropriately, the vertical interval. Its physical manifestation is the wide black horizontal bar
thats visible between image frames when your television screen or computer monitor picture rolls and requires adjustment of the vertical hold control. The big black bar corresponds to the time during which the signal carries no video information.
817
818
The vertical interval is a carryover from the early days of television when vacuum tube
electronics needed time to recover between fields and frames. This allowed voltages
inside the circuitry of the TV set to retreat to the proper levels to begin the next field.
Modern electronicssay, for example, those of televisions made in the last 30 years
dont really require the long duration of the vertical interval. Consequently, broadcasters
have found the time devoted to it useful for stuffing in extra information. Television stations add a vertical interval test signal (VITS) to monitor operation of their transmitters
and associated equipment. The text for the closed captioning system is also encoded during the vertical interval, as is all sorts of other miscellaneous data.
Video Integration
The signals between computers and their monitors dont follow the same standards used
by broadcast television and studio video systems. Nevertheless, these standard video signals have become an intrinsic part of both what computers do and what you expect from
your computer. After all, computers are now regularly used to produce videos, including
those for presentation on television, even on networks. At the other end of the channel,
computers display video imagesfrom a television using a TV adapter board or from a
DVD drive. Video images inevitably confront the computer display system, and dealing
with them can be a challenge.
Many display adapters make conventional video signals along with those meant for your
computer display. Add a video-capture board to your computer, and you can turn conventional video signals into digital form so you can edit home movies or package them on
CD or DVD. You can even watch television on your computer monitor using a technique
called video overlay.
Standards
When it comes to images, the standard most widely used stares you in the face, literally,
for several hours a day. The unblinking eye of the television set defines the most widely
used image communication system in the world. Considering when the basic television
standards were created, their longevity has been amazing, particularly compared to the
short tenure of computer standardsthe basic television signal was defined half a century
ago. Only recently has it come under threat by digital and high-definition technologies.
(The Federal Communications Commission hopes to turn off all conventional television
signals by 2007, substituting a new all-digital television standard.)
First a bit of definition. The word video means I see in Latin. The word television is a
hodge-podge derived from the Greek for distant and Latin for sight. Television is
what is broadcast or transmitted over a distance. Video is up close and personal, the signals inside a studio or your home.
Video Integration
When we talk of video among computers, however, we mean an electrical signal that
encodes an image in raster form, hence the term video board thats often used instead of
graphics adapter. When people involved with television speak of video, they mean a particular form of this signal, one with well-defined characteristics. A television transmitter modulates a carrier wave with this video signal to make the broadcasts to which you tune in.
Video signals range in frequency from zero to a half-dozen megahertz. Television signals
start out at 60MHz and extend upward to nearly ten times that. Television sets tune in television signals, receiving them on their antenna inputs. Monitors display video signals.
Although one standard currently dominates video signals in the United States and Japan,
other standards are used elsewhere in the world. In addition, a secondary standard termed
S-video appears in high-quality video applications.
NTSC
The most common form of video wears the designation NTSC, which stands for National
Television Standards Committee, an industry organization formed in the early 1950s to create a
single signal standard for color television. At the time, CBS had been broadcasting for over a
year with an electromechanical color system that essentially spun a color wheel in front of
the camera and a matching wheel in front of the monitor. RCA, owner of rival NBC, proposed an all-electronic alternative. The RCA system had the advantage that it was backwardly compatible with black-and-white television sets, whereas the CBS system was not.
The NTSC was formed chiefly to put an impartial stamp of approval on the RCA system.
In the RCA/NTSC system, each pixel gets scanned in each of the three primary colors.
Although studio equipment may pass along the three colors separately like the RGB signals in computers, for broadcast they are combined together with synchronizing signals to
create NTSC video.
The magic is in the combining process. The NTSC system packages three channels of
color into one using some clever mathematical transformations.
First, it transforms the color space. For compatibility with monochrome, NTSC combines
all three color signals together. This produces a signal called luminance, which encodes all
the brightness information in a television image. The luminance signal is essentially a
monochrome signal and produces an image entirely compatible with black-and-white television sets. The name of the luminance signal is often abbreviated as Y.
Next, the NTSC system creates two signals encoding color, or rather the differences
between the color signals and the luminance signal. One signal (called I in the NTSC system) encodes the difference between luminance and the red signal, and another (called Q)
encodes the difference between luminance and the blue signal. Subtract the first from
luminance, and you get red again. Subtract the other, and you get blue. Subtract both, and
the remainder is green.
819
820
Next, the NTSC system combines the two difference signals into a single signal that can
carry all the color information, called chrominance (abbreviated as C). Engineers used
quadrature modulation to combine the two signals into one. The result was that colors
were encoded into the chrominance signal as different phase angles of the signal.
Together the luminance and chrominance signals provided a guide to a map of colors, a
polar chart. The chrominance encodes the angle between the color and the X-axis of the
chart, and the luminance indicates the distance from the origin to the color.
Finally, to fit the chrominance signal in where luminance should only fit, the NTSC engineers resorted to putting chrominance on a subcarrier. That is, they modulated a carrier
wave with the chrominance signal and then added it to the luminance signal. Although
the subcarrier had much less bandwidth than the main luminance channel, the process
was effective because the human eye is less sensitive to color differences than brightness
differences.
The NTSC chose a frequency of 3.58MHz as the color subcarrier frequency. The
chrominance is an amplitude modulated signal on a 3.58MHz carrier. To avoid interference with the luminance signal, the NTSC process eliminates the carrier and lower sideband of the chrominance signal after the modulation process.
The NTSC process has two drawbacks: The luminance signal must be cut off before it
reaches 3.58KHz to avoid it interfering with the subcarrier. This frequency cap limits the
highest possible frequencies in the luminance signal, which means that the sharpness of
the image is reduced from what it would be when using the full bandwidth (4.5MHz for
the video signal) of the channel. Chrominance carries even less detail.
The basic frame rate of a color video signal is about 29.97 per second. Each frame is made
from two interlaced fields, so the field rate is 59.94Hz. Each frame is made from 525
lines, of which about 480 are visible, and the rest are devoted to vertical retrace. Ideally, a
studio image would have about 640 pixels across a line. However, the 3.58MHz bandwidth imposed by the NTSC color process constrains the luminance signal bandwidth to
400 to 450 pixels horizontally. Although that might sound paltry, a good home VCR may
be able to store images with only about half that resolution.
Black-and-white signals are different. They use a 30Hz frame rate (each with two fields)
and lack the color subcarrier. As a result, they can be sharper than color signals because
the entire 4.5MHz bandwidth can be devoted to the image.
Instead of NTSC, most of the world uses a color system called PAL, which stands for
Phase Alternating Line. France and most of the nations that once formed the USSR, such
as Russia and Ukraine, use a system called SECAM, which stands for Sequence Couleur
Memoire (in English, sequential color with memory). These video standardsand the
equipment that follows themare mutually incompatible.
Video Integration
S-Video
The constraints of NTSC color are required because of the need for backward compatibility. The color signal had to fit into exactly the same bandwidth as black and white. In
effect, NTSC gives up a bit of black-and-white resolution to fit in the color information.
Video signals that never make it to the airwaves need not suffer the indignities required by
the NTSC broadcast standard. Studio signals have always transcended broadcast standardsstudio RGB signals have full-bandwidth, high-resolution (640-pixel) images in
each of their three colors. To raise home viewing quality, VCR designers came up with a
way to get more quality in color signals by avoiding the NTSC process.
The part of the NTSC process that most limits visual quality is the squeezing of the
color signal onto its subcarrier. By leaving the video in two parts, separate luminance
and color signals, the bandwidth limitation can be sidestepped. This form of video is
termed S-video, short for separate video. High-end VCRs, camcorders, and monitors
use often use S-video signals.
Other than not modulating chrominance onto a subcarrier, the color-encoding method
used by S-video is identical to that of NTSC. The three RGB color signals are combined
into luminance and chrominance using exactly the same formulae. Although you cannot
substitute one signal for the other, the innards of S-video monitors need not be radically
different from those of NTSC displays. The level of quality is often quite visibly different.
S-video components may have twice the horizontal resolution as composite video.
Note that once a signal is encoded as NTSC, information is irretrievably lost. Theres no
point to decoding an off-the-air television signal to S-video. The only time S-video helps
is when you have a source of the signals that has never been NTSC encoded.
Video Overlay
To put active video in a window on your computer screen, many graphics adapters use a
technique called video overlay, which allows them to do most of the image processing in
hardware rather than software (where it would slow down the rest of your system).
The video overlay process borrows from an old television technology called chroma
keying. This process works by substituting one image for a key part of another image.
Typically the keyan area in the image being shown on the screenwould be identified
by its color or chroma. Hardware then substitutes another image for the key color. In
computers, the graphics adapter paints the television image into the key area, the windows, on the screen. Traditionally in television, the color of choice is a sky blue. This
color is preferred because its optically the opposite of average Caucasian flesh tones, so
its least apt to make parts of people disappear on the screen. Sometimes the process is
used in reversefor example, a weather reporter stands in front of a chroma-key blue
821
822
screen, and the background gets filled with a weather map or satellite photo. Television
people call this technique blue screening.
In video overlay, the driver software uses standard Windows instructions to paint a window on the screen, keyed to a special value (it need not be blue). The graphics adapter
intercepts the signals destined for your monitor and substitutes the video signal where it
finds the keyed windows. Your software and even your computers microprocessor never
need to deal with processing the video. In fact, the video never even makes it as far as the
expansion bus. It is isolated on the overlay board. You get full-motion video on your
screen with virtually no impact on the performance of your computer.
Video Capture
Gathering up video images so that they can be used by your programsvideo capture
requires hardware that combines aspects of a more traditional video board and a digital
video camera. Traditional video images are analog signals and require an analog-to-digital
converter (A-to-D converter) to put them in a form usable by your computer.
The A-to-D converter works by sampling the voltage level of the video at each pixel position of the video image and assigning a digital value to it. In most systems, the signal is
first decoded into your computers standard RGB format to determine the strengths of
individual colors before sampling. Typically the image gets stored in a buffer and is sampled from the buffer rather than from the real-time video signal. The buffer helps bridge
between the different timings and formats of the video signal and its target digital form.
Todays video-capture systems can snag active video in real time. Most can also snatch a
single frame to use as a still image. In single-frame mode, the board operates as a videoframe grabber.
Image Compression
Nothing eats memory and storage faster than full-motion, full-color video. The math is
enough to make the manufacturers drool. Full-color video, which requires three bytes per
pixel, at 640480 resolution, equals nearly 1MB of digital data per frame. At the 30 frames
per second used in the United States and most of the Western Hemisphere (or even the
25 frames per second standard in Europe and elsewhere), a video producer would easily
use up 1GB of hard disk space in storing less than one minute of uncompressed digital
video information.
Digital video systems with reasonable storage needs are possible only because of video
compression, much as digital photographs are made more compact with image compression.
Video compression takes the next step and analyzes not only the image data but the
changes in sequential images. Unlike data-compression systems, which reduce the size of
Image Compression
a file without losing any information, image- and video-compression systems are lossy.
That is, they throw away informationusually the part that is least perceptiblethat is
lost and can never be recovered.
Normal still-image compression programs work two dimensionally, analyzing areas and
reducing the data required for storing them. The most popular is called JPEG, which
stands for the Joint Photographic Experts Group, the group that developed the standard.
Video compression works three dimensionallyin addition to compressing individual
areas, it processes changes that occur between images in the time dimension. It takes
advantage of how little actually changes from frame to frame in a video image. Only the
changes get stored or transmitted; the static parts of the image can be ignored. For example, when someone moves against a backdrop, only the pixels in the moving character
need to be relayed to the data stream. The most popular form is called MPEG, for the
Motion Picture Experts Group, an organization similar to JPEG (part of the same overall
body) but separate from it.
Filters and Codecs

Compressing still- and video-image data streams is so different that developers use distinct terminologies when speaking of the conversion process. Moreover, they even handle
the conversion software differently.
The program routines that compress still images are usually termed filters. Most graphic
applications have several filters built in to handle a variety of different compression systems and file formats.
Video compression requires either a software- or hardware-based processor that is termed
a codec, short for coder/decoder. The most efficient software codecs are proprietary
designs that rely on patented technology. Each has its own advantages (such as speed of
processing, high compression ratio, or good image quality) that make it best suited for a
given type of application. Consequently, many codecs remain in common use. Most multimedia applications include the appropriate codec in their playback software or work with
those assumed to be installed in your operating system.
JPEG
JPEG is at its best compressing color images, because it relies on psycho-visual perception
effects to discard image data that you might not be able to perceive. It also works on
grayscale images but yields lower compression ratios at a given quality level. It does not
work well on monochrome (two-tone or black-and-white) images and requires that colormapped images be converted to a conventional, continuous-tone color format before processingwhich, of course, loses the compression effect of the color mapping.
823
824
JPEG processing involves several steps, some of which are optional. Several of these steps
may reduce the amount of detail in the image and therefore its quality. The JPEG standard allows you to select the amount of information thats thrown away in these steps so
you can control how well an image reconstructed from the compressed data will resemble
the original. One option is lossless, which throws away no information other than that
which would be redundant. This typically compresses an image file to 50 percent of its
original size. Even invoking lossy compression, you can reconstruct an image visually
indistinguishable from the original with a reduction to 33 percent of the original. The loss
becomes apparent somewhere around reductions of 5 to 10 percent of the original data
size. You can brute-force the image data down to 1 percent of the original size, although
the results will resemble more a mew work of computer art than whatever masterpiece
you started with.
MPEG
The videos that youre most likely to display on your computer use MPEG compression.
As with JPEG, MPEG is a committee working under the joint direction of the
International Standards Organization (ISO) and the International Electro-Technical
Commission (IEC). The formal name of the group is ISO/IEC JTC1 SC29 WG11the
stuff after the organization simply further delineates the organization (JTC stands for
Joint Technical Committee, Subcommittee 29, Work group 11). It began its life in 1988
under the leadership of Leonardo Chiariglione and Hiroshi Yasuda.
Despite the similarity of names, the JPEG and MPEG groups are separate and share few
members. Some of the technologies used by the two compression systems are similar, but
they are meant for different kinds of data. The most prominent point of divergence is that
MPEG achieves most of its data reduction by compressing in the time dimension, encoding only differences between frames in video data.
MPEG includes multiple standards for encoding not only video but also the accompanying audio. Over the years it has progressed through several levels with increasing sophistication and quality.
MPEG-1
The first MPEG standard, now usually called MPEG-1 but formally titled Coding of
Moving Pictures and Associated Audio for Digital Storage Media at up to about 1.5
MBit/s, became an international standard in October, 1992. It has four parts. The actual
compression of video and video signals is covered under International Standard 11172-2.
Related parts describe the compression of audio signals, synchronizing audio and video,
and testing for compliance with the standard.
MPEG-1 is used by CD-i (interactive compact discs) because it achieves a data rate that is
within the range of CD drives. To get down that low with the technology existing at the
Image Compression
time the standard was developed, the system had to sacrifice resolution. At best, an
MPEG-1 image on CD-i has about one-quarter the pixels of a standard TV picture.
MPEG also requires hefty processing power to reconstruct the moving image stream,
which is why CD-i players can display it directly to your TV or monitor, but only the
most powerful computers can process the information fast enough to get it to your display
without dropping more frames than an art museum in an earthquake. If youre used to the
stuff that pours out of a good VCR, this early MPEG looks marginal, indeed.
MPEG-2
MPEG-2 was meant to rectify the shortcomings of MPEG-1, at least in regard to image
quality. The most apparent difference appears on the screen. The most common form of
MPEG-2 extends resolution to true TV quality (720 pixels horizontally and 480 vertically)
while allowing for both standard and wide-screen formats (4:3 and 16:9 aspect ratios,
respectively). Although MPEG-2 benefits from advances in compression technology, this
higher quality also demands more data. The TV-quality image format requires a bandwidth
of about 4Mbps. Beyond that, the MPEG-2 standard supports resolutions into ionspheric
levels. All MPEG-2 chips are also required to step back and process MPEG-1 formats.
In addition to high-quality video, MPEG-2 allows for 5.1 audio channelsthat is, left and
right main channels (front), left and right rear channels (surround), and a special effects
channel for gut-thumping rumbles limited to no higher than 100Hz. (The .1 in the
channel description refers to the 100Hz limit.) MPEG-1 only allows for a single stereo
pair.
MPEG-3
What was initially MPEG-3 has been incorporated into MPEG-2. The concept behind
MPEG-3 was to make a separate system for High Definition TV for images with resolutions up to 1920 by 1080 pixels with a 30Hz frame rate. Fine-tuning the high levels of
MPEG-2 worked well enough for HDTV images that there was insufficient need to support a separate standard.
MPEG-4
Adopted as ISO/IEC 14496 in early 1999, MPEG-4 defines a standard for interactive
media. Although it incorporates a compression scheme (actually, several), it looks at images
entirely differently than does MPEG-2. Instead of compressing an entire scene as an
image, MPEG-4 makes the scene from video objects, each of which the standard allows to
be independently defined and manipulated. A single scene may incorporate several video
objects as well as a background. The standard also allows for conventional rectangular
images such as movie frames as a special class of video objects. Its compression is optimized for low rates suited to moving images through conventional modems for videophones or small-screen video conferencing. For example, such images may have low
resolution (about 176144 pixels) and a low frame rate, on the order of 10Hz.
825
826
In effect, MPEG-4 is a standardization of many of the features of a 3D accelerator (an

unremarkable convergence in that both are designed for the same purposethe effective
presentation of action video). It provides compression algorithms for video, textures, and
wire-frames as well as a system for manipulating objects, scenes, and sequences. In addition, it incorporates MPEG-J, an application program interface that allows combining
MPEG-4 data with Java code to make a cohesive multimedia playback environment.
Other MPEGs
MPEG-7 is a content-retrieval standard rather than a compression or storage standard,
called a Multimedia Content Description Interface. It will provide a standard for describing
different types of multimedia information that links the description to the content itself.
By searching the description, the link will allow you to quickly and efficiently find the
material that interests you. MPEG-7 is designed to let you search any kind of medium
from still images and graphics to audio to conventional and 3D movies. Its designers even
envision its extension to classifying facial expressions and personal characteristics.
MPEG-5 and MPEG-6 are not defined.
Computer Display Standards

The display systems used by todays computers follow several standards that dictate key
aspects of their signals, including the onscreen resolution, the number of colors on the
screen, and the signals used to connect monitors to your computer.
The starting point in a modern computer display is called VGA, named after IBMs pioneering Video Graphics Array, which was introduced in 1987. This little bit of history
remains the default mode of nearly every computer video system. Switch on your computer, and it begins life as a VGA system. Only after your operating system loads its video
drivers does it switch to the resolution you set. The first screen you see, usually the
Windows splash screen announcing your investment in making Bill Gates fortune, pops
on your monitor courtesy of VGA. When you cant get anything to work and your
Windows-based computer snarls into service in safe mode, VGA is what you see. Much as
you might want, you cant get away from VGA.
But VGA is still just for starters. Except for the lowliest of notebook systems, every computer today strives higher. The best display systems put more than six times as much
detail on your screen in more colors than you can name (or even distinguish). At one time,
standards were important for properly displaying these higher resolution modes. But
because the use of driver software has removed the need for hardware to exactly match
any set configuration, most of the older standards survive only as names that identify
onscreen resolutions or operating modes. Table 24.1 lists the most popular display standards for personal computers.
Display Adapters
TA B L E 2 4 . 1
Computer Video Standards
Designation
Name
Resolution
Maximum Colors
MDA
Monochrome Display Adapter
720 by 350
CGA
Color Graphics Adapter
640 by 200
16
EGA
Enhanced Graphics Adapter
640 by 350
256
VGA
Video Graphics Array
640 by 480
256/65,536
SVGA
SuperVGA
800 by 600
16.7 million
XGA
Extended Graphics Array
1024 by 768
16.7 million
SXGA
SuperXGA
1280 by 1024
16.7 million
Display Adapters
The hardware that changes your pulsing digital computers thoughts into the signals that
can be displayed by a monitor is called the display adapter. Over the years, the display
adapter has itself adapted to the demands of computer users, gaining color and graphics
capabilities as well as increasing its resolution and range of hues. In most machines, the
display adapter is a special expansion board that serves primarily to make graphic images;
hence, the display adapter is often called a graphics board. Because the graphics board sends
out signals in a form that resembles (but is not identical to) that of your home video system, it is often termed a video board. Notebook computers lack video boardsthey typically lack any conventional expansion boards at allbut all of them also include display
adapter circuitry on their motherboards.
No matter its name, the function of display adapter circuitry is the samecontrol. The
adapter controls every pixel that appears on your computer display. But there is one more
essential element. Just any control wont do. Give a room full of monkeys control of a million light dimmers (youll need a mighty large room or a special breed of small, social
simians), and the resulting patterns might be interestingand might make sense at about
the same time your simians have completed duplicating the works of Shakespeare. The
display adapter circuitry also organizes the image, helping you make sense from the chaos
of digital pulses in your computer. It translates the sense of your computers thoughts into
an image that makes sense to you.
The video circuitry of your computer, whether on a dedicated display adapter or part of
the motherboard circuitry in the chipset, performs the same functions. In its frame buffer
(or in main memory in systems using Unified Memory Architecture), it creates the image
your computer will display. It then rasterizes the memory-mapped image and converts the
digital signals into an analog format compatible with your monitor.
827
828
The modern video board usually has five chief circuits that carry out these functions,
although some boards lack some of these elements. A graphics accelerator chip builds the
image, taking commands from your software and pushing the appropriate pixel values into
the frame buffer. Memory forms the frame buffer that stores the image created on the
board. A video controller reads the image in the frame buffer and converts it to raster form.
A RAMDAC then takes the digital values in the raster and converts them into analog signals of the proper level. Finally, a video BIOS holds extension code that provides the video
functions your system needs as it boots up.
Accelerator Chips
Of all the chips on a video board, the most important is the graphics accelerator. The chip
choice here determines the commands the board understandsfor example, whether the
board carries out 3D functions in its hardware or depends on your computer to do the
processing of 3D effects. The speed at which the accelerator chip operates determines
how quickly your system can build image frames. This performance directly translates
into how quickly your system responds when you give a command that changes the screen
(for example, dropping down a menu) or how many frames get dropped when you play
back a video clip. The accelerator also limits the amount and kind of memory in the frame
buffer as well as the resolution levels of the images your computer can display, although
other video board circuits can also impose limits. In short, the graphics accelerator is the
most important chip in the entire video system.
That said, the accelerator is optional, both physically and logically. The oldest display
adapters lack accelerators; hence, they are not accelerated, which means your computers
microprocessor must execute all drawing instructions. In addition, even boards with accelerators may not accelerate all video operations. The board may lack the knowledge of a
specific command to carry out some video task, or the boards driver software may not
take advantage of all its features. In such circumstances, the drawing functions will be
emulated by a Hardware Emulation Layer (often abbreviated HEL) in your operating systemwhich means your microprocessor gets stuck with the accelerators drawing work.
As computers have evolved, the need for graphics processing has shifted back and forth
between the accelerator and your systems microprocessor. For example, the MMX
instructions of newer Pentium microprocessors overlap the functions of graphics accelerators. Streaming SIMD extensions add performance to most graphics operations, complementing MMX. In Unified Memory Architecture (UMA) computers, these technologies
can take the place of a dedicated graphics accelerator. In computers with frame buffers,
these features work in conjunction with the graphics accelerator. They speed up your
computers ability to calculate what images look likefor example, decompressing stored
images or calculating wire-frames for your drafting program. But the final work, actually
painting the images that will appear on your screen, relies on the graphics accelerator.
Display Adapters
The graphics accelerator is an outgrowth of an older chip technologythe graphics

coprocessor. An early attempt to speed up the display system, the graphics coprocessor was
introduced as a supplemental microprocessor optimized for carrying out video-oriented
commands.
The graphics coprocessor added speed in three ways. By carrying out drawing and imagemanipulation operations without the need for intervention by the microprocessor, the
coprocessor freed up the microprocessor for other jobs. Because the graphics coprocessor
was optimized for video processing, it could carry out most image-oriented operations
faster than could the microprocessor, even if the microprocessor was able to devote its full
time to image processing. The graphics coprocessor also broke through the bus bottleneck that was (at the time of the development of graphics coprocessor technology) choking video performance. When the microprocessor carried out drawing functions, it had to
transfer every bit bound for the monitor through the expansion busat the time, the slow
ISA bus. The coprocessor was directly connected to the frame buffer and could move
bytes to and from the buffer without regard to bus speed. The microprocessor only
needed to send high-level drawing commands across the old expansion bus. The graphics
coprocessor would carry out the command through its direct attachment to the frame
buffer.
The workstation market triggered the graphics coprocessor. Microprocessor-makers
altered their general-purpose designs into products that were particularly adept at manipulating video images. Because the workstation market was multifaceted, with each different hardware platform running different software, the graphics coprocessor had to be as
flexible as possibleprogrammable just like its microprocessor forebears.
These coprocessors joined the computer revolution in applications that demanded highperformance graphics. But the mass acceptance of Windows made nearly every computer
graphics intensive. The coprocessor was left behind as chipmakers targeted the specific
features needed by Windows and trimmed off the excessprogrammability. The result
was the fixed-function graphics coprocessor, exactly the same technology better known
now as the graphics accelerator.
The most recent evolution of graphics acceleration technology has produced the 3D accelerator. Rather than some dramatic breakthrough, the 3D accelerator is a fixed-function
graphics coprocessor that includes the ability to carry out the more common 3D functions
in its hardware circuitry. Just as an ordinary graphics accelerator speeds up drawing and
windowing, the 3D accelerator gives a boost to the 3D rendering. Nearly all of todays
video boards are equipped with a 3D accelerator. The technology is even built in to many
motherboard chipsets.
As with the microprocessors, graphics and 3D accelerators come in wide varieties with different levels of performance and features. Each maker of graphics accelerators typically
829
830
has a full line of products, ranging from basic chips with moderate performance designed
for low-cost video boards to high-powered 3D products aimed at awing you with benchmark numbers far beyond the claims of their competitors (and often, reality).
The performance and output quality of a graphics accelerator depends on a number of
design variables. Among the most important of these are the width of the registers it uses
for processing video data, the amount and technology of the memory it uses, the ability of
the chip to support different levels of resolution and color, the speed rating of the chip,
the bandwidth of its connection to your computer and display, and the depth and extent of
its command set, as well as how well those commands get exploited by your software.
Graphics accelerators work like microprocessors dedicated to their singular purpose, and
internally they are built much the same. The same design choice that determines microprocessor power also affects the performance of graphics accelerator chips. The internal
register width of a graphics accelerator determines how many bits the chip works with at a
time. As with microprocessors, the wider the registers, the more data that can be manipulated in a single operation.
The basic data type for modern graphics operations is 32 bitsthats the requirement of
24-bit True Color with an alpha channel. Many accelerators at least double that and can
move pixels two (or four) at a time in blocks. Todays best are full 128-bit processors, able
to operate on multiple pixels at a time.
Because the graphics or 3D accelerator makes the video circuitry of your computer a separate, isolated system, concerns about data and bus widths elsewhere in your computer are
immaterial. The wide registers in graphic accelerators work equally well regardless of
whether you run 16-bit software (Windows 95 through Windows Me) or 32-bit software
(Windows NT, Windows 2000, and Windows XP), no matter what microprocessor you
have or what bus you plug your video board into.
The design of a graphics accelerator also sets the maximum amount of memory that can
be used in the frame buffer, which in turn sets upper limits on the color and resolution
support of a graphics accelerator. Other video board circuit choices may further constrain
these capabilities. In general, however, the more memory, the higher the resolution and
the greater the depth of color the accelerator can manage.
The same graphics or 3D accelerator chip may deliver wildly different performance when
installed in different video boards, even boards with substantially similar circuitry. One of
the chief reasons for performance differences among different brands of video board is not
in the hardware but the software support. Drivers can radically alter the performance of a
given accelerator chip. After all, the chip only processes instructions. If the instructions
are optimized, the performance of the chip will be optimum. To be useful at all, a video
board must have drivers to match the operating system you want to use.
Display Adapters
Video Controllers
The primary job of the video controller in desktop computers that use picture tubes is to
serialize the data in display memory. The conversion often is as convoluted as a video
game maze. The resemblance between the memory map and the onscreen image is only
metaphoric. The rows and columns by which the frame buffer is organized have no relationship to the rows and columns of pixels on your monitor screen. The bytes of video
information are scattered between a handful of memory chips or modules, sliced into several logical pages, and liberally dosed with added-in features such as cursors and sprites.
Somehow, all the scattered bytes of data must get organized and find their way to the
monitor. In addition, the monitor itself must be brought under the control of the computer, synchronized in two dimensions.
The video controller generates the actual scanning signals. Using the regular oscillations of
a crystal, the controller generates a dot clock, a frequency corresponding to the rate at which
it will scan the data for the pixels to appear on the screen. The controller divides down this
basic operating frequency to produce the horizontal synchronizing frequency, and from
that, the vertical synchronizing frequency. From these frequencies, the controller can create a monitor signal that lacks only image data. In real time, the controller scans through
the memory addresses assigned to each pixel on the screen in the exact order in which each
pixel will appear on the screen. The time at which each address gets read exactly matches
the time and position the pixel data appears in the final video output signal.
In modern analog computer video systems, the controller doesnt read memory directly.
Rather, the digital data scanned by the controller from memory gets routed first through
the RAMDAC, which converts the data from digital to analog form. The video controller
then adds the analog data to the scanning signals to create the video signal that gets
passed along to your monitor.
The video controller may draw pixel data from places other than the frame buffer. The
circuits used by computers generate the cursor that appears on the screen in text modes.
Or it may add in the bit-values associated with a sprite. By rerouting its scan, it can make
hardware windows.
In the language of computer engineering, the part of the video circuitry that performs the
actual scanning operation is called the CRT controller, because the signals it generates actually control the sweep of the electron beam in the CRT or picture tube of your monitor.
In the first computers, the CRT controller was a separate integrated circuit, the 6845
made by Motorola. This chip originally was not designed for computers but as a generalized scan-maker for any sort of electronic device that might plug in to a television set or
monitor. The engineers who designed the first computers chose it because it was a readily
available, off-the-shelf product that made the development of video boards relatively easy
and cheap. Long ago most video board manufacturers switched to custom-designed and
831
832
manufactured CRT controllers, often part of other computer video circuitry. Even the
most advanced of these chips emulate the 6845 in their basic operating modes. When
software calls upon the video system to operate in one of the original computers modes to
display text or low-resolution graphics, all CRT controllers react the same way to basic
hardware instructions.
RAMDACs
Modern computer monitors use analog signals so that the signals supplied them do not
limit the range of color they can display. The data stored in your computers frame buffer
is digital becausewell, everything in your computer is digital. Moreover, no convenient
form of analog memory is available. As a result of this divergence of signal typesdigital
in and analog outyour video board must convert the digital data into analog form compatible with your monitor. The chip that performs this magic is termed a digital-to-analog
converter. Sometimes it may be referred to as a RAMDACRAM for random access
memorybecause its digital data originates in memory.
RAMDACs are classified by the number of digital bits in the digital code they translate.
The number of bits translates into the number of signal levels that can appear in its output signal. For example, an eight-bit RAMDAC converts the levels encoded in eight-bit
digital patterns into 256 analog levels. In a monochrome system, each one of those levels
represents a shade of gray.
In color systems, each primary color or channel requires a separate DAC, a total of three.
Video RAMDACs usually put all three converter channels into a single package, although
some older video boards may use separate DAC chips for each color channel. Total up the
number of bits across all three channels of each RAMDAC, and youll get the number of
bit-planes of color a system can displayits palette. Most of the RAMDACs in todays
video systems have three eight-bit channels, allowing them to generate the 16.7 million
hues of True Color.
RAMDACs are also speed rated. The RAMDAC chip must be fast enough to process each
pixel that is to be displayed on the screen. The higher the resolution of the image you
want to display, the higher the speed required from your RAMDAC. The required speed
corresponds directly to the dot-clock (the number of pixels on the screen times the refresh
rate). To accommodate high-resolution displays, some RAMDACs are rated 200MHz and
higher. They dont have to enter the gigahertz stratosphere inhabited by microprocessors,
because such speeds are well beyond the needs of any practical video resolution.
Memory
On a video board, memory mostly means frame buffer. Every video board includes a good
dose of some kind of RAM for holding the bitmap of the image that appears on the
Displays
screen. In addition, 3D accelerators need memory for their special operations. Doublebuffering, as the name implies, doubles the memory needs by putting two separate frame
buffers to work. Z-buffering and working memory for the calculations of the 3D accelerator also increase the memory needs of the video board.
The requirements of the graphics or 3D accelerator determine the type of memory
required. The manufacturer of the video board sets the amount actually included on the
board. Some manufacturers provide sockets to allow you to later upgrade to increase the
resolution or color depth capabilities of your video system. As memory prices have fallen
to the point that sockets are an appreciable fraction of the cost of adding memory, manufacturers have resorted to providing separate board models with differing memory
dosages, all soldered down and not upgradable.
Because of the low prices of memory, most display adapters include substantially more
memory than even the largest frame buffers require. The memory on display adapters no
longer limits the resolution capabilities of the board. But more memory allows the display
adapter to implement more speed-enhancing technologies. In other words, a display
adapter with more memory generally will be faster.
Displays
You cannot see data. The information that your computer processes is nothing but ideas,
and ideas are intangibleno matter whether in your mind or your computers. Whereas
you can visualize your own ideas, you cannot peer directly into the pulsing digital thought
patterns of your computer. You probably have no right to think that you couldif you
cant read another persons thoughts, you should hardly expect to read the distinctly nonhuman circuit surges of your computer.
The display is your computers line of communication to you, much as the keyboard
enables you to communicate with it. Like even the best of friends, the display doesnt tell
you everything, but it does give you a clear picture, one from which you can draw your
own conclusions about what the computer is doing.
Background
Although the terms are often used interchangeably, a display and a monitor are distinctly
different. A display is the image-producing device itself, the screen that you see. The monitor is a complete box that adds support circuitry to the display. This circuitry converts the
signals set by the computer (or some other device, such as a videocassette recorder) into
the proper form for the display to use. Although most monitors operate under principles
like those of the television set, displays can be made from a variety of technologies,
including liquid crystals and the photon-glow of some noble gases.
833
834
Because of their similar technological foundations, monitors to a great extent resemble the
humble old television set. Just as a monitor is a display enhanced with extra circuitry, the
television is a monitor with even more signal-conversion electronics. The television incorporates into its design a tuner or demodulator that converts signals broadcast by television
stations or a cable television company into about the same form as those signals used by
monitors. Beyond the tuner, the television and monitor work in much the same way.
Indeed, some old-fashioned computer monitors work as televisions as long as they are
supplied the proper signals.
New monitors have developed far beyond their television roots, however. They have
greater sharpness and purity of color. To achieve these ends, they operate at higher frequencies than television stations can broadcast.
Computer displays and monitors use a variety of technologies to create visible images. A
basic bifurcation once divided the displays of desktop computers and those of laptop
machines. All notebook computers and a growing number of desktop machines use the
new technology, the flat-panel display. Most use large liquid crystal displays (LCDs). The
remaining desktop computers use monitors based on cathode ray tube (CRT) technology.
The old-technology monitors earn their desktop home with low prices, higher resolution,
and bigger screens than their LCD counterparts.
The glaringly apparent difference between thin LCD panels and fat cathode ray tubes
hides far greater internal dissimilarities. About the only characteristic the two technologies
share is that they make images. CRTs and LCDs differ not only in how they form images
but also how they are viewed and how they are best connected to your computer.
Cathode Ray Technology

The oldest electronic image-generating system still in use is the cathode ray tube. The
name is purely descriptive. The device is based on a special form of vacuum tubea glass
bottle that is partially evacuated and filled with an inert gas at very low pressure. The tube
of the CRT is hardly a tube but more flask-shaped, with a thin neck that broadens like a
funnel into a wide, nearly flat face. Although CRTs appear to be made like simple bottlesin fact, people in the monitor business sometimes refer to CRTs as bottlestheir
construction is surprisingly complex and involves a variety of glasses of many thicknesses.
The face of the typical CRT, for example, often is about an inch thick.
The cathode in the CRT name is a scientific term for a negatively charged electrode. In a
CRT, a specially designed cathode shoots a beam or ray of electrons toward a positively
charged electrode, the anode. (Electrons, having a negative charge, are naturally attracted
to positive potentials.) Because it works like a howitzer for electrons, the cathode of a
CRT is often called an electron gun.
Displays
The electrons race on their way at a substantial fraction of the speed of light, driven by
the high-voltage potential difference between the cathode and anode, sometimes as much
as 25,000 volts.
At the end of their flight to the anode, the electrons crash into a layer of a coating made
from phosphor compounds that has the wondrous property of converting the kinetic
energy of the electrons into visible light.
Physical Characteristics
The CRT is a physical entity that you can hold in your hand, drop on the floor, and watch
shatter. Little of its design is by chancenearly every design choice in making the CRT
has an effect on the image you see.
Four elements of the CRT exert the greatest influence on the kind and quality of image
made by a monitor. The phosphors chosen for the tube affect the color and persistence of
the display. The electron guns actually paint the image, and how well they work is a major
factor in determining image sharpness. In color CRTs, the shadow mask or aperture grille
limits the ultimate resolution of the screen. The face of the screen and the glare reflected
from it affect both image contrast and how happy you will be in working with a monitor.
Phosphors
At the end of the electrons short flight, from the gun in the neck of a CRT to the inside
of its wide, flat face, lies a layer of a phosphor-based compound with a wonderful propertyit glows when struck by an electron beam. The image you see in a CRT is the glow
of the electrically stimulated phosphor compounds, simply termed phosphors in the industry. Not all the phosphorous compounds used in CRTs are the same. Different compounds and mixtures glow in various colors and for various lengths of time after being
struck by the electron beam.
The type of phosphor determines the color of the image on the screen. Several varieties of
amber, green, and whitish phosphors are commonly used in monochrome displays. Color
CRT displays use three different phosphors painted in fine patterns across the inner surface of the tube. The patterns are made from dots or stripes of the three additive primary
colorsred, green, and bluearrayed next to one another. A group of three dots is called
a color triad or color triplet.
One triad of dots makes up whats called a picture element, often abbreviated as pixel
(although some manufacturers prefer to shorten picture element to pel).
The makers of color monitors individually can choose each of the three colors used in
forming the color triads on the screen. Most monitor-makers have adopted the same
phosphor family, which is called P22 (or B22, which is the same thing with a different
nomenclature), so the basic color capabilities of most multihued monitors are the same.
835
836
The color monitor screen can be illuminated in any of its three primary colors by individually hitting the phosphor dots associated with that color with the electron beam. Other
colors can be made by illuminating combinations of the primary colors. By varying the
intensity of each primary color, the tube can display a nearly infinite spectrum.
Monochrome displays have their CRTs evenly coated with a single, homogenous phosphor so that wherever the electron beam strikes, the tube glows in the same color. The
color of the phosphors determines the overall color that the screen glows.
Just as important to screen readability as the phosphor colors is the background color of
the display tube. The background area on a color screenthat is, the space in between the
phosphor dotsis called the matrix. It is never illuminated by the electron beam. The
color of the matrix determines the color screens background color (usually termed matrix
color)what it looks like when the power is off, either pale gray, dark green-gray, or nearly
black. Darker and black matrixes give an impression of higher contrast to the displayed
images. Lighter-gray matrixes make for purer whites. The distinctions are subtle, however,
and unless you put two tubes side-by-side, youre unlikely to appreciate the difference.
Color Temperature
If your work involves critical color matching, the color temperature of your monitor can
be an important issue. White light is not white, of course, but a mixture of all colors. Alas,
all whites are not the same. Some are richer in blue, some in yellow. The different colors
of white are described in their color temperature, the number of kelvins (degrees Celsius
above absolute zero) that a perfect luminescent body would need to be to emit that color.
Like the incandescence of a hot iron horseshoe in the blacksmiths forge, as its temperature gets higher the hue of a glowing object shifts from red to orange to yellow and on to
blue-white. Color temperature simply assigns an absolute temperature rating to these
colors. Figure 24.1 shows the effect of color temperature.
FIGURE 24.1
The color temperatures associated with
various conditions.
Kelvins
Conditions
Overcast skylight
New phosphor standard
Old phosphor standard
Average daylight
Direct sunlight
Halogen lamps
Incandescent lamps
Candlelight
For example, ordinary light bulbs range from 2,700 to 3,400 kelvins. Most fluorescent lights
have noncontinuous color spectra rich in certain hues (notably green) while lacking others,
which makes assigning a True Color temperature impossible. Other fluorescent lamps are
designed to approximate daylight with color temperatures of about 5,000 kelvins.
Displays
Monitors are designed to glow with the approximate color temperature of daylight rather
than incandescent or fluorescent light. Unfortunately, not everyone has the same definition of daylight. Noonday sun, for instance, ranges from 5,500 to 6,000 kelvins. Overcast
days may achieve a color temperature of 10,000 kelvins because the scattered blue glow of
the sky (higher color temperature) dominates the yellowish radiation from the sun.
The colors and blend of the phosphors used to make the picture tube screen and the relative strengths of the electron beams illuminating those phosphors determine the color temperature of a monitor. Some engineers believe the perfect day is a soggy, overcast afternoon
suited only to ducks and Englishmen and opt to run their monitors with a color temperatures as high as 10,000 kelvins. Others, however, live in a Kodachrome world where the
color temperature is the same 5,300 kelvins as a spring day with tulips in the park.
Many monitors now add color temperature to their setup menus, so you can choose
whether you want 5000-kelvin or 8500-kelvin whiteor something else entirely. Unless
you have a need to make precise color matches, the temperature setting you use is only a
matter of personal preference.
Electron Guns
To generate the beams that light the phosphors on the screen, a CRT uses one or more
electron guns. An electron gun is an electron emitter (a cathode) in an assembly that draws
the electrons into a sharp, high-speed beam. To move the beam across the breadth of the
tube face (so that the beam doesnt light just a tiny dot in the center of the screen), a group
of powerful electromagnets arranged around the tube, called the yoke, bend the electron
beam in the course of its flight. The magnetic field set up by the yoke is carefully controlled and causes the beam to sweep each individual display line down the face of the tube.
Monochrome CRTs have a single electron gun that continuously sweeps across the screen.
Most color tubes have three guns, although some color televisions and monitors boast
one-gun tubes, which more correctly might be called three guns in one.
The gun count depends on the definition of a gun. Like all color CRTs, the one-gun tubes
have three distinct electron-emitting cathodes that can be individually controlled. The
three cathodes are fabricated into a single assembly that allows them to be controlled as if
they were generating only a single beam.
In a three-gun tube, the trio of guns is arranged in a triangle. So-called one-gun tubes
arrange their cathodes in a straight line, often earning the epithet inline guns. In theory,
inline guns should be easier to set up, but as a practical matter, excellent performance can
be derived from either arrangement.
The three guns in a color CRT emit their electrons simultaneously, and the three resulting beams are steered together by force of a group of electromagnets around the neck of
the tube called the yoke. Monitors provide individual adjustments, both mechanical and
837
838
electrical, for each of the three beams to ensure that each beam falls exactly on the same
triplet of color dots on the screen as the others. Because these controls help the three
beams converge on the same triad, they are called convergence controls. The process of
adjusting them is usually termed color alignment.
Convergence
The three electron beams inside any color monitor must converge on exactly the right
point on the screen to illuminate a single triad of phosphor dots. If a monitor is not
adjusted properlyor if it is not designed or made properlythe three beams cannot
converge properly to one point. Poor convergence results in images with rainbow-like
shadows and a loss of sharpness and detail. Individual text characters no longer appear
sharply defined but become two- or three-color blurs. Monochrome monitors are inherently free from such convergence problems because they have but one electron beam.
Figure 24.2 shows the difference between good and poor convergence.
FIGURE 24.2
Excellent (left) and
poor convergence on
a monitor screen.
Convergence problems are a symptom rather than a cause of monitor deficiencies.

Convergence problems arise not only from the design of the display but also from the
construction and setup of each individual monitor. It can vary widely from one display to
the next and may be aggravated by damage during shipping.
The result of convergence problems is most noticeable at the screen periphery because
thats where the electron beams are the most difficult to control. When bad, convergence
problems can be the primary limit on the sharpness of a given display, having a greater
negative effect than wide dot-pitch or low bandwidth (discussed later).
Many monitor-makers claim that their convergence is a given fraction of a millimeter at a
particular place on the screen. If a figure is given for more than one screen location, the
center of the screen invariably has a lower figuretighter, better convergencethan a
corner of the screen.
The number given is how far one color may spread from another at that location. Lower
numbers are better. Typical monitors may claim convergence of about 0.5 (one-half)
millimeter at one of the corners of the screen. That figure often rises 50 percent higher
than the dot-pitch of the tube, making the convergence the limit on sharpness for that
particular monitor.
Misconvergence problems often can be corrected by adjustment of the monitor, usually
using the monitors internal convergence controls. A few, high-resolution (and high-cost)
Displays
monitors even have external convergence adjustments. But adjusting monitor convergence
is a job for the specialistand that means getting a monitor converged can be expensive,
as is any computer service call.
Many monitor-makers now claim that their products are converged for life. Although this
strategy should eliminate the need to adjust them, it also makes it mandatory to test your
display before you buy it. You dont want a display thats been badly converged for life.
Purity
The ability of a monitor to show you an evenly lit screen that does not vary in color
across its width is termed purity. A monitor with good purity will be able to display a pure
white screen without a hint of color appearing. A monitor with poor purity will be tinged
with one color or another in large patches. Figure 24.3 illustrates the screens with good
and poor purity.
FIGURE 24.3
Comparison of good
and bad monitor
purity.
Excellent purity
Poor purity
Poor purity often results from the shadow mask or aperture grille of a cathode ray tube
becoming magnetized. Degaussing the screen usually cures the problem. Most larger
monitors have built-in automatic degaussers.
You can degauss your monitor with a degaussing loop designed for color televisions or
even a bulk tape eraser. Energize the degaussing coil or tape eraser in close proximity to
the screen, then gradually remove the coil to a distance of three or more feet away before
switching it off. The gradually declining alternating magnetic field will overpower the static field on the mask, and the gradual removal of the alternating field prevents the strong
field from establishing itself on the mask.
Shadow Masks
Just pointing the electron beams at the right dots is not enough, because part of the beam
can spill over and hit the other dots in the triplet. The result of this spillover is a loss of
color puritybright hues become muddied. To prevent this effect and make images as
sharp and colorful as possible, most color CRTs used in computer displays and televisions
alike have a shadow maska metal sheet with fine perforations in itlocated inside the
display tube and a small distance behind the phosphor coating of the screen.
The shadow mask and the phosphor dot coating on the CRT screen are critically arranged
so that the electron beam can only hit phosphor dots of one color. The other two colors
of dots are in the shadow of the mask and cannot be seen by the electron beam.
839
840
The spacing of the holes in the shadow mask to a great degree determines the quality of
the displayed image. For the geometry of the system to work, the phosphor dots on the
CRT screen must be spaced at the same distance as the holes in the mask. Because the
hole spacing determines the dot spacing, it is often termed the dot-pitch of the CRT.
The dot-pitch of a CRT is simply a measurement of the distance between dots of the
same color. It is an absolute measurement, independent of the size of the tube or the size
of the displayed image.
The shadow mask affects the brightness of a monitors image in two ways. The size of the
holes in the mask limits the size of the electron beam getting through to the phosphors.
Off-axis from the gunsthat is, toward the corners of the screenthe round holes appear
oval to the gun and less of the beam can get through. As a result, the corners of a shadow
mask screen are often dimmer than the center, although the brightness difference may not
be distinguishable.
The mask also limits how high the electron beam intensity can be in a given CRT. A
stronger beamwhich makes a brighter imageholds more energy. When the beam
strikes the mask, part of that energy is absorbed by the mask and becomes heat, which
raises the temperature of the mask. In turn, this temperature rise makes the mask expand
unpredictably, distorting it minutely and blurring the image. To minimize this heatinduced blur, monitor-makers are moving toward making shadow masks from materials
that have a low coefficient of thermal expansion. That is, they change size as little as possible with temperature. The alloy Invar is favored for shadow masks because of its capability to maintain a nearly constant size as it warms.
Aperture Grilles
With all the problems associated with shadow masks, you might expect someone to come
up with a better idea. Sony Corporation did exactly that, inventing the Trinitron picture
tube.
The Trinitron uses an aperture grilleslots between a vertical array of wiresinstead of a
mask. The phosphors are painted on the inner face of the tube as interleaved stripes of the
three additive primary colors. The grille blocks the electron beam from the wrong stripes,
just as a shadow-mask blocks it from the wrong dots. The distance between two sequential
stripes of the same color is governed by the spacing of the slots between the wiresthe
slot-pitch of the tube. Because the electron beam fans out as it travels away from the electron gun and because stripes are farther from the gun than is the mask, the stripes are
spaced a bit farther apart than the slot-pitch. Their spacing is termed screen-pitch. For
example, a 0.25-millimeter slot-pitch Trinitron might have a screen-pitch of 0.26 millimeters. Figure 24.4 shows how dot-pitch and slot-pitch are measured.
Displays
FIGURE 24.4
Dot-pitch
Slot-pitch
Measuring dot-pitch
and slot-pitch.
Shadow mask
Aperture grille
The wires of the aperture grille are quite thick, about two-thirds the width of the slotpitch. For example, in a Trinitron with a 0.25-millimeter slot-pitch, the grille wires measure about 0.18 millimeters in diameter because each electron beam is supposed to
illuminate only one-third of the screen. The wires shadow the other two-thirds from the
beam to maintain the purity of the color.
The aperture grille wires are held taut, but they can vibrate. Consequently, Trinitron
monitors have one or two thin tensioning wires running horizontally across the screen.
Although quite fine, these wires cast a shadow on the screen that is most apparent on
light-colored screen backgrounds. Some people find the tensioning wire shadows objectionable, so you should look closely at a Trinitron before buying.
Trinitrons hold a theoretical brightness advantage over shadow-mask tubes. Because the
slots allow more electrons to pass through to the screen than do the tiny holes of a
shadow mask, a Trinitron can (in theory) create a brighter image. This added brightness is
not borne out in practice. However, Trinitrons do excel in keeping their screens uniformly
bright. The aperture grille wires of a Trinitron block the beam only in one dimension,
and they dont impinge as much on the electron beam at the screen edges.
Required Dot-Pitch
Regardless of whether a monitor uses a shadow mask with a dot-pitch or an aperture grille
with a slot-pitch, the spacing of image triads on the screen is an important constituent in
monitor quality. A monitor simply cannot put dots any closer together than the holes in
the mask or grille allow. Its easy to compute the pitch necessary for a resolution level in a
computer system. Just divide the screen size by the number of dots required to be displayed.
For example, a typical small display measures 14 inches diagonally. A horizontal line
stretching across the full width of such a screen would measure 11.2 inches or about 285
millimeters. To properly display an SVGA image (800 pixels by 600 pixels) would require
a dot-pitch of 0.36 millimeters or smaller (that is, 285/800 of a millimeter or less). Often a
monitors image is somewhat smaller than full screen width and such displays require an
even finer dot-pitch. The larger the display, the coarser the dot-pitch can be for a given
level of resolution.
841
842
Line Width
Another factor limits the sharpness of monitor images: the width of the lines drawn on the
screen. Ideally, any vertical or horizontal line on the screen will appear exactly one pixel
wide, but in practical monitors, the width of a line may not be so compliant. If lines are
narrower than one pixel wide, thin black lines will separate adjacent white lines and wide
white areas will be thinly striped in black. If the line width exceeds the size of a pixel, the
displays ability to render fine detail will be lost.
The ideal line width for a monitor varies with the size of the screen and the resolution
displayed on the screen. As resolution increases, lines must be narrower. As screen size
goes up (with the resolution constant), line width must increase commensurately. You can
calculate the required line width the same way as for the preceding dot-pitch example.
The line width should equal the maximum dot-pitch you calculate. In other words, for the
preceding 14-inch screen at SVGA resolution, youd want a line width of 0.36 millimeters.
Several factors influence the width of lines on the screen. The monitor must be able to
focus its electron beam into a line of ideal width. However, width also varies with the
brightness of the beambrighter beams naturally tend to expand out of focus.
Consequently, when you increase the brightness of your monitor, the sharpness of the
image may decline. For this reason, test laboratories usually make monitor measurements
at a standardized brightness level.
Screen Curvature
Most CRTs have a distinctive shape. At one end, a narrow neck contains the electron gun
or guns. Around the neck fits the deflection yoke, an external assembly that generates the
magnetic fields that bend the electron beams to sweep across the inner surface of the wide
face of the tube. The tube emerges from the yoke as a funnel-like flaring, which enlarges
to the rectangular face of the screen itself. This face often (but becoming much less common) is a spherically curving surface.
The spherical curve of the face makes sense for a couple of reasons. It makes the distance
traveled by the electron beam more consistent at various points on the screen, edge to
center to edge. A truly flat screen would require the beam to travel farther at the edges
than at the center and would require the beam to strike the face of the screen obliquely,
resulting in image distortion. Although this distortion can be compensated for electrically,
the curving screen helps things along.
In addition, the CRT is partly evacuated, so normal atmospheric pressure is constantly
trying to crush the tube. The spherical surface helps distribute this potentially destructive
force more evenly, making the tube stronger.
Screen curvature has a negative side effect. Straight lines on the screen appear straight
only from one observation point. Move your head closer, farther away, or to one side, and
the supposedly straight lines of your graphic images will bow this way and that.
Displays
Technology has made the reasons underlying spherical curved screens less than compelling. The geometry of inline guns simplifies tube construction and alignment sufficiently that cylindrically curved screens are feasible. They have fewer curvilinear problems
because they warp only one axis of the image. Trinitrons characteristically have faces with
cylindrical curves. Most shadow-mask tubes have spherical faces.
In the last few years, the technical obstacles to making genuinely flat screens have been
surmounted. Besides the easier geometry offered by true flat-screen CRTs, they bring
another benefit. The curved screen on conventional tubes reflects light from nearly the
entire area in front of the tube. No matter where you point the display, its sure to pick up
some glare. True flat-screen tubes reflect glare over a very narrow angle. If you point a
flat-screen display slightly downward, most of the room glare will reflect down onto your
keyboard and away from your eyes.
Resolution Versus Addressability
The resolution of a video system refers to the fineness of detail that it can display. It is a
direct consequence of the number of individual dots that make up the screen image and is
therefore a function of both the screen size and the dot-pitch.
Because the size and number of dots limit the image quality, the apparent sharpness of
screen images can be described by the number of dots that can be displayed horizontally
and vertically across the screen. For example, the resolution required by the Video
Graphics Array system in its standard graphics mode is 640 dots horizontally by 480 vertically. Modern display systems may produce images with as many as 16001200 dots in
their highest resolution mode.
Sometimes, however, the resolution available on the screen and that made by a computers
display adapter are not the same. For example, a video mode designed for the resolution
capabilities of a color television set hardly taps the quality available from a computer monitor. On the other hand, the computer-generated graphics may be designed for a display
system thats sharper than the one being used. You might, for instance, try to use a television in lieu of a more expensive monitor. The sharpness you actually see would then be
less than what the resolution of the video system would have you believe.
Resolution is a physical quality of the video display systemthe monitorthats actually
being used. It sets the ultimate upper limit on the display quality. In color systems, the
chief limit on resolution is purely physicalthe convergence of the system and the dotpitch of the tube. In monochrome systems, which have no quality-limiting shadow masks,
the resolution is limited by the bandwidth of the monitor, the highest frequency signal
with which it can deal. (Finer details pack more information into the signals sent from
computer system to monitor. The more information in a given time, the higher the frequency of the signal.)
843
844
Addressability is essentially a bandwidth measurement for color monitors. It indicates how

many different dots on the screen the monitor can point its electron guns at. It ignores,
however, the physical limit imposed by the shadow mask. In other words, addressability
describes the highest quality signals the monitor can handle, but the full quality of those
signals is not necessarily visible to you onscreen.
Anti-Glare Treatment
Most mirrors are made from glass, and glass tries to mimic the mirror whenever it can.
Because of the difference between the index of refraction of air and that of glass, glass is
naturally reflective. If you make mirrors, thats great. If you make monitorsor worse yet,
use themthe reflectivity of glass can be a big headache. A reflection of a room light or
window from the glass face of the CRT can easily be brighter than the glow of phosphors
inside. As a result, the text or graphics on the display tend to wash out or be obscured
by the brightness.
The greater the curvature of a monitor screen, the more apt it is to have a problem with
reflections. You cant change the curve of your monitors face, but anti-glare treatments
can reduce or eliminate reflections from the monitor face. Several glare-reduction technologies are available, and each varies somewhat in its effectiveness.
Mesh
The lowest tech and least expensive anti-glare treatment is simply a fabric mesh, usually
nylon. The mesh can either be placed directly atop the face of the screen or in a removable frame that fits about half an inch in front of the screen. Each hole in the mesh acts
like a short tube, allowing you to see straight in at the tube, but cutting off light from the
sides of the tube. Your straight-on vision gets through unimpeded, while glare that angles
in doesnt make it to the screen.
As simple as this technique is, it works amazingly well. The least expensive after-market
anti-glare system uses mesh suspiciously similar to pantyhose stretched across a frame.
Unfortunately, this mesh has an unwanted side effect. Besides blocking the glare, it also
blocks some of the light from the screen and makes the image appear darker. You may
have to turn the brightness control up to compensate, which may make the image bloom
and lose sharpness.
Mechanical
Glare can be reduced by mechanical meansnot by a machine that automatically intercepts glare before it reaches the screen, but by mechanical preparation of the screen surface. By lightly grinding the glass on the front of the CRT, the face of the screen can be
made to scatter rather than reflect light. Each rough spot on the screen that results from
the mechanical grinding process reflects light randomly, sending it every which direction.
A smooth screen reflects a patch of light all together, like a mirror, reflecting any bright
light source into your eyes. Because the light scattered by the ground glass is dispersed,
Displays
less of it reaches your eyes and the glare is not as bright. However, because the coarse
screen surface disperses the light coming from inside the tube as well as that reflected
from the tube face, it also lessens the sharpness of the image. The mechanical treatment
makes text appear slightly fuzzier and out of focus, which to some manufacturers is a
worse problem than glare.
Coating
Engineers can reduce the glare reflected from a CRT screen by applying a coating with
dispersive or interference properties to the screen. A dispersive coating forms a rough film
on the face of the CRT. This rough surface acts in the same way as a ground-glass screen
would, scattering light that would otherwise reflect back as glare. An optical interference
coating cancels out reflections.
Interference coatings work by reflecting light between multiple thin layers of a compound, such as magnesium fluoride. By precisely controlling the thickness of this coating,
light can be made to reflect in such a way its waves cancel out. The fluoride coating is
made to be a quarter the wavelength of light (usually of light at the middle of the spectrum). Light going through the fluoride and reflecting from the screen thus emerges from
the coating out of phase with the light striking the fluoride surface, visually canceling out
the glare. Multiple layers are tuned to the different frequencies of the colors of visible
light. Camera lenses are coated to achieve exactly the same purpose, the elimination of
reflections. A proper interference coating can minimize glare without affecting image
sharpness or brightness.
Polarization
Light can be polarizedthat is, its photons can be restricted to a single plane of oscillation. A polarizing filter allows light of only one polarization to pass. Two polarizing filters
in a row can be arranged to allow light of only one plane of polarization to pass (by making the planes of polarization of the filters parallel), or the two filters can stop light
entirely when their planes of polarization are perpendicular.
The first filter lets only one kind of light pass; the second filter lets only another kind of light
pass. Because none of the second kind of light reaches the second filter, no light gets by.
When light is reflected from a surface, its polarization is shifted by 90 degrees. This physical principle makes polarizing filters excellent reducers of glare.
A sheet of polarizing material is merely placed a short space in front of a display screen.
Light from a potential source of glare goes through the screen and is polarized. When it
strikes the display and is reflected, its polarization is shifted 90 degrees. When it again
reaches the filter, it is out of phase with the filter and cannot get through. Light from the
display, however, only needs to go through the filter once. Although this glow is polarized,
there is no second screen to impede its flow to your eyes.
845
846
Every anti-glare treatment has its disadvantage. Mesh makes an otherwise sharp screen
look fuzzy, because smooth characters are broken up by the cell structure of the mesh.
Mechanical treatments are expensive and tend to make the screen appear to be slightly
fuzzy or out of focus. The same is true of coatings that rely on the dispersion principle.
Optical coatings, polarizing filters, and even mesh suffer from their own reflections. The
anti-glare material itself may add its own bit of glare. In addition, all anti-glare treatmentspolarizing filters in particulartend to make displays dimmer. The polarizing filter actually reduces the brightness of a display to one-quarter its untreated value.
Even with their shortcomings, however, anti-glare treatments are amazingly effective. They
can ease eyestrain and eliminate the headaches that come with extended computer use.
Deflection Angle
Another difference between CRTs is their deflection angle, which measures the maximum
degree the tube allows its electron beam to bend. This corresponds to the angle at which the
tube flares out. A tube with a narrow deflection angle will be long with a small screen. A
wide deflection angle permits large screens with shorter tubes. Ideally, the best tube would
be the shortest, but as the deflection angle increases, it becomes more difficult to control the
electron beam and make a perfect image. Consequently, lower deflection angles usually produce better images. Improved technology has allowed deflection angles to increase while
maintaining image quality. This allows monitors to be shorter and take up less desk space.
A monitor can deflect the electron beam inside its CRT both left and right of the centerline path the beam would take without deflection. By custom, the deflection angle listed
for a given CRT is the total deflection from maximum left to maximum right. Until the
late 1990s, nearly all computer monitors used CRTs with deflection angles of 90 degrees.
Television sets, which can get by with less image quality, typically use 110-degree tubes. In
1999, computer monitors with 100-degree tubes began to enter the market.
Image Characteristics
The electronics of a monitor control the size, shape, and other aspects of the image displayed on a monitors screen. The image qualities are defined and characterized in a number of ways. The most rudimentary is screen sizethe bigger your monitor screen, the
larger the images it can make. Because of the underscanning common among computer
monitors, however, the actual image size is almost always smaller than the screen. The
aspect ratio of the image describes its shape independent of its size. Most monitors give
you a variety of controls to alter the size and shape of the image, so you are the final
arbiter of what things look like on your monitor screen.
Screen Size
The most significant measurement of a CRT-based monitor is the size of its screen.
Although seemingly straightforward, screen size has been at best an ambiguous measurement and at worst downright misleading.
Displays
The confusion all started with television, where confusion often begins. The very first
television sets had round CRTs, and their size was easy to measuresimply the diameter
of the tube. When rectangular tubes became prevalent in the 1950s, the measurement
shifted to the diagonal of the face of the tube. The diagonal was, of course, the closest
equivalent to the diameter of an equivalent round tube. It was also the largest dimension
that a television manufacturer could reasonably quote.
Unlike television images, which usually cover the entire face of the CRT, computer monitors limit their images to somewhat less. The image is most difficult to control at the
edges of the screen, so monitor-makers maintain higher quality by restricting the size of
the image. They mask off the far edges of the CRT with the bezel of the monitor case.
That bezel means that no image can fill the entire screenat least no image that you can
entirely see. The tube size becomes irrelevant to a realistic appraisal of the image. Some
monitor-makers persisted in using it to describe their products. Fortunately most of the
industry recognized this measuring system as optimistic exaggeration and began using
more realistic diagonal measurement of the actual maximum displayable image area.
VESA adopted the diagonal of the maximum image area as the measurement standard in
its Video Image Area Definition standard, version 1.1, which it published on October 26,
1995. This standard requires that the screen image area be given as horizontal and vertical
measurements of the actual active image area when the monitor is set up by the manufacturer using the manufacturers test signals. The dimensions must be given in millimeters,
with an assumed maximum variance of error of plus or minus two percent. Wider tolerances are allowed but must be explicitly stated by the manufacturer. In no case can the
expressed image dimensions exceed the area visible through the monitor bezel.
Because the aspect ratio of computer monitor displays is 4:3, computation of the horizontal and vertical screen dimensions from the diagonal is easy. The diagonal represents the
hypotenuse of a 3-4-5 right triangle, and that ratio applies to all screen sizes. Table 24.2
lists the dimensions for the most common nominal screen sizes.
TA B L E 2 4 . 2
Nominal CRT Screen Dimensions
Diagonal
Horizontal
Millimeters
Vertical
Inches
Horizontal
Vertical
14 inches
284
213
11.2
8.4
15 inches
305
229
12
16 inches
325
244
12.8
9.6
17 inches
345
259
13.6
10.2
20 inches
406
305
16
12
21 inches
427
320
16.8
12.6
847
848
Portrait displays, which are designed to give you a view more like the printed sheets that
roll out of your laser printer and into envelopes, merely take an ordinary CRT and turn it
on its side. The market is not large enough to justify development of custom CRTs for
portrait applications. Moreover, the 4:3 aspect ratio works fine because the active image
on a sheet of letterheadthe space actually occupied by printing once you slice off the
top, bottom, left, and right marginsis about eight by ten inches, a nearly perfect fit on a
standard picture tube. When measuring the images on these portrait displays, horizontal
becomes vertical, and all measurements rotate 90 degrees.
Overscan and Underscan
Two monitors with the same-size screens may have entirely different onscreen image sizes.
Composite monitors are often afflicted by overscanthey attempt to generate images larger
than their screen size, and the edges and corners of the active display area may be cut off.
(The overscan is often designed in so that as the components inside the monitor age and
become weaker, the picture shrinks down to normal sizelikely over a period of years.)
Underscan is the opposite conditionthe image is smaller than nominal screen size. For a
given screen size, an overscanned image will appear larger at the expense of clipping off the
corners and edges of the image as well as increasing distortion at the periphery of the
image. Underscanning wastes some of the active area of the monitor screen. Figure 24.5
illustrates the effects of underscanning and overscanning on the same size screen.
FIGURE 24.5
Underscan and overscan compared.
Underscan is perfectly normal on computer displays and does not necessarily indicate any
underlying problems unless it is severefor example, when it leaves a two-inch black
band encircling the image. Underscanning helps keep quality high because image geometry is easier to control nearer to the center of the screen than it is at the edges. Pulling in
the reigns on the image can ensure that straight lines actually are displayed straight.
Moreover, if you extend the active image to the very edge of the bezel and you change
your viewing position so that you are not facing the screen straight on, the edge of the
image may get hidden behind the bezel. The glass in the face of the screen is thicker than
you might think, on the order of an inch (25 millimeters), enough that the third dimension will interfere with your most careful alignment.
On the other hand, whereas overscan gives you a larger image and is the common display
mode for video systems, it is not a good idea for computer monitor images. Vital parts of
the image may be lost behind the bezel. You may lose the first character or two from each
line of type of one edge of a drafting display to overscan. With video, however, people
Displays
prefer to see as big an image as possible and usually pay little attention to what goes on at
the periphery. Broadcasters, in fact, restrict the important part of the images they deal
with to a safe area that will be completely viewable, even on televisions with substantial
overscan.
Aspect Ratio
The relationship between the width and height of a monitor screen is termed its aspect
ratio. Today, the shape of the screen of nearly every monitor is standardized, as is that of
the underlying CRT that makes the image. The screen is 1.33 times wider than it is high,
resulting in the same 4:3 aspect ratio used in television and motion pictures before the
wide-screen phenomenon took over. Modern engineers now prefer to put the vertical
number first to produce aspect ratios that are less than one. Expressed in this way, video
has a 3:4 aspect ratio, a value of 0.75.
The choice of aspect ratio is arbitrary and a matter of aesthetics. According to classical
Greek aesthetics, the Golden Ratio with a value of about 0.618 is the most beautiful. The
exact value of the Golden Ratio is irrational, (SQRT(5)-1)/2. Its beauty is mathematical as
well as aesthetic, the solution to the neat little equation x+1 = 1/x. (Hardly coincidental,
expressed as a ratio of horizontal to vertical, the Golden Ratio becomes 1.618, the solution to x-1 = 1/x.)
Various display systems feature their own aspect ratios. The modern tendency is toward
wider aspect ratios. For example, High Definition Television (HDTV) stretches its aspect
ratio from the 3:4 of normal video and television to 9:16. The normal negatives you make
with your 35 mm camera have a 4:6 aspect ratio. (These numbers are often reversedfor
example, 16:9 for HDTVbut they refer to the same screen shape, which has a longer
horizontal axis than vertical axis.) The reason video is so nearly square carries over from
the early days of television when cathode ray tubes had circular faces. The squarer the
image, the more of the circular screen was put to use. Figure 24.6 compares the aspect
ratios of three common display systems.
FIGURE 24.6
Aspect ratios of
display systems.
The image on your monitor screen need not have the same aspect ratio as the tube, however. The electronics of monitors separate the circuitry that generates the horizontal and
vertical scanning signals and results in their independent control. As a result, the relationship between the two can be adjusted, and that adjustment results in an alteration of the
aspect ratio of the actual displayed image. For example, by increasing the amplification of
the horizontal signal, the width of the image is stretched, thus raising the aspect ratio.
849
850
Image Sizing
Normally you set up your monitor to display the best possible image at the resolution you
use most often. Most people leave their monitors locked at one resolution level.
Those who change resolution settings often run into a problemthe monitor produces
different size images at different resolution levels. Images with more dots are larger than
those with fewer.
Monitor-makers developed autosizing technology to compensate for changes in resolution.
Autosizing works by examining the video signal and automatically adjusting the monitor
to compensate for different line widths and screen heights. True autosizing works regardless of the signal going to the monitor and scales the image to match the number of display lines.
Image Distortion
Between the electron guns and the phosphors in a cathode ray tube, the electron beam
passes through an electronic lens and deflection system that focuses and steers the beam
to ensure that its path across the screen is the proper size and in the proper place. The
electronics of the monitor control the lens and deflection system, adjusting it throughout the sweep of the electron beam across the screen. In addition to other chores, the
electronics must compensate for the difference in the path of the electron beam at different screen positions. Modern monitors do a magnificent job of controlling beam
placement.
When the control system is not properly adjusted, however, the image may exhibit any of
a number of defects. Because these defects distort the image from its desired form, they
are collectively called image distortion.
The two most common forms of image distortion are barrel distortion and pincushion
distortion. Barrel distortion causes vertical or horizontal lines in the image to bow outward
so that the center of the lines lies closer to the nearest parallel edge of the screen.
Pincushion distortion causes the vertical or horizontal lines in the image to bow inward so
that the center of the lines is closer to the center of the screen. Figure 24.7 shows these
two kinds of image distortion.
FIGURE 24.7
Barrel and pincushion
distortion.
No distortion
Barrel
Pincushion
Displays
Barrel and pincushion distortion arise from the same causeimproper image compensationand are essentially opposites of one another. Overcompensate for pincushion distortion and you get barrel distortion. Collectively, the two are sometimes simply called
pincushioning, no matter which way the lines bow.
Pincushioning is always worse closer the edges of the image. All monitors have adjustments to compensate for pincushioning, although these adjustments are not always available to you. They may be hidden inside the monitor. Other monitors may include
pincushioning adjustments in their control menus. Technicians usually use test patterns
that display a regular grid on the screen to adjust monitors to minimize pincushioning.
You can usually use a full-screen image to adjust the pincushioning controls so that the
edges of the desktop background color are parallel with the bezel of your monitor.
Less common is trapezoidal distortion, which leaves lines at the outer edge of the screen
straight but not parallel to the bezel. In other words, instead of your desktop being a rectangle, it is a trapezoid with one side shorter than its opposite side. As with pincushioning,
all monitors have controls for trapezoidal distortion but not all make them available to
you as the user of the monitor. If your monitor does have an external control for trapezoidal distortion, you adjust it as you do for pincushioning.
Image Controls
A few monitors (far from a majority) make coping with underscan, overscan, and odd
aspect ratios simply a matter of twisting controls. These displays feature horizontal and
vertical size (or gain) controls that enable you to adjust the size and shape of the image to
suit your own tastes. With these controlsproviding they have adequate rangeyou can
make the active image touch the top, bottom, and sides of the screen bezel, or you can
shrink the bright area of your display to a tiny (but geometrically perfect) patch in the
center of your screen.
Size and position controls give you command of how much screen the image on your
monitor fills. With full-range controls, you can expand the image to fill the screen from
corner to corner or reduce it to a smaller size that minimizes the inevitable geometric distortion that occurs near the edges of the tube. A full complement of controls includes one
each of the following: horizontal position (sometimes termed phase), vertical position, horizontal size (sometimes called width), and vertical size (or height).
A wide control range is better than a narrow one. Some monitors skimp on one or more
controls and limit you in how large you can make the onscreen image. Worse, sometimes
a monitor-maker doesnt include a control at all. For example, some monitors have no
horizontal size controls. As a result, you cannot adjust both the size and aspect ratio of the
image.
851
852
The optimum position for these controls is on the front panel, where you can adjust them
and view the image at the same time. Controls on the rear panel require you to have gorillalike arms to reach around the monitor to make adjustments while checking their effect.
Image controls come in two types: analog and digital. Analog controls are the familiar
old knobs like you find on vintage television sets. Twist one way and the image gets bigger; twist the other and it shrinks. Analog controls have one virtuejust by looking at
the knob you know where they are set, whether at one or the other extreme of their
travel. The control itself is a simple memory system; it stays put until you move it again.
Analog controls, however, become dirty and wear out with age, and they usually enable
you to set just one value per knobone value that must cover all the monitors operating
modes.
Digital controls give you pushbutton control over image parameters. Press one button,
and the image gets larger or moves to the left. Another compensates in the opposite direction. Usually digital controls are linked with a microprocessor, memory, and mode-sensing
circuitry so that you can preset different image heights and widths for every video standard your monitor can display.
Digital controls dont get noisy with age and are more reliable and repeatable, but you
never know when you are approaching the limit of their travel. Most have two-speed
operationhold them in momentarily and they make minute changes; keep holding them
down and they shift gears to make gross changes. Of course, if you dont anticipate the
shift, youll overshoot the setting you want and spend a few extra moments zeroing in on
the exact setting.
Most CRT-based displays also carry over several controls from their television progenitors. Nearly every computer monitor has a brightness control, which adjusts the level of
the scanning electron beam; this in turn makes the onscreen image glow brighter or dimmer. The contrast control adjusts the linearity of the relationship between the incoming
signal and the onscreen image brightness. In other words, it controls the brightness relationship that results from different signal levelshow much brighter high-intensity is. In
a few displays, both the brightness and contrast functions are combined into a single
picture control. Although a godsend to those who might get confused by having to
twiddle two knobs, the combined control also limits your flexibility in adjusting the
image to best suit your liking.
Flat-Panel Display Technology

Throughout the age of television, predictions of the future have almost universally
included big, flat TV screens. Almost as soon as the first CRTs cast their blue spells in living rooms across the world, everyone knew they were only temporary technology.
Displays
Something better would soon come along. Of course, at the same time pundits predicted
an autogyro in every garage.
Although engineers never pursued the dream of the personal autogyro, they have
attacked the flat-screen challenge with verve and a variety of approaches. None of the
results quite takes television to reaches, along with that autogyro, that are still whispering siren-like from the land of to-be. CRTs are impractical for portable computers, as
anyone who has toted a forty-pound first-generation portable computer knows. The glass
in the tube itself weighs more than most of todays portable machines, and running a
CRT steals more power than most laptop or notebook machines budget for all their circuitry and peripherals.
LCD
The winner in the display technology competition was the liquid crystal display, the infamous LCD. Unlike LED and gas-plasma displays, which glow on their own emitting photons of visible light, LCDs dont waste energy by shining. Instead, they merely block light
otherwise available. To make patterns visible, they either selectively block reflected light
(reflective LCDs) or the light generated by a secondary source either behind the LCD
panel (backlit LCDs) or adjacent to it (edgelit LCDs). The backlight source is typically an
electroluminescent (EL) panel, although some laptops use cold-cathode fluorescent (CCF)
panels for brighter, whiter displays with the penalty of higher cost, greater thickness, and
increased complexity.
Nematic Technology
A number of different terms describe the technologies used in the LCD panels themselves, terms such as supertwist, double-supertwist, and triple-supertwist. In effect, the
twist of the crystals controls the contrast of the screen, so triple-supertwist screens are
contrastier than ordinary supertwist.
The history of laptop and notebook computer displays has been led by innovations in
LCD technology. Invented by RCA in the 1960s (General Electric still receives royalties
on RCAs basic patents), LCDs came into their own with laptop computers because of
their low power requirements, light weight, and ruggedness.
An LCD display is actually a sandwich made from two plastic sheets with a very special
liquid made from rod-shaped or nematic molecules. One important property of the
nematic molecules of liquid crystals is that they can be aligned by grooves in the plastic to
bend the polarity of light that passes through them. More importantly, the amount of
bend the molecules of the liquid crystal give to the light can be altered by applying an
electrical current through them.
Ordinary light has no particular orientation, so liquid crystals dont visibly alter it. But
polarized light aligns all the oscillations of its photons in a single direction. A polarizing
853
854
filter creates polarized light by allowing light of a particular polarity (or axis of oscillation)
to pass through. Polarization is key to the function of LCDs.
To make an LCD, light is first passed through one polarizing filter to polarize it. A second
polarizing filter, set to pass light at right angles to the polarity of the first, is put on the
other side of the liquid crystal. Normally, this second polarizing filter stops all light from
passing. However, the liquid crystal bends the polarity of light emerging from the first filter so that it lines up with the second filter. Pass a current through the liquid crystal and
the amount of bending changes, which alters in turn the amount of light passing through
the second polarizer.
To make an LCD display, you need only to selectively apply current to small areas of the
liquid crystal. The areas to which you apply current are dark; those that you dont are
light. A light behind the LCD makes the changes more visible.
Over the past few years, engineers have made several changes to this basic LCD design to
improve its contrast and color. The basic LCD design outlined thus far is technically
termed twisted nematic (TN) technology. The liquid molecules of the TN display in their
resting state always bend light by 90 degrees, exactly counteracting the relationship
between the two polarizing panels that make up the display.
By increasing the bending of light by the nematic molecules, the contrast between light
and dark can be increased. An LCD design that bends light by 180 to 270 degrees is
termed a supertwist nematic (or simply supertwist) display. One side effect of the added twist
is that the appearance of color artifacts results in the yellowish green and bright blue hues
of many familiar LCD displays.
This tinge of color can be canceled simply by mounting two supertwist liquid crystals
back-to-back so that one bends the light the opposite direction of the other. This design
is logically termed a double-supertwist nematic (or simply double-supertwist) display. This
was once used for laptop computers but is not currently used for several reasonsnot the
least of which is the lack of color. In addition, because two layers of LCD are between
you and the light source, double-supertwist panels appear darker or require brighter
backlights for adequate visibility. Thats why most notebook computers need backlights
for their screens.
Triple-supertwist nematic displays instead compensate for color-shifts in the supertwist
design by layering both sides of the liquid crystal with thin polymer films. Because the
films absorb less light than the twin panels of double-supertwist screens, less backlight
and less backlight poweris required for the same screen brightness.
Displays
Cholesteric Technology
In 1996, the Liquid Crystal Institute of Kent State University developed another liquid
crystal technology into a workable display system and began its commercial manufacture.
Termed cholesteric LCDs, this design uses crystals that switch between transmissive and
reflective states instead of twisting. These changes are more directly visible and require no
polarizers to operate. In that polarizing panels reduce the brightness of nematic displays
by as much as 75 percent, cholesteric LCDs can be brighter. Early screens are able to
achieve high contrast ratios without backlights.
Cholesteric screens have a second advantage. They are bi-stable. That is, maintaining a
given pixel in either the transmissive or reflective phase requires no energy input. Once
switched on, a pixel stays on until switched off. The screen requires power only to change
pixels. In fact, a cholesteric screen will retain its last image even after it is switched off.
Power usage in notebook computer applications are likely to be as low as 10 percent of
nematic panels.
Passive Matrix
Nematic LCDs also come in two styles, based on how the current that aligns their
nematic molecules is applied. Most LCD panels have a grid of horizontal and vertical conductors, and each pixel is located at the intersection of these conductors. The pixel is
darkened simply by sending current through the conductors to the liquid crystal. This
kind of display is called a passive matrix.
Active Matrix
The alternate design, the active matrix, is more commonly referred to as Thin Film
Transistor (TFT) technology. This style of LCD puts a transistor at every pixel. The transistor acts as a relay. A small current is sent to it through the horizontal and vertical grid,
and in response the transistor switches on a much higher current to activate the LCD pixel.
The advantage of the active matrix design is that a smaller current needs to traverse the
grid, so the pixel can be switched on and off faster. Whereas passive LCD screens may
update only about half a dozen times per second, TFT designs can operate at ordinary
monitor speedsten times faster. That increased speed equates to faster responsefor
example, your mouse wont disappear as you move it across the screen.
The disadvantage of the TFT design is that it requires the fabrication of one transistor for
each screen pixel. Putting those transistors there requires combining the LCD and semiconductor manufacturing processes. Thats sort of like getting bricklayers and carpenters
to work together. Even so, manufacturers have lowered production costs so that nearly all
LCDs in new computers and monitors use active matrix designs.
855
856
Double-Scanned Displays
To achieve the quality of active matrix displays without paying the price, engineers at one
time opted to up the scan on passive panels. Double-scanned passive displays work
exactly like the name says: They scan their screens twice in the period that a normal
screen is scanned only once. Rather than go over each pixel two times, a double-scanned
display divides its screen into two halves and scans the two halves at the same time. The
idea is something like the interlacing of CRT screens, lowering the required scanning
frequency, but the arrangement and effect are different. Double-scanned displays split
the screen in the middle into upper and lower halves. The split means that each pixel
gets twice as long for updates as would be the case if the whole screen were scanned at
the same frequency. As a result, double-scanning can eke out extra brightness, contrast,
and speed. These screens do not, however, reach the quality level set by active matrix
screens. Consequently, they have dropped from favor in step with the drop in activematrix pricing.
Sometimes manufacturers use the abbreviation DSTN to indicate double-scanned displays.
It is an acronym for double-scanned supertwist nematic. HPA, which stands for high performance addressing, is an improved double-scanned technology. Note that both DSTN and
HPA display systems use passive matrix technology.
Response Time
The LCD panel equivalent of persistence is response time. Charging and discharging individual pixels requires a finite period, and the response time measures this period. The
time to charge and the time to discharge a given pixel can be, and often are, different, and
theyre typically individually specified. For example, the off times of some active screens
may be twice that of the on time.
The ambient temperature can have dramatic effects on the response time of an LCD
panel. At freezing, the response time of a panel may be three times longer (slower) than at
room temperature.
At room temperature, an active matrix display pixel has a response time on the order of 10
to 50 milliseconds.
Bad Pixels
One of the major issues in making LCD panels is yield. That is, the number of good, fully
functional panels that can be made in each batch. Out of 100 screens manufactured, for
example, 30 may work properly, resulting in a yield of 30 percent. Early in production,
yield might not be any higher than that, although as the technology matures, yields have
been improving.
One way of improving yields is to lower the standard for what you call fully functional.
Some manufacturers believe that a screen with a few pixels that do not work properly still
Displays
is a usableand more importantly, salablescreen. These bad pixels may remain constantly lit or may never light. In any case, you get a screen with a few dots that persistently stay one color.
Constantly lit pixels of a single color appears to be the most common defect. Although
these one-color pixels stand out against an overall black screen, they disappear against the
default white Windows background.
Color Issues
The displayable range of colors of todays LCD panels is more limited than that of CRT
displays. Although the color shortfall recommends against using a desktop LCD panel in
color-critical applications such as prepublishing, the range is more than broad enough to
present realistic full-color images. In fact, unless you need to do Pantone matching, youre
unlikely to notice the lack of color range.
A related problem with LCD panels has been control of color temperature. Early LCD
panels have had almost random color temperaturesthey were set at whatever the commonly available backlights would permit. Modern displays have advanced to the point that
many now offer selectable color temperatures, typically a daylight value (in the range of
5000 kelvins), a bluish hot temperature (in the 9000-kelvin range), and an adjustable
user-defined temperature.
Transmissive Displays
Most of todays LCD panels are transmissivethat is, light is transmitted through the
entire sandwich that makes up the panel. The light has to come from somewhere, and that
source usually takes the form of an EL panel called a backlight. Although EL panels are
more efficient than incandescent lights, they still consume a substantial amount of power.
In notebooks computers, for example, the EL panel may require about one-third of the
entire energy budget of the computer. For this reason, most notebook computer manufacturers restrict the brightness of the display when a computer operates from batteries.
Dimming the screen can greatly increase battery life.
Most display manufacturers make the EL panel a separately replaceable element of their
computer monitors (although most displays require a service technician to replace the
backlight panel). These panels dim with age (at least early models do), making the overall
display dimmer. Such displays can often be rejuvenated by replacing the backlight.
Even the brightest backlights pale in comparison to daylight. Consequently, transmissive
displays do not work well outdoors. Although you can work with one in the shade, youll
end up raising the backlight brightness to a level that will quickly consume your battery
reserves.
857
858
Reflective Displays
A reflective display reflects ambient light from the crystals in the screen. The brighter the
ambient light, the brighter the screen. Reflective screens are at their best outdoors and are
almost impossible to read in dim light. Reflective screens are more energy efficient
because they have no need for a power-hungry backlight.
Reflective technology is mostly reserved for the small displays of toys and appliances and are
often monochrome. The technology does not work well for large color computer displays.
A half-bred technology called transflective combines both transmissive and reflective qualities. In bright light, the screen acts as a reflective display but allows a backlight to
takeover and make the screen operate in transmissive mode. Although still in the developmental stage, transflective screens hold promise for extending battery life for computers
used in outdoor environments because the backlight serves only part-time.
Some handheld computers and Personal Digital Assistants attempt to bridge the gap
between battery-saving reflective technology and power-draining transmissive technology
by sidelighting their small screens. These machines use a reflective screen augmented by a
light shining through its edge. Although the technology does not yield a bright screen, the
sidelit screens are readable and long battery life compensates for some of the squinting.
Electronics
The image you see onscreen is only part of the story of a complete display system. The
video signals from your computer must be amplified and processed by the electronics
inside the monitor to achieve the right strength and timing relationships to put the proper
image in view.
The basic electronic components inside a monitor are its video amplifiers. As the name
implies, these circuits simply increase the strength (amplify) of the approximately one-volt
signals they receive from your computer to the thousands of volts needed to drive the
electron beam from cathode to phosphor. Monochrome monitors have a single video
amplifier; color monitors, three (one for each primary color).
In an analog color monitor, these three amplifiers must be exactly matched and absolutely
linear. The input and output of each amplifier must be precisely proportional, and they
must be the same as the other two amplifiers. The relationship between these amplifiers is
called color tracking. If it varies, the color of the image on the screen wont be what your
software had in mind.
The effects of poor color tracking are all bad. You lose precision in your color control. This
is especially important for desktop publishing and presentation applications. With poor
color tracking, the screen can no longer hope to be an exact preview of what eventually
appears on paper or film. You may even lose a good fraction of the colors displayable by
your video system.
Displays
What happens is that differences between the amplifiers cause one of the three primary
colors to be emphasized at times and de-emphasized at others, casting a subtle shade on
the onscreen image. This shading effect is most pronounced in gray displaysthe dominate color or colors tinge the gray.
Although you dont have to worry about color tracking in a monochrome display, the
quality of the amplifier nevertheless determines the range of grays that can be displayed.
Aberrations in the amplifier cause the monitor to lose some of its grayscale range.
The relationship between the input and output signals of video amplifiers is usually not
linear. That is, a small change in the input signal may make a greater than corresponding
change in the output. In other words, the monitor may exaggerate the color or grayscale
range of the input signalcontrast increases. The relationship between input and output
is referred to as the gamma of the amplifier. A gamma of one would result in an exact correspondence of the input and output signals. However, monitors with unity gammas tend
to have washed out, pastel images. Most people prefer higher gammas, in the range 1.5 to
1.8, because of their contrastier images.
Synchronizing Frequency Range

At one time, engineers locked computer monitors to one standard, much as video monitors lock to standard NTSC or PAL video signals. In order to access higher resolutions,
however, monitor-makers abandoned such fixed frequency monitors and opted for multiscanning monitors that could lock on a wide range of signals. Almost all new computer
monitors use multiscanning technology.
Coping with todays variety of signal standards makes mandatory a monitors ability to synchronize with the widest possible range of synchronizing frequencies. You have two frequencies to worry about. The vertical synchronizing frequency, sometimes called the refresh
rate or frame rate, determines how often the complete screen is updated. The horizontal synchronizing frequency (or horizontal scan rate) indicates the rate at which the individual scan
lines that make up the image are drawn. These frequency ranges are important to you
because they determine with which video standards the monitor can work.
The starting point for the scan rate of a modern monitor is the 31.5KHz used by the
VGA system. All current monitors require compatibility with this scan rate to make legible displays when your computer boots up. (Under the DVI standard, discussed later, new
digitally interfaced monitors can emulate this mode, although DVI itself will make scan
rate issues irrelevant.) A higher maximum scan rate is always better.
The lowest frame rate normally required is the 59Hz used by some early VESA modes,
although today youll only run into such low refresh rates with old display adapters. Youll
want a new monitor that supports the highest possible frame rate, as high as 85Hz under
some VESA standards.
859
860
Scan rate, refresh rate, and resolution interact. Table 24.3 lists the scanning frequencies of
most common computer display systems.
TA B L E 2 4 . 3
Scanning Frequencies Specified by Monitor Standards
Standard
Resolution
Vertical Sync
(Frame Rate)
Horizontal
Sync (Line Rate)
MDA
720350
50Hz
18.3KHz
CGA
640200
60Hz
15.75KHz
EGA
640350
60Hz
21.5KHz
MCGA (Graphics)
640480
60Hz
31.5KHz
MCGA (Text)
720400
70Hz
31.5KHz
VGA (Graphics)
640480
60Hz
31.5KHz
VGA (Text)
720400
70Hz
31.5KHz
Macintosh
640480
67Hz
35.0KHz
XGA-2
640480
75Hz
39.38KHz
VESA
640480
75Hz
37.5KHz
Apple Portrait
640870
76.5Hz
70.19KHz
VESA guideline
800600
56Hz
35.5KHz
VESA guideline
800600
60Hz
37.9KHz
VESA standard
800600
72Hz
48.1KHz
VESA standard
800600
75Hz
46.875KHz
RasterOps
1024768
75.1Hz
60.24KHz
Supermac
1024768
75.1Hz
60.24KHz
VESA guideline
1024768
60Hz
48.3KHz
VESA standard
1024768
70.1Hz
56.5KHz
VESA standard
1024768
75Hz
60KHz
8514/A
1024768
44Hz *
35.5KHz
XGA
1024768
44Hz *
35.5KHz
XGA-2
1024768
75.8Hz
61.1KHz
Apple 2-page
1152870
75Hz
68.68KHz
VESA standard
12801024
75Hz
80KHz
Displays
Note that two old standards (the legacy 8514/A and XGA systems) have very low frame
rates, 44Hz, because they are interlaced systems (discussed next). To properly display
these signals, a monitor must have sufficient range to synchronize with the field rate of
these standards, which is twice the frame rate.
Interlacing
Most computer display systems use progressive scanning. That is, the system starts by displaying the first (top) line in the video image and then displays each subsequent line, one
after the other down the screen. Televisions and video systems use interlaced scanning.
Instead of scanning lines sequentially from top to bottom, each frame of the image is broken in half into two fields. One field consists of the odd-numbered lines of the image; the
other the even-numbered lines. The electron beam sweeps across and down, illuminating
every other line, and then starts from the top again and finishes with the ones it missed on
the first pass.
This technique achieves an apparent doubling of the frame rate. Instead of sweeping down the
screen 30 times a second (the case of a normal television picture), the top-to-bottom sweep
occurs 60 times a second. Whereas a 30-frames-per-second rate would noticeably flicker, the
ersatz 60-frames-per-second rate does notat least not to most people under most circumstances. Some folks eyes are not fooled, however, so interlaced images have earned a reputation of being flickery. Figure 24.8 compares progressive and interlaced scanning.
FIGURE 24.8
Progressive
Interlaced
Progressive versus
interlaced scanning.
Frame
First field
Second field
Interlacing is used on computer display signals to keep the necessary bandwidth down. A
lower frame rate lowers the required bandwidth of the transmission channel. Although
interlaced scanning should have been forgotten along with the 8514/A and original XGA
display systems, it occasionally pops up in computer displaystypically low-priced monitors that claim high resolutions. Be wary of any monitor that specifies its maximum resolution as interlaced.
Bandwidth
Perhaps the most common specification usually listed for any sort of monitor is bandwidth, which is usually rated in megahertz. Common monitor bandwidths stretch across a
wide rangefigures from 12 to 200MHz are sometimes encountered.
In theory, the higher the bandwidth, the higher the resolution and sharper the image displayed. In the case of color displays, the dot-pitch of the display tube is the biggest limit
861
862
on performance. In a monochrome system, however, bandwidth is a determinant of overall sharpness. The computer display standards do not demand extremely wide bandwidths.
Extremely large bandwidths are often superfluous.
The bandwidth necessary in a monitor is easy to compute. A system ordinarily requires a
bandwidth wide enough to address each individual screen dot plus an extra margin to
allow for retrace times. (Retrace times are those periods in which the electron beam
moves but does not displayfor instance, at the end of each frame when the beam must
move from the bottom of the screen at the end of the last line of one frame back up to the
top of the screen for the first line of the next frame.)
A typical color display operating under the VGA standard shows 288,000 pixels (a 729400pixel image in text mode) 70 times per second, a total of 20.16 million pixels per second. An
800600-pixel Super VGA display at 75Hz must produce 36 million pixels per second.
Synchronizing signals require their own slice of monitor bandwidth. Allowing a wide margin of about 25 percent for retrace times, it can thus be seen that for most computer
applications, a bandwidth of 16MHz is acceptable for TTL monitors, and 10MHz of
bandwidth is sufficient for sharp composite video displays and figures well within the
claims of most commercial products. For VGA, 25MHz is the necessary minimum.
Multiplying the dot-clock by 25 percent yields an acceptable estimate of the bandwidth
required in a monitor. For the standards IBM promulgated, the company listed actual
bandwidth requirements. Table 24.4 summarizes these requirements and calculated estimates for various computer display standards.
TA B L E 2 4 . 4
Dot-Clocks and Recommended Bandwidths for Video Standards
Video Standard
Dot-Clock
Recommended Bandwidth
MDA
12.6MHz
16.3MHz
CGA
7.68MHz
14.3MHz
EGA
13.4MHz
16.3MHz
PGC
18.4MHz
25MHz
VGA (350- or 480-line mode)
18.4MHz
25MHz
VGA (400-line mode)
20.2MHz
28MHz
8514/A
34.6MHz
44.9MHz
VESA 800600, 75Hz
36MHz
45MHz
VESA 1024768, 75Hz
60MHz
75MHz
VESA 12801024, 75Hz
100MHz
125MHz
Displays
Although these estimates are calculated using the dot-clock of the display, the relationship
between dot-clock and bandwidth is not as straightforward as the calculations imply.
Because in real-world applications the worst-case display puts an illuminated pixel next to
a dark one, the actual bandwidth required by a display system should be the dot-clock plus
system overhead. A pair of on/off pixels exactly corresponds to the up and down halves of
a single cycle. The higher bandwidth you calculate from the dot clock allows extra bandwidth that gives greater image sharpnesssharply defining the edges of a pixel requires
square waves, which contain high-frequency components. Consequently, the multiplication of the dot-clock by display overhead offers a good practical approximation of the
required bandwidth, even though the calculations are on shaky theoretical ground.
Energy Star
Compared to some of the things you might connect to your computer, a monitor consumes a modest amount of electricity. A laser printer can draw as much as a kilowatt when
its fuser is operating. A typical computer requires 100 to 200 watts. A typical monitor
requires only about 30 watts. Unlike the laser fuser, however, your monitor may stay on
all day long, and an office may have hundreds of them, each continually downing its
energy dosage. Those watts add up, not just in their power consumption but also their
heat production that adds to the load on the office air conditioning.
To help cut the power used by computer equipment, the Environmental Protection
Agency started the Energy Star program, which seeks to conserve power while computers
and their peripherals are in their idle states. For monitors, Energy Star means that the
monitor powers down to a one of two lower power conditions or shuts off entirely when
its host computer has not been used for awhile.
Energy Starcompliant monitors have four operating modes: on, standby, suspend, and
off. During normal operation when the monitor displays an active image generated by
your computer, it is on. In standby mode, the monitor cuts off the electron beam in its
CRT and powers down some of its electronics. It keeps the filament or heat of the CRT
(the part that has to warm up to make the tube work) hot so that the monitor can instantly
switch back to its on state. In suspend mode, the filament and most of the electronics of
the monitor switch off. Only a small portion of the electronics of the monitor remain
operational to sense the incoming signals, ready to switch the monitor back on when the
need arises. This conserves most of the power that would be used by the monitor but
requires the CRT to heat up before the monitor can resume normal operation. In other
words, the monitor trades rapid availability for a reduction in power use. In off mode, the
monitor uses no power but requires you to manually switch it on.
To enable your computer to control the operating mode of your monitor without additional electrical connections, VESA developed its Display Power Management Standard
(DPMS). This system uses the two synchronizing signals your video board supplies to
863
864
your monitor to control its operating mode. To signal the monitor to switch to standby
operation, your video card switches off only its horizontal sync signal. To signal the monitor to switch to suspend mode, your video board cuts both the video signal and the vertical sync but leaves the horizontal sync on. In off mode, all signals are cut off. Table 24.5
summarizes these modes.
TA B L E 2 4 . 5
VESA Display Power Management Summary
Monitor
State
Video
Vertical
Sync
Horizontal
Sync
DPMS
Recovery
Time
Power
Savings
On
On
On
On
Mandatory
None
None
Standby
On
On
Off
Optional
Short
Minimal
Suspend
Off
Off
On
Mandatory
Longer
Substantial
Off
Off
Off
Off
Mandatory
Warm-up
Maximum
Advanced operating systems monitor your system usage and send out the standby
and/or suspend signals when you leave your system idle for a predetermined time. Your
video driver software controls the DPMS signals. Note that screensaver software defeats
the purpose of DPMS by keeping your system active even when it is not in use. You can
trim your power usage by relying on DPMS rather than a screensaver to shut down
your monitor.
Many monitors made before the DPMS standard was created often incorporate their own
power-saving mode thats initiated by a loss of the video signal for a predetermined time.
In other words, these monitors sense when your screen is blank and start waiting. If you
dont do something for, say, five minutes, the monitor switches to standby or suspend
mode, no matter the state of the synchronizing signals. The DPMS system is designed so
that these monitors, too, will power down (although only after the conclusion of both the
DPMS and their own internal timing cycles).
To enable the DPMS system under Windows, you must tell the operating system that
your monitor is Energy Star compliant. If your monitor is in the Windows database,
Windows will know automatically whether the monitor is compliant. You can also check
the Energy Star compliance in the Change Display Settings screen. You must also activate
the DPMS screen blanker from the Screen Saver tab of your Display Properties screen.
Identification
The prevalence of multiscanning monitors with wildly different capabilities makes getting
the most from your investment a challenge. You must be able to identify not only the display capabilities of the monitor, but also those of your video board, and then make the
Displays
best possible match between them. If youre wrong, you wont get everything that youve
paid for. Worse, you might not see anything intelligible on your screen. Worse still, you
face a tiny chance of actually harming your monitor or video board.
Hard-Wired Coding
The problem is neither new nor one that arose with multiscanning systems. When the
VGA system was new, IBM developed a rudimentary system for letting its computers
determine the type of monitor that was connectedlimited, of course, to the monitors
IBM made. The system had limited capabilities. It could identify whether the monitor was
monochrome or color and whether it met merely the VGA standard or advanced into
higher-resolution territory. At the time, IBM only offered four monitors, and that modest
range defined the extent of the selection.
The IBM scheme was to use three of the connections between the monitor and the video
board to carry signals identifying the monitor. These first signals were crudea simple
binary code that put the signal wire either at ground potential or with no connection.
Table 24.6 lists this coding system.
TA B L E 2 4 . 6
Monitor-Identification Coding Used by IBM
Display Type
Size
IBM Model
ID 0
ID 1
ID 2
Monochrome
12 inch
8503
NC
Ground
NC
Color
12 inch
8513
Ground
NC
NC
Color
14 inch
8512
Ground
NC
NC
Hi-resolution
15 inch
8514
Ground
NC
Ground
Display Data Channel

This rudimentary standard was not up to the task of identifying the wide range of monitor
capabilities that became available in the years after the introduction of VGA. Yet adding
true Plug-and-Play capabilities to your computer requires automatically identifying the
type of monitor connected to your computer so that the display adapter (and the rest of the
system) can be properly configured. To meet this challenge, VESA developed the Display
Data Channel (DDC), an elaborate monitor-identification system based on the same connections as the early IBM system but with greatly enhanced signals and capabilities.
Through the DDC, the monitor sends an Extended Display Identification (EDID) to your
computer. In advanced form, the DDC moves data both ways between your monitor and
your computer using either the I2C or ACCESS.bus serial interfaces. The DDC2B standard
uses I2C bus signaling on two of the wires of the monitor connection to transfer data both
865
866
ways between monitor and its attached video board. DDC2AB uses a full ACCESS.bus connection and protocol, which allow you to connect other computer peripherals (for example,
your keyboard) to your monitor rather than the system unit.
All levels of the DDC system gain information about your monitor in the same way.
The monitor sends out the EDID as a string of serial bits on the monitor data line, pin
12. Depending on the level of DDC supported by the monitor, the EDID data stream is
synchronized either to the vertical sync signal generated by the video board present on
pin 14 of 15-pin video connectors or to a separate serial data clock (SCL) thats on pin
15 in DDC2 systems. One bit of data moves with each clock cycle. When the system
uses vertical sync as the clock, the data rate will be in the range of 60 to 85Hz. With
DDC-compliant monitors, your video board can temporarily increase the vertical sync
frequency to up to 25KHz to speed the transmission of this data. Using the SCL signal
when both the video board and monitor support it, data rates as high as 100KHz are
possible.
The serial data takes the form of nine-bit sequences, one per byte. The first eight bits
encode the data, the most significant bit first. The last bit can be either a zero or one, at
the choice of the monitor manufacturer. The only restriction is that the ninth bit must
have the same value for every byte.
The DDC system sends data from the monitor to your display adapter in 128-byte blocks.
The first of these is the Extended Display Identification or EDID block. It is optionally
followed by an Extended EDID block or additional proprietary manufacturer data blocks.
Table 24.7 lists the structure of the basic EDID.
TA B L E 2 4 . 7
Basic EDID Structure
Start Byte
Length
Description
8 bytes
Header
10 bytes
Vendor/product identification
18
2 bytes
EDID version/revision
20
15 bytes
Basic display parameters/features
35
19 bytes
Established/standard timings
54
72 bytes
Timing descriptions 4 (18 bytes each)
126
1 byte
Extension flag
127
1 byte
Checksum
Displays
The header is always the same data pattern and serves to identify the EDID information
stream. The vendor identification is based on EISA manufacturer identifications. The
product identification is assigned by the manufacturer. It includes the month and year of
manufacture of the monitor.
The basic display parameters that EDID relays to your system include the maximum size
of your monitors display, expressed as the largest width and height of the image area.
Your applications or operating system can use this information to automatically set the
proper scaling for fonts displayed on the screen. The timing data includes a bit representing the ability to support each of the various VESA standards so that your system can
determine the possible modes and frequencies your monitor can use.
In addition to basic DDC support, VESA provides for two higher levels of standardization. The DDC2B system uses the Philips I2C signaling system to transfer data bidirectionally across the interface. The DDC2AB system includes full ACCESS.bus support
that supplies a low-speed serial interconnection bus suitable for linking such peripherals as
keyboards and pointing devices through the monitor.
Because standard expansion buses do not provide suitable connections for routing
ACCESS.bus signals, VESA has defined an auxiliary connector for the purpose, a five-pin
Berg connector. Table 24.8 lists the signal assignments of this connector.
TA B L E 2 4 . 8
Access Bus Connector Signal Assignments
Pin
Function
Ground
Mechanical key
Serial data (SDA)
+5V ACCESS.bus supply voltage
Serial data clock (SCL)
Monitors that are compliant with DDC use the same connectors as ordinary VGA displays. All the active video and synchronizing signals are located on the same pins of the
connector, no matter the DDC level the monitor uses (or even if it doesnt use DDC at
all). The only difference is the definition of the monitor-identification signals. DDC1
video boards sacrifice the Monitor ID Bit 1 pin, number 12, as a channel to receive
identification data from your monitor. DDC2 systems make this signal bidirectional and
take over pin 15 for use carrying the clock signal. In any case, pin 9 may be used to supply five volts for running accessory devices. Table 24.9 lists the signal assignments of the
VGA 15-pin connector under DDC.
867
868
TA B L E 2 4 . 9
VESA Display Data Channel Signal Assignments
Pin
DDC1 Host
DDC2 Host
DDC1/2 Display
Red video
Red video
Red video
Green video
Green video
Green video
Blue video
Blue video
Blue video
Monitor ID bit 2
Monitor ID bit 2
Optional
Return
Return
Return
Red video return
Red video return
Red video return
Green video return
Green video return
Green video return
Blue video return
Blue video return
Blue video return
+5V (optional)
+5V (optional)
+5V load (optional)
10
Sync return
Sync return
Sync return
11
Monitor ID bit 0
Monitor ID bit 0
Optional
12
Data from display
Bidirectional data
Bidirectional data
13
Horizontal sync
Horizontal sync
Horizontal sync
14
Vertical sync
Vertical sync
Vertical sync
15
Monitor ID bit 3
Data clock (SCL)
Data clock (SCL)
Manual Configuration
If your monitor or video board does not support any level of the DDC specification, you
will be left to configure your system on your own. In general, you wont need to perform
any special configuration to a multiscanning monitor to match it to your video board if
the signals from the video board are within the range of the monitor. Thats the whole
point of the multiscanning displayit adjusts itself to accommodate any video signal.
That said, you may not get the most from your monitor. You might slight its refresh rate
and quickly tire your eyes with a flickering display. Worse, you might exceed its capabilities and end up with a scrambled or blank screen.
Windows includes its own configuration process that attempts to optimize the signals of
your compliant video adapter with your monitor type. Windows already knows what video
board you haveyou have to tell it when you install your video drivers. Without a DDC
connection, however, Windows is cut off from your monitor, so you must manually indicate the brand and model of monitor youre using.
Displays
You make these settings by clicking the Change Display Type button in the Settings tab of
your Display properties menu. Click the button, and you see a dialog box like that shown
in Figure 24.9.
FIGURE 24.9
The Change Display
Type dialog box in
Windows.
Interfaces
The video system in your computer must somehow link up with the rest of its circuitry. In
most display systems, the connection is intimate. It gives your computers microprocessor
the power to directly intervene and inject data, such as an image bitmap, right into the
frame buffer. In nearly all of todays computers, the necessary link is an accelerated graphics port (AGP).
Although based on the PCI expansion bus, AGP delivers substantially higher performance
because it incorporates dedicated pipelined access to main memory. Thanks to its wide
bandwidth, the design allows an accelerator chip on an AGP expansion board to use main
memory for video functions such as z-buffering, texture-mapping, and alpha-blending.
The AGP may be embedded and hidden in the circuitry of your computer when the display system is part of the motherboard or chipset. Otherwise, AGP takes the form of a
single special slot on your desktop computers motherboard.
Accelerated Graphics Port

The AGP design originated in Intel Labs and was officially released as a standard
(version 1.0) on July 31, 1996. As you might expect from a new bus design from Intel, it is
based on the PCI design and follows basic PCI protocol and signaling. Because of its high
speedpotentially four times that of todays PCIit uses its own, incompatible connector
869
870
design and imposes tight restrictions on design layouts. On May 4, 1998, Intel revised the
AGP design to version 2.0. The improvements were twofold. By adding a transfer mode,
Intel created a new quadruple-speed mode for AGP. In addition, the new specification
allows for lower-voltage operation. In addition to the 3.3-volt design of AGP 1.0, the revision allows for 1.5-volt operation as well.
Although based on revision 2.1 of the PCI specifications, the AGP design goes beyond
them with three innovations: pipelined memory that eliminates wait states, separate
address and data lines on the bus to cut the necessary number of cycles per transfer, and a
timing specification allowing operation at 133MHz across a 32-bit bus, thus allowing for
throughput on the order of 500MBps. The actual clock speed of the AGP bus is 66MHz,
matching the PCI 2.1 specs. To achieve the higher data rates, the AGP system relies
on more exotic technologies, discussed in the 2 transfer mode and 4 transfer mode
sections, later.
The AGP shares the same physical address space for memory with the PCI system and
microprocessor. It uses a 32-bit address bus that can handle up to a 4GB address range.
Matching its 32-bit data bus, AGP transfers always move data in eight byte blocks, double
the granularity of standard PCI transfers (which slices data into four bytes blocks). To
accommodate smaller data types, the AGP interface can also use PCI transfer modes.
Despite its PCI foundation, AGP has its own set of commands and transfer protocol. The
AGP interface hardware supports both the AGP transfers and traditional PCI transfers as
separate modes.
Unlike PCI, where address and data cycles are intimately linked, the AGP modes disconnect them. This allows other data to intervene in between. This design allows the system
to keep the pipeline filled. Each access request across the AGP is sent to a queue while it
awaits the transfer of its associated data. The AGP devices track the access requests and
data in their hardware. Access requests in the queue are prioritized into high- and lowpriority subqueues. PCI transfers have their own queue.
The AGP design allows for a maximum of 256 transactions in the pipeline at once,
although hardware capabilities will further constrain the pipeline. When the Plug-andPlay system sets up the AGP board, the AGP hardware reports the maximum depth that
the pipeline can support in its host computer.
On the other hand, the AGP transfer modes allow access only to memory. They do not
link to input/output ports or configuration memory. In addition, to help the AGP achieve
its high speed, its design cuts it free from cache snooping, so changes made through AGP
are not necessarily reflected in the memory cache.
Displays
Architecture
AGP clearly shows the new role of video in the future of the computer. Through AGP,
video is given its own dedicated interface and connection. It does not share a bus with
expansion boards or even memory. It has a direct reach to the high holy sanctum of the
computer, the microprocessor and chipset.
In effect, the AGP is a fourth bus in a modern computer, added to the expected compatibility bus (ISA), high-speed expansion bus (PCI), and memory bus. The signals for AGP
arise in the same system-control logic that links to the PCI bridge and main memory in
the computer. As shown in Figure 24.10, the AGP design gives the graphics chip of a
computer the same direct connection with the chipset as honors the microprocessor, PCI
bridge, and memory system.
FIGURE 24.10
PCI
Bridge
Block diagram
showing the role of
AGP in modern system architecture.
Microprocessor
AGP
Chipset
Graphic
Chip
PCI slots
ISA slots
Main
Memory
Frame buffer
AGP operates solely between two devices. By design, one of these must be the system
memory controller, which is normally part of the computers chipset. To the AGP system,
the chipset is the target device, and AGP transfers are controlled by the graphics board at
the other end of the connection, which is the master. In PCI transfer modes, the host
chipset operates as a PCI sequencer, which is capable of service as both master and target.
The video board only needs to operate as a PCI target, although it may also have master
functions as well.
Operation
Intel envisioned that the AGP would be used in two different ways. Its DMA model pictured AGP would be used as a fast channel for moving large blocks of data between a dedicated frame buffer and main system memory. Your computers graphic chip would work
conventionally, working in the memory on the expansion board, carrying out most operations directly in the frame buffer. The system would use DMA transfers across the AGP
to move bit-image data from system memory to the frame buffer. The execute model sees
871
872
the AGP as a means of connecting the graphics accelerator to main memory. The accelerator chip on the video board can use main memory for execution of high-level functions
such as antialiasing and 3D rendering. The graphics chip would use the AGP to draw data
from system memory and hold intermediate results, loading the final data of the image
into the local frame buffer.
In either case, transfers are mediated in the same way. First, either end of the connection
negotiates to use the interface.
When operating in its native AGP mode, the graphics controller starts each transaction by
making an access request to the system board chipset. The chipset then schedules its
action on the request, working around other tasks that involve the bus. The chipset can
accumulate several graphics requests and begin to process them while waiting for the
actual data transfer to take place. In effect, the processing steps involved in the transfer
requests are pipelined much like instructions in an advanced microprocessor.
The transfer requests in the pipeline are split into two queues, one with high priority and
one with low priority, for each read and write request. The AGP system puts the responsibility for tracking the status of all the transfers on the graphics controller requesting
them, and the controller determines the number of requests waiting in each queue. The
chipset processes the requestfor example, transferring data from system memory to the
frame bufferas free time in the system allows it. It can process requests from the various
queues as system resources allow, so it may interleave the processing of read and write
requests.
When operating in its PCI mode, the graphics controller acts like any PCI device.
However, the PCI transfers are not pipelined like the native AGP transactions.
The signals used for AGP transfers are built on a foundation of the PCI bus. The AGP
standard augments the PCI standard with special control signals Intel terms sideband signals.
They alter the timing of the PCI bus to allow higher data throughput. The PCI data and
bus control lines still channel information during AGP transfers, but a new signal termed
PIPE# defines the AGP transfer, much as the FRAME# signal defines PCI transfers.
When the PIPE# signal is active, the system operates in the AGP 2 transfer mode and
moves data across the bus twice during each cycle. Request queuing, pipelining, parallel
execution, and the double-time transfer cycle all contribute to the higher throughput of
the AGP interface.
AGP-defined protocols (for example, pipelining) are overlaid on the PCI bus at a time and
in a manner that a PCI bus agent (non-AGP) would view the bus as idle. Both pipelined
access requests (read or write) and resultant data transfers are handled in this manner. The
AGP interface uses both PCI bus transactions without change, as well as AGP pipelined
transactions as defined herein. Both of these classes of transactions are interleaved on the
Displays
same physical connection. The access request portion of an AGP transaction (bus command, address, and length) is signaled differently than is a PCI address phase. The information is still transferred on the AD and C/BE# signals of the bus, as is the case with PCI
6, but is identified or framed with a new control signal, PIPE#, in a similar way to which
PCI address phases are identified with FRAME#.
2 Transfer Mode
For high-performance systems in which the 266MBps potential of basic AGP is inadequate, the original AGP specification allowed for higher-speed transfers in its 2 transfer
mode. Using the proven technique called double-clocking, 2 mode pushes the maximum
potential throughput of the AGP system to 533MBps.
Double-clocking makes the clock signal do double duty, keying transfers to both its leading and trailing edges. That is, the transition marking the beginning of the clock signal
triggers one transfer across the bus, and the transition ending the clock pulse triggers
another. In most logic systems and basic AGP mode, only the leading edge of the clock
signals valid data, allowing the making of a transfer.
To improve the reliability of 2 transfer mode, it uses three strobe signals rather than the
clock itself. The clock signal originates in your computer as part of its array of PCI signals. In contrast, the device sending data generates the strobe signals. When your computer sends data to the AGP board, it creates the strobes. When the AGP board sends
data to your computer, it creates the strobes. Because the sending device generates both
the data and strobe signals, the timing relationship between the two can be more precisely
controlled.
The strobe signals appear only within the interconnection between the AGP transmitting
and receiving circuits and the hardware connection between them. In AGP terminology,
this is the inner loop. The rest of the AGP system is the outer loop.
The specification defines individual strobe signals for the low and high 16 bits of the AGP
bus as well as the sideband. The AGP circuitry issues the strobe signals at the same frequency as the AGP clock, but the specification more rigidly defines the timing of the
strobes. The high and low phases of the strobe signal are equal, so leading and trailing
edges of the signals are evenly spaced and transfers keyed to the edges occur uniformly.
4 Transfer Mode
In enhancing the performance of AGP, the Intel engineers faced a major problem. The
original design already strained at the high-frequency limit of the connector system.
Increasing the speed of AGP would require a redesign with expensive new technologies.
To avoid most of these issues, the Intel engineers kept their changes inside the inner loop,
designing a more elaborate strobe system.
873
874
To create 4 transfer mode, Intels engineers doubled the frequency of the strobe signals
and double-clocked the data keyed to them. This change again doubles the potential
throughput of the system, yielding a maximum rate of 1066MBps.
With the increase in strobe frequency comes a need for a tighter definition of the timing
of the signals. Intel created the necessary precision by adding three more strobe signals.
Each one is paired with one of the 2 transfer mode signals but is the opposite polarity,
turning each strobe into a differential pair. When one of the strobes in a pair transitions
from high to low, the other moves from low to high. At one instant during the paired
transitions, at approximately midway between them, the voltage values of the two signals
are equalon a timing chart, the values cross. The 4 transfer mode keys each transfer
to one of these crossings. Because these crossings occur twice in each strobe cycle and
the strobe frequency is twice the basic AGP clock frequency, transfers occur at four times
the basic AGP rate.
Intel took another step to help AGP achieve the speed required by 4 transfer mode. It
lowered the voltage of the signals, which helps deal with the capacitance inherent in any
bus system. All AGP boards that operate in AGP 4 transfer mode must use 1.5-volt
signaling.
UMA
Unified Memory Architecture (UMA) is an alternate way to design a computer for lowercost construction. It actually can have a beneficial effect on video performance. Because the
frame buffer is integrated into the main memory of the computer, bitmaps dont have to
cross any I/O bus at all. Of course, the memory used by the frame buffer cannot also be used
by programs, but todays memory prices make this loss of storage almost inconsequential.
To provide a common standard to which designers could create the electrical and logical
interface between the video system using UMA and the core logic of a computer, VESA
developed its own standard called VESA Unified Memory Architecture (VUMA). First published in March, 1996, VUMA allows for adding external devices to your computer that
can share direct access to its main memory system to use it as a frame buffer. Although
UMA has become popular in low-cost computers, even with this standard, add-on devices
that use the technology have not become popular.
Monitor Interfacing and Connectors

The video connection carries the image between your video board and monitor.
Monitors can be grouped by the display standard they support, mostly based on the display adapter card they are designed to plug into. One basic guide that helps you narrow
down the compatibility of a display just by inspecting its rear panel is the input connector
Displays
used by the monitor. After all, if you cannot plug a monitor into your computer, odds are
it is not much good to you.
Youll likely encounter one of three styles of connector on the back of your computer.
These include legacy connectors used by display systems too ancient to worry about, analog system used by the majority of todays display systems, and new digital interfaces that
are beginning to appear on the market and mark the look (and connection) of things to
come.
Legacy Connectors
The first video connector used by computers was the nine-pin D-shell connector, as shown
in Figure 24.11. It is now used only by legacy video systems and serves mostly as a warning
that you may encounter trouble in trying to match your video system to a monitor.
FIGURE 24.11
A nine-pin D-shell jack
used by legacy video
systems.
Analog Connection Systems

Three styles of connectors are shared by different computer video standards. By name,
these three connectors are the RCA-style pin jack, the nine-pin D-shell, and the 15-pin
high-density D-shell. In addition, some high-resolution monitors use three or more
BNC connectors for their input signals.
Pin Jacks
The bulls-eye jack used on stereo and video equipment is used by most manufacturers for
the composite video connections in computer display systems, although a wealth of monitors and television sets made by innumerable manufacturers also use this connector. This
connector does give you many choices for alternate displaysthat is, if you dont mind
marginal quality.
Composite monitors (those dealing with the composite video and NTSC color only) rank
among the most widely available and least expensive in both color and monochrome. Even
better quality television sets have such jacks, shown in Figure 24.12, available.
FIGURE 24.12
A pin jack used by
composite video
signals.
875
876
Although you can use any composite video display with a CGA or compatible color card,
the signal itself limits the possible image quality to okay for monochrome, acceptable for
40-column color, and unintelligible for 80-column color. Nevertheless, a composite video
displayalready a multipurpose devicebecomes even more versatile with a computer
input.
Daisy-Chaining
A side benefit of pin plug/composite video displays is that most have both input and output jacks. These paired jacks enable you to daisy-chain multiple monitors to a single video
output. For example, you can attach six composite video monitors to the output of your
computer for presentations in the classroom or boardroom.
In many cases, the jacks just loop through the display (that is, they connect together). The
display merely bridges the input video signal and alters it in no other manner. You can
connect a nearly unlimited number of monitors to these loop-through connections with
no image degradation. Some monitors, however, buffer their outputs with a built-in video
amplifier. Depending on the quality of the amplifier, daisy-chaining several of these monitors can result in noticeable image degradation.
One way to tell the difference is by plugging the output of the display into the output of
your computer. Most amplifiers dont work backwards, so if the display has a buffering
amplifier, nothing appears onscreen. If you do get an image comparable to the one you
get when plugging into the input jack, the signal just loops through the display.
Analog Voltage Level
The specifications of composite monitors sometimes include a number describing the
voltage level of the input signal. This voltage level can be important when selecting a
composite display because all such monitors are essentially analog devices.
In analog monitors, the voltage level corresponds to the brightness the electron beam displays onscreen. A nominal one-volt peak-to-peak input signal is the standard in both the
video and computer industries and should be expected from any composite monitor. The
VGA system requires a slightly different level0.7 volts.
Termination
For proper performance, a composite video signal line must be terminated by an impedance of 75 ohms. This termination ensures that the signal is at the proper level and that
aberrations do not creep in because of an improperly matched line. Most composite input
monitors (particularly those with separate inputs and outputs) feature a termination switch
that connects a 75-ohm resistor across the video line when turned on. Only one termination resistor should be switched on in any daisy-chain, and it should always be the last
monitor in the chain.
Displays
If you watch a monitor when you switch the termination resistor on, youll notice that
the screen gets dimmer. Thats because the resistor absorbs about half the video signal.
Because composite video signals are analog, they are sensitive to voltage level. The termination cuts the voltage in half and consequently dims the screen by the same amount.
Note that the dim image is the proper one. Although bright might seem better, its not.
It may overload the circuits of the monitor or otherwise cause erratic operation.
Composite monitors with a single video input jack and no video output usually have a termination resistor permanently installed. Although you might try to connect two or more
monitors to a single CGA composite output (with a Y cable or adapter), doing so would
be unwise. With each additional monitor, the image gets dimmer (the signal must be split
among the various monitors), and the CGA adapter is required to send out increasing current. The latter could cause the CGA to fail.
VGA Connectors
The most common connector on computer monitors is the 15-pin high-density D-shell connector. Originally put in use by IBM for its first VGA monitors, this connector has
become adopted as an industry standard for all but the highest performance computer displays.
Because the signals generated by VGA are so different from those of previous IBM display
systems, IBM finally elected to use a different, incompatible connector so the wrong monitor wouldnt be plugged in with disastrous results. Although only nine connections are
actually needed by the VGA system (11 if you give each of the three video signals its own
ground return as IBM specifies), the new connector is equipped with 15 pins. Its roughly
the same size and shape as a nine-pin D-shell connector, but before IBMs adoption of it,
this so-called high-density 15-pin connector was not generally available. Figure 24.13
shows this connector.
FIGURE 24.13
A 15-pin mini D-shell
connector used by
VGA and similar video
systems.
In addition, to allow for four video signals (three primary colors and separate sync) and
their ground returns, the VGA connector provides a number of additional functions. In
the original VGA design, it enabled the coding of both monitor type and the line count of
the video signal leaving the display adapter. The modern adaptation of the connector to
the VESA DDC standard redefines several pins for carrying data signals, as noted earlier.
Table 24.10 lists the signal assignments of the VGA connector.
877
878
TA B L E 2 4 . 1 0
VGA and SuperVGA Connector Pinout
Pin
Function
Red video
Green video
Blue video
Reserved
Ground
Red return (ground)
Green return (ground)
Blue return (ground)
Composite sync
10
Sync return (ground)
11
VESA display data channel
12
Reserved
13
Horizontal sync
14
Vertical sync
15
VESA display data channel
IBMs 8514 and 8515 displays as well as 8514/A and XGA display adapters also use the
same connector, even though they at times use different signals. Again, however, IBM has
incorporated coding in the signals to ensure that problems do not arise. The 8514/A and
XGA adapters can sense the type of display connected to them and do not send out conflicting signals. The 8514 and 8515 monitors operate happily with VGA signals, so problems do not occur if they are plugged into an ordinary VGA output.
The advantage of the 15-pin connector is convenience. One cable does everything. On
the downside, the connector is not arranged for proper high-speed operation, and its deficiencies can limit high-frequency performance, which in video terms equates to sharpness
when operating at high resolutions and refresh rates. Consequently, the highest resolution
systems often forego the 15-pin connector for separate BNC connectors for each video
channel.
BNC Connectors
True high-resolution systems use a separate coaxial cable for every signal they receive.
Typically, they use BNC connectors to attach these to the monitor. They have one very
Displays
good reason: Connectors differ in their frequency-handling capabilities, and capacitance

in the standard 15-pin high-density D-shell connector can limit bandwidth, particularly as
signal frequencies climb into the range above 30MHz. BNC connectors are designed for
frequencies into the gigahertz range, so they impose few limits on ordinary video signals.
Monitors can use either three, four, or five BNC connectors for their inputs. Figure 24.14
shows an array of five BNC connectors on the rear of a monitor.
FIGURE 24.14
A monitor with five
BNC connectors for
input.
A three-connector system integrates both horizontal and vertical synchronizing signals

with the green signal. The resulting mix is called sync-on-green. Others use three connectors for red, green, and blue signals and a fourth for the horizontal and vertical sync signals combined together. This scheme is called composite sync. Five-connector systems use
three color signals, one for horizontal sync, and one for vertical sync. These are called separate sync systems.
Digital Video Interface

Flat-panel displays made the need for a high-performance digital display interface obvious. Analog-interface flat panels work but are only an interim solution.
Among the other problems with the standard analog links so popular on todays monitors,
one stands in the way of getting truly high-resolution images on your monitorlimited
bandwidth. The now-dominant VGA connector has a maximum bandwidth of about
150MHz, so it already constrains the quality of some display systems. Conventional highfrequency video connectors (primarily BNC) require a separate cable for each video signal
and allow no provisions for auxiliary functions such as monitor identification.
Achieving a single standard has been problematic for the industry, however. In the last few
years, various groups have developed at least three specifications, all promoted as standards. Despite the infighting among the promoters of each, the three are not in competition, however, but represent the evolution of what will likely be the dominant standard for
the future, the Digital Video Interface (DVI).
In late 1998, Intel rounded up the usual suspects to work on yet another digital display
standard. The result was the Digital Display Working Group, initially comprising representatives from both the Plug and Display and Digital Flat Panel (DFP) connector
campsIntel, Compaq, Fujitsu, Hewlett-Packard, IBM, NEC, and Silicon Image.
The working group took the better features of each of the two earlier standards and
developed the Digital Video Interface as kind of the greatest hits of monitor connectors.
879
880
The standard was adopted and published by VESA as DVI version 1.0 on April 2, 1999.
Number Nine Visual Technology demonstrated the first product to use the interface, the
SR9 graphics adapter, on May 19, 1999.
DVI shares the connector design and some of the non-video data signals with Plug and
Display, but it breaks off the integral USB port from the connector. Although the DVI
standard does not preclude USB in displays, to do so a separate connection must be used
for the USB signals.
Although DVI uses the same Transition Minimized Differential Signaling (TMDS) signals
as Plug and Display and DFP, it doubles the effective bandwidth of these signals with the
simple expedient of using two of them. Both are locked to the single clock channel provided by the interface, so they are automatically synchronized to each other. Each color
signal has two signals, which are interleaved so that data is spread between the two channels of a given color, creating what is essentially a parallel connection. Display systems
may use either the one-channel system or the two-channel system, but the DVI specification draws a hard line between the two. Signals requiring a bandwidth of less than
165MHz must use the single-channel system. Systems requiring larger bandwidths must
use two channels.
The three signal paths of DVI are not named with colors, although they correspond to
red, green, and blue data. Each signal path actually carries three interleaved channels of
data. The signal path used by blue data, for example, also carries a horizontal synchronizing signal and a vertical synchronizing signal. The two channels in the remaining pair of
signal paths are currently reserved and have no assigned function.
Analog video is an optional part of DVI. Rather than shape, the DVI connector uses the
cross-pin in the connector as a key for the presence of analog signals. The same connector
can be used for all-analog or all-digital systems. Analog signals, when present, are located
in the four quadrants around the keying cross in the connector.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
25
Audio Systems
Of the five senses, most people only experience personal computers with
four: touch, smell, sound, and sight. Not that computers are tasteless
although a growing body of software doesnt even aspire that highbut
most people dont normally drag their tongues across the cases of their
computers.
Touch is inherent in typing and pushing around a mouse or digitizer
cursor.
Smell is more limited stillwhat you appreciate in opening the box
holding your new computer or what warns you when the fan in the
power supply stops, internal temperatures rise, and roasting resistors and
near inflammatory components begin to melt.
Most interactions with computers involve sight: What you see on the
monitor screen and, if youre not a touch-typist, a peek down at the keyboard. High-resolution graphics make sight perhaps the most important
part of interacting with any computeror at least the most expensive.
To really experience your computer, however, you need to get aural with it.
You need an added sensual dimensionsound. In fact, top-quality sound
distinguishes todays computers from previous generations (at least those
modern computers meant for home use rather than as office workstations).
Today, a computer can generate sounds on its own, acting like a music
synthesizer or noise generator, and it can control external devices that do
the same thing through a MIDI interface. It can record or sample sounds
on any standard computer medium (the hard disk being todays preferred
choice) with sonic accuracy every bit as good (even better) than commercial stereo CDs. It can file all of your music and dole it out to your MP3
884
player. Moreover, all the sounds it makes and stores can be edited and manipulated: Tones
can be stretched; voices shifted; noises combined; music mixed. It can play back all the
sounds it makes and records with the same fidelity, pushing the limits of even the best
stereo systems.
Sound
Sound is different from what your computer normally deals with. The electrical signals
inside and at its ports, even the light from your monitor, are electro-magnetic phenomena.
Sound is purely physical. It is simply the rapid change in air pressure.
When a physical object vibrates, it forces air to move along its undulations. After all, air
and the object cannot take up the same place at the same time. Air is pushed away from
the place the object bows out and rushes into the empty place where it bends in. The
movements are tiny to your eye, but on the molecular level, they are huge.
Air, being made from physical molecules, has its own inertia. It cant immediately move
out of the way of a vibrating object but instead gets squeezed, which raises the pressure
between molecules. Where the air gets sucked toward the vibrating object, the pressure
goes down. Consequently, the air around a vibrating object is repeatedly squeezed and
expanded as the object moves. The pressure changes affect the adjoining air, so the
changes in pressure move away from the vibrating object as a series of high-and-low pressure waves. Your ear detects the changes in air pressure as sound.
Unlike light, which happily traverses the vacuum of space, sound requires a medium (usually air) for transmission. The speed of sound depends not on the moving object but on
the density of the medium. The higher the density, the faster the sound movesfor
example, sound propagates faster in water than air.
Because the waves of sound spread out as they move away from a vibrating object, the
intensity of the sound pressure declines with distance. Unconstrained, this decline would
follow the famous inverse-square law, because as the sound travels in one dimension, it
must spread over two. By confining or directing the air channel, however, you can alter
the rate of this decay.
Fitting its digital signals into this world of pressure waves is a challenge for the computer in
several ways. The computer needs a convenient form for manipulating sound. Fortunately,
sound has an analog in the electronic world called analog audio, which uses electrical signals
to represent the strengths of the pressure waves. Computers turn these electrical signals into
digital audio thats compatible with microprocessors, other digital circuits, and sound systems.
Of course, the computer is not limited to sounds supplied by othersit can create the digital signals itself, a process termed synthesis. To turn those digital signals into something that
approaches soundback to audio againyour computer uses its own audio circuitry or a
CHAPTER 25 AUDIO SYSTEMS
Analog Audio
soundboard that includes both a digital-to-analog converter and an amplifier. Finally, your
computer plugs into loudspeakers, which convert the audio into pressure waves once again.
Analog Audio
As a human sensation, sound is an analog phenomenon. It has two primary characteristicsloudness (or amplitude) and frequencythat vary over a wide range with an infinite
range of variations between its limits. Sounds can be loud or soft or any gradation in
between. Frequencies can be low or high or anything in between.
Frequency
Sound frequencies are measured in hertz, just like the frequencies of computer clocks or
radio signals. The range of frequencies that a human being can hear depends on age and
sex. Younger, female ears generally have wider ranges than older, male earsmaybe thats
why some older adults miss the nuances of rap music. Maybe not. In any event, most
sources ignore individual differences and list the range of human hearing as being from 20
hertz to 15,000 hertz (or as high as 20,000 hertz if your ears are particularly good).
Lower frequencies correspond to bass notes in music and the thumps of explosions in theatrical special effects such as exploding spaceships and particularly tenacious indigestion.
High frequencies correspond to the treble in music, the bright, even harsh sounds that
comprise overtones in musicthe brightness of strings, the tinkle of bells, the sharp edge
of crashing glassas well as hissy sounds, such as sibilants in speech, the rush of waterfalls, and overall background noise.
Low frequencies have long wavelengths, in the range of ten feet (three meters) for middle
bass notes. The long wavelengths allow low frequencies to easily bend around objects and,
from a single speaker, permeate a room. Moreover, human hearing is not directionally
sensitive at low frequencies. You cannot easily localize a low frequency source. Acoustical
designers exploit this characteristic of low frequencies when they design low-frequency
loudspeakers. For example, because you cannot distinguish the locations of individual lowfrequency sources, a single speaker called a subwoofer is sufficient for very low frequencies,
even in stereo and multichannel sound systems.
High frequencies have short wavelengths, measured in inches or fractions of an inch (or
centimeters). They can easily be blocked or reflected by even small objects. Human hearing is acutely sensitive to the location of higher frequency sources.
Amplitude
Amplitude describes the strength or power of the sound. The amplitude of sound traveling through the air is usually expressed as its sound pressure level (SPL). The threshold
of human hearing is about 0.0002 microbarswhich means a pressure change of
885
886
1/5,000,000,000th (one five-billionth) of normal atmospheric pressure. In other word, the

ear is a sensitive detector of pressure changes. Were it any more sensitive, you might hear
the clink of Brownian motion as dust particles ricochet through the air.
In audio systems, electrical signals take the place of sound pressure waves. These signals
retain the characteristic frequencies of the original sounds but their amplitude refers to
variations in electrical strength. Usually the voltage in an audio system represents the
amplitude of pressure of the original sound waves.
Decibels
A term that youll usually see engineers use in measuring amplitude loudness is the decibel
(dB). Although the primary unit is actually the bel, named after Alexander Graham Bell,
the inventor of the hydrofoil (and yes, the telephone), engineers find units of one-tenth
that quantity to be more manageable.
The decibel in itself represents not a true measuring unit but a relationship between two
measurements. The bel is the ratio between two powers expressed as a logarithm. For
example, a loud sound source may have an acoustic power of one watt, whereas a somewhat softer source may only generate one milliwatt of power, a ratio of 1000:1. The logarithm of 1000 is 3, so the relationship is 3 bels or 30 decibelsone watt is 30 decibels
louder than one milliwatt.
In addition to reducing power relationships to manageable numbers, decibels also approximately coincide with the way we hear sounds. Human hearing is also logarithmic, which
means that something twice as loud to the human ear does not involve twice the power.
For most people, for one sound to appear twice as loud as another, it must have ten times
the power. Expressed as dB, this change is an increase in a level of three dB because the
logarithm of 10 is 0.3, so the relationship is 0.3 bels or 3 decibels.
Engineers also use the decibel to compare sound pressures and voltages. Because sound
pressures represent power, the math is the same as for electrical power. But for voltages,
the decibel relationship is different. Power varies as the square of the voltage, so a doubling in voltage results in four times more power (four is two squared). With logarithms,
increasing a quantity by the power of two doubles a logarithm. Consequently, a doubling
of voltage represents not a 3 dB change but one of 6 dB.
Most commonly youll see dB used to describe signal-to-noise ratios and frequency
response tolerances. The unit is apt in these circumstances because it is used to reflect
relationships between units. Sometimes, however, people express loudness or signal levels
as a given number of dB. This usage is incorrect and meaningless because it lacks a reference value. When the reference unit is understood or specified, however, dB measurements are useful.
Analog Audio
Any unit may be used as a reference for measurements expressed in dB. Several of these
have common abbreviations, as listed in Table 25.1.
TA B L E 2 5 . 1
Reference Units in dB Measurements
Abbreviation
Reference Unit
0 dBj
1 millivolt
0 dBv
1 volt
0 dBm
1 milliwatt
0 dBk
1 kilowatt
In sound systems, the dBm system is most common. The electronic industry adopted (in
May, 1939) a power level of one milliwatt in a 600 ohm circuit as the standard reference
for zero dBm.
Youll also encounter references to volume units (VU), as measured by the classic VU
meter. A true VU meter is strictly defined, and the zero it indicates reflects a level 4 dB
above 0 dBm. The meters on tape and cassette recorders have VU designations but are
not, strictly speaking, VU meters. They are usually referenced to a specific recording level
on the tape. Similarly, meters on radio transmitters are often calibrated in VU, but zero
corresponds not to the input line level but the output modulation level0 VU usually
means 100-percent modulation.
Distortion
No one is perfect, and neither are audio amplifiers. All analog audio amplifiers add subtle
defects to the sound called distortion. In effect, distortion adds unwanted signalsthe
defectsto the desired audio. The most common way to express distortion is the ratio
between the unwanted and wanted signals expressed as a percentage. In modern low-level
circuits, the level of added distortion is vanishingly small, hundredths of a percent or less,
and is often lost in the noise polluting the signal. Only if your hearing is particularly acute
might you be able to detect the additions.
In power amplifiers, the circuits that produce the high-level signals used to operate nonpowered loudspeakers, the addition of distortion can rise to levels that are not only
noticeable but also objectionable, from one-tenth to several percent. Power amplifier distortion also tends to increase as the level increasesas you turn the volume up.
Better amplifiers produce less distortion. This basic fact has a direct repercussion when it
comes to getting the best sound quality from a computer. The soundboards typically
found in computers produce a lot of distortion. You can often get appreciably better audio
quality by plugging your stereo system (which is designed for low distortion even at high
887
888
power levels) into the low-level outputs of your soundboard so that the audio signal avoids
the soundboards own power amplifier. In fact, better-quality soundboards often lack
power amplifiers in the recognition that any included circuitry is apt to be of lower sound
quality due to the restrictions of installing them on a circuit board with limited power
supply capacity.
Digital Audio
Computers, of course, use digital signals, as do many modern stereo components such as
Compact Disc players and Digital Audio Tape systems. Once a sound signal has been
translated from analog into digital form, it becomes just another form of data that your
computer can store or compute upon. Digital technology adds new terms to the audio
vocabulary and raises new concerns.
Digital recording of sound turns music into numbers. That is, a sampling circuit examines
audio waveforms thousands of times every second and assigns a numerical value to the
strength of the sound every time it looks; it then records the numbers. To reproduce the
music or noise, a computers sound system works backward. It takes the recorded numbers
and regenerates the corresponding signal strength at intervals exactly corresponding to
those at which it examined the original signal. The result is a near-exact duplication of the
original audio.
The digital recording process involves several arbitrary variables. The two most important
are the rate at which the original audio signal is examinedcalled the sampling rateand
the numeric code assigned to each value sampled. The code is digital and is defined as a
given number of bits, the bit-depth or resolution of the system. The quality of sound reproduction is determined primarily by the values chosen for these variables.
Sampling Rate
The sampling rate limits the frequency response of a digital recording system. The highest frequency that can be recorded and reproduced digitally is half the sampling frequency.
Why? Start by taking the worst case, when the sampling frequency and the frequency
being sampled are the same. The sample would then occur at exactly the same place in the
audio wave with each sample. The numbers in the digital code would be the same, and
when reproduced the system would not produce a tone but a constant, unvarying voltagedirect current. The sampling frequency thus acts like an audio stroboscope. As the
frequency being sampled goes up past half the sampling frequency, the tone reconstructed
from the digital code actually goes down, reaching zero at the sampling frequency. Go
even higher and the regenerated tone starts increasing againfrom zero.
This top frequency that can be sampled (that is, half the sampling frequency) is often
called the Nyquist frequency, after Harry Nyquist, a Swedish-born American scientist
Digital Audio
working at Bell Labs who first published the explanation of the limit in 1928. Higher frequencies become ambiguous and can be confused with lower frequency values, producing
distortion. To prevent problems, frequencies higher than half the sampling frequency
must be eliminatedfiltered outbefore they are digitally sampled. Because frequencies
beyond the Nyquist frequency masquerade as lower frequencies when they get regenerated, the regenerated tones are said to alias, and filters that remove the high frequencies
before sampling are called antialiasing filters.
Because no audio filter is perfect, most digital audio systems have antialiasing filters with
cutoff frequencies somewhat lower than the Nyquist frequency. The Compact Disc digital
audio system is designed to record sounds with frequencies up to about 15KHz, and it
uses a sampling rate of 44.1KHz and a Nyquist frequency of 20.05Hz. Table 25.2 lists the
sampling rates in common use in a variety of applications.
TA B L E 2 5 . 2
Common Digital Sampling Rates
Rate (Hz)
Application
5563.6
Apple Macintosh, lowest quality
7418.3
Apple Macintosh, low quality
8000
Telephone standard
8012.8
NeXT workstations
11,025
PC, low quality (1/4th CD rate)
11,127.3
Apple Macintosh, medium quality
16,000
G.722 compression standard
18,900
CD-ROM/XA long-play standard
22,050
PC, medium quality (1/2 CD rate)
22,254.5
Basic Apple Macintosh rate
32,000
Digital radio, NICAM, long-play DAT, HDTV
37,800
CD-ROM/XA higher-quality standard
44,056
Professional video systems
44,100
Basic CD standard
48,000
DVD, Audio Codec 97, Professional audio recording
96,000
DVD at highest audio quality
The odd numbers used by some of the standards are often less arbitrary than they look.
For example, the 22,254.5454 Hz rate used by Apple Macintosh system matches the horizontal line rate of the video display of the original 128K Macintosh computer system.
889
890
For Mac people, thats a convenient number. The 44,056 rate used by some professional
video systems is designed to better match the sampling rate to the video frame rate.
Resolution
The number of bits in a digital code or bit-depth determines the number of discrete values
it can record. For example, an eight-bit digital code can represent 256 distinct objects, be
they numbers or sound levels. A recording system that uses an eight-bit code can therefore record 256 distinct values or steps in sound levels. Unfortunately, music and sounds
vary smoothly rather than in discrete steps. The difference between the digital steps and
the smooth audio value is heard as distortion. This distortion also adds to the noise in the
sound recording system. Minimizing distortion and noise means using more steps. Highquality sound systemsthat is, CD-quality soundrequire a minimum of a 16-bit code.
High-quality systems use 20- or 24-bit codes.
Bandwidth
Sampling rate and resolution determine the amount of data produced during the digitization process, which in turn determines the amount that must be recorded. In addition, full
stereo recording doubles the data needed because two separate information channels are
required. The 44.1KHz sampling frequency and 16-bit digital code of stereo CD audio
result in the need to process and record about 150,000 bits of data every second, about
9MB per minute.
For full CD compatibility, most newer soundboards have the capability to digitize at the
CD level. Intels Audio Codec 97 specification requires a bit more, a 48K sampling rate,
and undoubtedly stereophiles will embrace the extraordinarily high optional 96K sampling
rate allowed by the DVD standard. For most computer operations, however, less can be
betterless quality means less data to save in files and ship across the Internet. The relatively low quality of loudspeakers attached to computers, the ambient background noise in
offices, and the noise the computer and its fan and disks make themselves make the
nuances of top-quality sound inaudible anyways.
To save disk space and processing time, computer sound software often gives you the
option of using less resource-intensive values for sampling rate and bit-depth. Most computer sound systems support 22 and 11KHz sampling; some offer other intermediate values, such as 8, 16, or 32KHz. You can trim your data needs in half simply by making
system sounds monophonic instead of stereo. (Note that such savings are not quite so
straightforward once audio is compressed, such as when using MP3.)
If you are making original recordings of sounds and music, you will want to use as high a
rate as is consistent with your computers resources. Often the application will dictate your
format. For example, if you want to use your CD-R drive to master audio CDs, youll
Digital Audio
need to use the standard CD format, stereo 16-bit quantization at a 44.1KHz sampling
rate. On the other hand, the best tradeoff between quality and bandwidth for Internetbound audio is 11KHz sampling with 8-bit quantization. If you have an eye to the future
and an ear of gold, go for 96KHz sampling (if your computer allows it).
Note that the format of the data sets a limit on quality without determining the actual
quality of what you will hear. In other words, you cannot do better than the quality level
you set through choice of bit-depth and sampling rate. The shortcomings of practical
hardware, particularly inexpensive sound systems and loudspeakers, dictate that the quality
of the sound that actually makes it into the air will be less realistic than the digital format
may allow.
Transmission
Moving digital audio around is the same as sending any digital signal. You can move files
through any convenient port.
You can usually move CD-quality audio through a 10Base-T network with no problem.
For example, you can put WAV files on a server and play them back smoothly on any
computer connected to the network. That is, unless you put your computer to work on
other jobs that are extremely resource intensive or your network bears heavy traffic.
USB provides a quick and easy connection and is coming into use for digital recording
and editing. The bandwidth of USB is, however, constrained. After allowing for handshaking and other transmission overhead, you can expect to route about six CD-quality
digital audio signals simultaneously through a USB 1.0 connection.
The Web provides a big challenge for audio. Real-time audio playback across the Web is
inevitably a big compromise, one that relies on heavy compression that restricts bandwidth, which in turn limits frequency response and guarantees a high noise level.
If you dont need real-time audio playbackif you are collecting music rather than listening to ityou can move any kind of digital audio through any channel. Ordinary telephone connections make the time involved prodigious, especially if you want to move raw
CD-quality audio for an album collection. Compressing audio files trims the time to
transmit them through such connections, which is why most music now moves through
the Internet in compressed form using aurally friendly algorithms such as MP3 (discussed
next) and newer, related systems.
Compression
The Internet is not the only place where the size of digital audio files becomes oppressive.
At about 10MB per minute, audio files quickly grow huge. Archiving more than a few
sound bites quickly becomes expensive in terms of disk space.
891
892
To squeeze more sound into a given amount of storage, digital audio can be compressed
like any other data. Actually, audio lends itself to compression. The more efficient algorithms take into account the special character of digital sound. The best rely on psychoacoustic principles, how people actually hear sound. They discard inaudible information,
not wasting space on what they cannot hear.
The algorithms for compressing and decompressing digital audio are called codecs, short
for coder/decoder. Several have become popular for different applications. Windows
includes several, invisibly selecting the proper one when necessary to play back an audio
file. When recording, your software will prompt you to select the codec to use if a selection is available.
The Internet has brought other compression systems to the forefront. Although nothing
inherent in the Web requires them, because they reduce the size of audio files (and hence
their transmission time), they have made audio distribution through the Web practical.
More than that, high-quality compressed audio and the Web are revolutionizing how
music gets distributed.
As excitement about the Web was reaching its peak in the year 2000, developing a new
audio compression system spurred more investment than perhaps any other area of Web
development. Every promoter had a new and better system that he hoped to make a new
industry standard. Although most of these efforts claimed high degrees of compression (and
thus small file sizes), the chief ingredient in each was rights management, the ability to control duplication of music files to protect publishers copyrights. Some systems (such as
RealAudio) allow you to listen from the Web without the ability to save music. Others only
let you play downloaded files on a single system after youve paid for them. All these rightsmanagement systems are based on digital encryption. The Digital Millennium Copyright
Act makes breaking these codes or even publishing how to break the codes illegal.
The most important open standardand the one that inspired the newer compression
systemsis MP3, shorthand for MPEG, Level 3. (It does not stand for MPEG-3there
is no such standard, at least not yet). Although usually regarded as a video standard, the
MPEG standards discussed in Chapter 24, Display Systems, also describe the audio that
accompanies the moving images. The applications of MPEG audio are widespreadits
compression system is used by Digital Compact Cassettes, digital broadcasting experiments, and the DVD.
MPEG audio is not one but a family of audio coding schemes based on the human perception of sound. The basic design has three layers that translate directly into sound quality. The layers, numbered 1 through 3, form a hierarchy of increasing complexity that
yield better quality at the same bit-rate. Each layer is built on the previous one and incorporates the ability to decode signals coded under the lower layers. Table 25.3 summarizes
the MPEG audio layers.
Digital Audio
TA B L E 2 5 . 3
MPEG Layers and Bit-Rates Compared
Layer
Allowed Range
Target or
Optimum
Sample Application
32 to 448Kbps
192Kbps
Digital Compact Cassette
32 to 384Kbps
128Kbps
MUSICAM (Broadcasting)
32 to 320Kbps
64Kbps
DVD and Internet sound
As the layer number increases, the encoding becomes more complex. The result is a
greater amount of compression. Because greater compression requires more processing,
there is apt to be more latency (signal delay) as the layer number increases.
The layer number does not affect perceived sound quality. All layers permit sampling frequencies of 32, 44.1, or 48Kbps. No matter the layer, the output quality is dependent on
the bit-rate allowedthe higher the bit-rate, the higher the quality. The different standards allow higher quality to be maintained at lower bit-rates. At their target bit-rates, all
three layers deliver sound quality approaching that of CDs.
Unlike other standards, MPEG does not define compression algorithms. Instead, the layers provide standards for the data output rather than how that output is achieved. This
descriptive approach allows developers to improve the quality of the algorithms as the
technology and their discoveries permit. Header information describes the level and
methodology of the compression used in the data that follows.
MPEG is asymmetrical in that it is designed with a complex encoder and a relatively simple decoder. Ordinarily you will decode files. Only the producer or distributor of MPEG
software needs an encoder. The encoding process does not need to (and often does not)
take place in real time. All layers use a polyphase filter bank with 32 sub-bands. Layer 3
also adds a modified discrete cosine transform (MDCT) that help increase its frequency
resolution.
MP3 takes advantage of the high compression afforded under the MPEG audio standard
and uses it as the basis for a file system, which serves as a basis for todays MP3 hardware.
The advantage of MP3 is simply compression. It squeezes audio files into about onetwelfth the space raw digital audio data would require. As a result, music that would nominally require a 50MB file under the WAV format only takes about 4MB. Smaller files
means less transmission time so that cuts and entire albums can reasonably be sent across
the Internet. This also allows a substantial length of music (an hour or more) to be
encoded into solid-state memory and carried about in a no-moving-parts player.
Better still, by squeezing the size of the MP3 file, the data rate required for playing back a
file in real time can be similarly reduced. Instead of requiring approximately 1.2Mbps to
893
894
move two CD-quality audio channels, MP3 files need only 64Kbps for near-CD-quality
playback. Although thats not slow enough for real-time playback through a 56K modem
(remember, MP3 files are already compressed so modem-based compression cannot
appreciably speed up transfer rates), real-time playback is possible with ISDN terminal
adapters (with a good 128K connection and fast server), cable modems, and ADSL connections. With light network traffic and a 56K modem, you can expect to download an
MP3 file from the Web in two to four times its actual playing time.
Because MP3 is part of the MPEG-1 standard, it accommodates only stereo. MPEG-2 is
designed for surround sound using up to eight channels. In the works is a new level of
compression, often referred to as MP4, which will extend MP3-like compression to surround sound systems.
The tradeoff with gaining the compactness of MP3 files is data processing. Decoding the
audio data from an MP3 file requires a substantial amount of microprocessor power, so
you cant use a vintage computer you stuck on your shelf a decade ago as an audio component. Machines more than a couple of generations old cant process MP3 in real time.
Typically, youll need at least a 100MHz Pentium microprocessor, although a 300MHz
Pentium II system make a great audio-playback (and MP3-encoding) computer. In other
words, save your old computer for your stereo system.
You can recognize MP3 files by their filename extension of MP3. Note that some encoding programs that generate MP3 files allow you to choose the level of processing
typically Level 1, 2, or 3. These selections correspond to the levels under the MPEG standard and represent higher degrees of compression (smaller files and lower data rates) as
the numbers ascend, but only Level 3 creates true MP3 files. Nearly all MP3 decoders
will handle whatever level you choose, so go for the lower numbers if you want the
highest quality and dont mind giving up more disk space.
Looped Sound
When you want to include sound as a background for an image that your audience may
linger over for a long but predetermined amount of time (for example, to add background
sounds to a Web page), you can minimize the storage and transmission requirements for
the audio by using looped sound. As the name implies, the sound forms an endless loop, the
end spliced back to the beginning. The loop can be as short as a heartbeat or as long as
several musical bars. The loop simply repeats as long as the viewer lingers over the Web
page. The loop only requires as much audio data as it takes to code a single pass, no matter how long it plays.
No rule says that the splice between the end and beginning of the loop must be inconspicuous, but if you want to avoid the nature of the loop becoming apparent and distracting, it
should be. When looping music, you should place the splice so that the rhythm is continuous and regular. With random sounds, you should match levels and timbre at the splice.
Digital Audio
Most computer sound editors will allow you to finely adjust the splice. Most Internet
sound systems support looped sounds.
Internet Sound Systems

Go online and youll be confronted with a strange menagerie of acronyms describing
sound systems promising everything from real-time delivery of rock concerts, background
music from remote radio stations a zillion miles away, and murky audio effects percolating
in the background as you browse past pages and pages of uninteresting Web sites. All this
stuff is packaged as digital audio of some kind; otherwise, it could never traverse the
extent of the Internet. Rather than straight digital audio, however, it is processed and
compressed into something that fits into an amazingly small bandwidth. Then, to get it to
you, it must latch on to its own protocol. Its amazing that anything gets through at all,
and, in truth, some days (and connections) are less amazing than others.
The biggest hardware obstacle to better Internet sound is bandwidth. Moving audio digitally consumes a huge amount of bandwidth, and todays typical communications hardware
is simply not up to the chore. A conventional telephone conversation with a frequency
response of 300 to 3000 hertzhardly hi-figets digitized by the telephone system into a
signal requiring a bandwidth of 64,000 bits per second. That low-fidelity data is a true
challenge to cram through a modem that has but a 28,800 or even a 33,400 bits per second data rate. As a result, all Internet sound systems start with data compression of some
kind to avoid the hardware-imposed bandwidth limits.
The Internet poses another problem for audio systemsthe Web environment itself is
rather inhospitable to audio signals. From its inception, the Net was developed as an asynchronous packet-switched network. Its primary protocol, TCP, was not designed for the
delivery of time-critical isosynchronous data such as live audio. When youre downloading
a file (or a Web page, which is essentially a file as well), it doesnt matter whether a packet
gets delayed, but a late packet in an audio stream is less than worthlessits an interruption that can ruin whatever youre listening to. Some Internet sound systems abandon
TCP for audio transfer and use the UDP protocol instead, which can complicate matters
with systems and firewalls designed expressly for TCP. Other Internet sound systems rely
on TCP, citing that it ensures top audio quality in transferred files.
Several streaming audio players are available for download from the Web. The three most
popular are the audio side of Apples QuickTime (www.apple.com/quicktime/), RealAudio
from Real.com (wwww.real.com), and Microsofts Windows Media Player (included with current versions of Windows and available from www.microsoft.com/windows/windowsmedia/).
Multichannel Sound
When sound recording first became possible, people were so thrilled to hear anything at
all that small issues such as sound quality were unimportant to them. A tiny voice against
a scratchy background sounded as real as a loving coo whispered in their ears. As new
895
896
technologies such as electronic recording processes brought more life-like qualities to

reproduced sound, people discovered that recordings had a missing dimension. In real life,
they could hear depth, but the single horn on top of the old Victrola compressed the
sound down to a single point.
Trying to recover the depth of sound recording has occupied the minds of engineers and
researchers almost since the advent of electronic recording in the 1920s. Their aim has
been to reproduce the entire listening experience so that you dont just hear the music but
feel like you are immersed in it, sitting in a concert hall rather than your living room.
Stereo
The most natural improvement was stereophonic recording. The idea was simple, even
intuitive. Scientists knew that people were able to localize sound because they had two
ears. At the time, they didnt know exactly what allowed human hearing to determine
locations, but they found that reproducing music through two-speaker systems with separately recorded signals created a miraculous effect: Instead of hearing sound from one
speaker or the other, the ears of the listeners were fooled into hearing sounds coming
from the entire space between the speakers.
Binaural
Closely related to stereo is binaural recording. As with stereo, binaural uses two recording
channels, but binaural requires a special recording method for creating those signals that
uses tiny microphones placed inside the ears of a dummy head. Playback, too, differs from
ordinary stereo. In its purest form, a binaural system requires headphones instead of loudspeakers. In effect, the binaural system records the sounds that reach the dummys ears
directly into your ears.
The difference between stereo and binaural is astounding. Unlike simple stereo systems,
binaural creates a convincing illusion of three-dimensional space. Listen to a conventional
stereo recording on headphones, and all the sounds appear to emanate from inside your
head. The only spread is between your ears. With a binaural recording, the sound appears
to come from all around, outside your head, surrounding you. You are convincingly transported to the place and position of the dummy head used in making the original recording.
Without headphones, the binaural illusion disappears. To most people, a binaural recording played through loudspeakers sounds like a conventional stereophonic recording.
The development of binaural recording paralleled conventional stereo. Although initially
more a laboratory curiosity, in the 1960s and early 1970s record companies released a few
commercial binaural discs and radio stations transmitted a few binaural broadcasts of classical music and, rarely, rock concerts. Unfortunately, the artificial head recording technique
was incompatible with the multitrack studio tapes used for the vast majority of popular and
rock discs, so the technology never transcended its status as a laboratory curiosity.
Digital Audio
Quadraphonic
An abortive attempt at creating a three-dimensional sound stage appeared in the form of
quadraphonic recording in the early 1970s. The idea was simple: To create listening space
around and behind the listener, two recording channels and speakers dedicated to them
were added behind the listeneressentially it was stereo squared.
The problem with quad was not the concept but the available technology for broadcasting
and recording discs. At the time, quad was invariably a compromise. Because of the limitations of analog technology, engineers were invariably forced to squeeze in four channels
where only two were meant to fit. The most common method involved a sacrifice in
channel separation. To achieve compatibility with conventional stereo, the engineers combined the front and back information for each of the two stereo channels. They then piggybacked difference informationthe back channel minus the front channelon top of
the front-plus-back signal using a form of phase modulation (quadrature modulation,
which, despite the similarity of names, is not related to the quadraphonic concept). These
systems provided a front-to-back channel separation of about six decibelsmuch less than
the 20 to 30 dB of separation for the front channels but sufficient (or so the engineers
said) for providing the illusion of depth.
Only tape systems could provide four full-bandwidth and completely separate audio channels. At the time, however, the only legitimate tape systems with quality suitable for music
recording used cumbersome open-reel tape. The audio cassette was still in its infancy.
Although the Compact Disc would have permitted true quadraphonic recording with four
equal channels (although with a halving of playing time), it came too late. By the time of
the introduction of the CD, consumer-level quad was long a dead issue. It had never made
it into Disco let alone the 1980s. Worse, quad bore the stigma of unfulfilled promises along
with skepticism about its origins. Many people suspected the stereo industry introduced
and promoted quad solely to sell more stereo equipment at a time when hardware sales
were flagging. Because of its bad name, the advent of technology capable of handling quad
was not sufficient to resurrect it, and modern systems carefully avoided any association with
it. Note that no new technology professing to be surround sound uses four speakers.
3D Sound
The notion that two channels of sound are all that you need for two ears is hard to shake.
Somehow with just two ears most people are able to perceive sound in three dimensions
you can tell not only whether a sound source is left or right of you but also whether it is
above, below, or anywhere else. Two ears shouldnt be sufficient for 3D, but (obviously)
they are. Binaural recording and playback shouldnt work.
As early as the first studies of stereo sound, scientists have puzzled over this anomaly.
They discovered that two ears alone arent enough. The ears had to be coupled to a powerful computerthe human brain. The brain exploited subtle differences in the signals
897
898
received in each ear, differences caused by the odd shape of the human ear (in some people odder than others), and extracted enough information not only for depth but height
and everything else.
Once they figured out how people could deduce depth from the signals from two ears,
engineers went to work reversing the process. They figured if they altered the signal
before it was played back, they could add cues that would fool listeners into thinking they
heard sound in three dimensions when played through two speakers.
Surround Sound
Although surround sound is a relatively new innovation for home listening, it actually predates two-channel stereo by decades. The first commercial presentation of surround
sound came with Walt Disneys animated feature film, Fantasia, released in 1941 with a
six-channel soundtrack. Although not known then as surround, similar multichannel
formats became popular in cinematic productions during the 1950s as movie producers
attempted to fight the defection of their audience to the small screens of television sets.
They stretched their already-giant movie screens with Cinerama and Cinemascope and
filled the auditoriums with six or more sound channels. Although these systems were formally classed as stereo, industry insiders began to refer to the speakers in the sides and
rear of the auditoriums as surrounds, and soon the technology became surround sound.
Although surround sound does not inherently imply any specific number of channels,
the most popular format made four the basic standard. In 1976, Dolby Laboratories introduced Dolby optical stereo sound, a system that used noise reduction to coax high-quality
sound from optical sound tracks. Until then, most movies with surround sound used magnetic soundtracks, a technology that adds to the cost and complication of printing films.
Optical soundtracks are printed in the same operation as the images. Fitting four channels
into the optical soundtrack required matrixing, the same technique that put quadraphonic
sound on vinyl phonograph discs.
The difference between this four-channel form of surround sound and quad is the
arrangement of channels. Quad uses four equal speakers in the corners of the listening
spacea square. Surround uses a diamond arrangement with two primary speakersone
left, one righta center speaker, and a rear speaker. Not only is this arrangement effective
in producing an illusion of an all-encompassing sound field, it is also more amenable to
the matrixing technique that combines four channels into two. Videocassettes labeled
Dolby Surround use this technology.
Dolby Pro Logic adds signal steering to the basic surround sound arrangement. That is, it is
able to emphasize (or exaggerate) the surround sound effect by selectively altering the balance between channels in response to coding in the original two-channel signal. Both
Dolby Surround and Dolby Pro Logic are analog technologies, although the latter uses
digital logic to control the balance between channels.
Digital Audio
In discussing surround sound, the arrangement of speakers is often abbreviated in what

looks like a fractiontwo numbers separated by a slash. The first number represents the
speakers in front of the listener; the latter number, the speakers behind the listener. Table
25.4 lists the common surround sound configurations.
TA B L E 2 5 . 4
Surround Sound Channel Configurations
Common
Designation Name
Speakers
Used
Front
Left
Front
Mid
Left
Front
Front
Center
Mid
Right
Front
Right
Rear
Left
Rear Rear
Center Right
1/0
Mono
2/0
Stereo
3/0
Centerchannel
stereo
2/1
Threechannel
surround
2/2
Quadraphonic
3/2
Standard
surround
5/2
Enhanced
surround
Further complicating the designations is a decimal component sometimes listed in the

count of channels (for example, 5.1 or 7.1). The decimal indicates an additional sound
source that doesnt quite reach to being a full-range channel because of limited bandwidth. Sometimes this extra channel is termed the low frequency effects (LFE) channel. Its
frequency range is limited to that of nondirectional low frequencies (below 100 to 150
Hz), as would be used to power a subwoofer. Nearly all theater-style surround sound systems use an LFE channel to accentuate explosions and other impacts so that their low-frequency components might be felt as readily as they are heard.
AC-3
More commonly known as Dolby Digital (Dolby Labs holds patents on the perceptual
encoding system it uses), AC-3 is a 5.1 channel sound system. In other words, it uses
five full-bandwidth (20 Hz to 20KHz) channels along with one reduced-bandwidth
899
900
(20 Hz to 120 Hz) channel dedicated to low frequency effects. Speakers for a fullfledged AC-3 system are arrayed in the 3/2 configuration with three in front (left,
center, and right) and two in the rear (left and right). The subwoofer can be put in any
convenient location. AC-3 also allows for other configurations, including plain mono
and stereo.
AC-3 is the standard sound format of movies distributed on DVD, having been approved
for that purpose in December, 1995. In the DVD system, all six channels of AC-3 are discrete, separately recorded and isolated from the others. The full-bandwidth channels are
encoded with a sampling rate of 48KHz and a depth of 16 bits.
In operation, AC-3 compresses the raw audio data encompassing all the channels down to
a bit-rate that can range from 64Kbps to 448Kbps. Typically, in stereo the bit-rate will be
192Kbps. In full 5.1 channel configuration, the bit-rate runs about 384Kbps.
AC-3 is the standard for sound on NTSC-based DVD-Video discs. In addition, it has also
been adopted as the sound system for the American digital television transmission standard (ATSC). As such, the full text of the AC-3 standard is available from the ATSC Web
site at www.atsc.org.
DTS
Digital Theater Systems (DTS) began as a proprietary system for providing top-quality
digital surround sound to motion picture theaters. Using the same encoding and compression system as it applied to professional applications, DTS created an alternative audio
format for digital source material. The DTS system is an option for both audio and video
DCDs. Nearly all DVD players lack the facility for decoding DTS signals, although you
can usually connect an auxiliary processor to handle them.
As with most of the sound formats for DVD, the DTS system began as an enhancement
for theatrical films. The professional DTS system syncs a CD-based digital playback system with movie projectors using a timecode encoded along with conventional sound
tracks on the film itself. The timecode allows the digital audio source to be exactly locked
to the film even after editing or repair (cutting and splicing) of the film.
In DVD form, the DTS sound stream encodes up to 5.1 channels sampled at 48KHz with a
depth of up to 20 bits. It allows all standard channel combinations from mono to 3/2 surround with an LFE channel. The chief advantage of DTS is that it uses a lower compression
ratio (about 4 to 1) than does Dolby Digital. As a result, it holds the potential of delivering
higher-quality audio. The bit stream may have a data rate from 64Kbps to 1536Kbps.
SDDS
By acquiring Columbia Pictures, Sony became a major player in the motion picture
business. The company is active not only in software but has also developed its own
high-quality digital surround system called Sony Dynamic Digital Sound (SDDS). Based
Digital Audio
on the same compression system used by the zombie Minidisc system, SDDS can encode
up to eight channels in a 1280Kbps stream. Typically it samples 5.1 or 7.1 channels at
48KHz and a depth of 16 bits. SDDS is an optional DVD format.
Capturing Sounds
If youre not content using the sound given you by others and dont want to go to the trouble of synthesizing your own, you may look longingly at all the noises already available to
you that just arent in computer-compatible form. You might have a stack of old vinyl
albums you want to transfer to CD; you might want to move a cut from a CD into an MP3
file for more compact listening on your notebook computer while your travel; or you may
want to steal a sound from a television program or movie to spice up your computers otherwise drab responses (my system has Homer Simpson crying Doh! every time I hit the
wrong key). Equipped with a modern soundboard and suitable software, your computer can
readily capture any audio source you have for whatever purpose you want.
As with most computing matters, however, you have several choices in capturing sounds.
You can directly transfer the digital data from CD, you can digitize analog data from conventional audio sources, or you can make original recordings. Each technique has its own
methodology and produces different results.
CD-Ripping
Audio Compact Discs already contain digital audio information, so you might think that
capturing the data would be easy. After all, your computer already can tap its built-in CD
drive to get the data it needs to set up programs. Matters are not quite so simple, however. The format for audio CDs is substantially different from the Yellow Book standard
used by digital data CDs. The differences extend to formatting and error correction.
Simply put, your ears tolerate audio errors much better than digital systems tolerate errors
in code. Although a data error may be inaudible to you when you listen to a CD playing
back, the data stream may include errors that will send digital circuits into conniptions. If
your computer cannot detect the audio errors, they will be faithfully encoded as digital
data in the proper format but with the wrong value. Transient audio errors become a permanent part of the dataand the digitally encoded audio that you listen to.
The ability for a CD to deliver audio in digital form to your computer is called digital
audio extraction. The process requires special software that has come to be known as the
CD ripper. Similarly, the extraction process is often called CD-ripping.
CD drives differ widely in their ability to yield up pure audio data. Most early drives cannot faithfully extract audio data. They are unable to properly frame that audio data into
digital form, resulting in jitter. The aural result is distortion in the digital audio, most
often in the form of loud clicks that repeat throughout a transfer, sort of like someone has
thoroughly rubbed 80-grit sandpaper over your favorite vinyl album.
901
902
Note that the quality of digital audio extraction is primarily dependent on the CD drive
that you use. Older ATAPI-based (IDE) CD drives were poor at digital audio extraction.
Music ripped using them contained numerous objectionable clicks and pops, if you could
rip from them at all. Todays drives do a much better job. The only time youre likely to
encounter clicks and pops is when ripping copy-protected commercial CDs. The noise is
intentional.
CD-ripping is fast. With a fast microprocessor your computer will likely rip through CDs
at the fastest speed at which your CD drive is capable. In other words, you may be able to
rip CD files in from one-sixth to one-thirty-second the time it would take the cut to play
back.
Analog Capture
The alternative to ripping digital audio is to capture audio in analog form and convert it
to digital. All soundboards have analog-to-digital converter circuits that carry out this
process, which is nothing more than ordinary audio sampling. You only need supply the
analog source to your soundboard and specify the parameters of the samplebit-rate and
bit-depth.
Internal CD
If your source material is CD, you can use the internal CD drive of your computer for
analog capture even if it is not capable of digital audio extraction. Use the mixer software
supplied with your soundboard to specify the CD drive as the source and then mute or
switch off other sources. Your soundboard will then sample only the audio from your CD.
The quality you get from analog capture will be lower than with direct extraction. Your
samples will pick up all noise and distortion present in the analog circuitry of your soundboard, which can be substantial. For example, while the noise floor of CD audio is about
96 dB below peak level, going through the analog audio circuitry of both the CD drive
and soundboard may increase the noise so that it is only 60 to 70 dB below peak audio.
This level of noise is often not objectionable and can actually sound superior to audio
extracted from a marginal CD drive that is rife with clicks and pops. In fact, this noise
level is comparable with optimum FM radio reception. If you are a purist, you can remove
most of the residual noise with the digital noisereduction functions available in some
audio-editing programs.
Another disadvantage of analog capture of CD audio is speed. Analog capture is a realtime process. Sampling a cut from a CD will require precisely the playback time of the
cut. In addition, you may need to edit what youve captured, trimming the silence from
the beginning and end of the cut.
Digital Audio
External Components
You can also capture sounds from external components using the Auxiliary Input thats available on most (but not all) soundboards. The Auxiliary Input functions exactly like the input
to a cassette recorder. The audio signals you plug in there are routed through the mixer
function of the soundboard to its analog-to-digital converter to be sampled. Once the audio
signals are digitized, you can use the audio data exactly like that ripped from CD.
You can plug an external audio CD player directly into the Auxiliary Input of your soundboard. Better still, you can connect the recorder outputs of your receiver to the Auxiliary
Input of your soundboard and use your computer as if it were a cassette recorder, using
the receiver to select the external audio source you want to record.
The mixer software supplied with most soundboards also has a MIDI input that allows
you to translate MIDI filesboth those you make and those you obtain elsewhereinto
digital audio files that you can play back like any other music file. You can make recordings by selecting the MIDI source and using any sound-recording software. Note, however, that most soundboards use their own MIDI circuitry to synthesize the sounds, so the
capabilities and idiosyncrasies of your soundboard will determine the musical voices in the
MIDI recordings that you make.
Microphone Inputs
The worst way to try to capture audio from a CD is to shove a microphone in front of a
speaker. Not only does the quality suffer the indignity of conversion from digital to analog
format, but it also suffers the substantial shortcomings of both the speaker and the microphone. You might just as well listen to a friend play the CD over the telephone.
Using microphones is the only way to capture live sounds, be it your own speech or the
band infesting your garage that no amount of warfarin will deter. Most soundboards have
microphone inputs that allow you to plug directly in and start recording simply by selecting the microphone (or mic) source in your soundboards mixer program.
That said, you should be aware that most soundboard microphone inputs are designed
primarily for speech. Although they are able to accept a wide frequency range, they have a
high noise level because their circuits are immersed inside the noisy electronics of your
computer. You can obtain better quality recordings by using an external microphone preamplifier such as that in a mixing console (or board).
Note that most soundboard microphone inputs are stereophonic and use stereophonic
miniature jacks with three connectionstip, ring, and sleeve. A single monophonic microphone with a standard miniature plug (tip and sleeve) usually wont work with these
inputs. Adapters are readily available to let you use a single microphone with a stereo
input or to combine the signals from two microphones into a single plug to match your
soundboard. Modern microphones designed for computer applications are usually
equipped with stereo plugs to match the inputs of soundboards.
903
904
Legal Issues
Commercial recordingsin general that means all the CDs that youve boughtare protected by copyright laws. You should not post what youve captured or sampled on the
Internet. You should not use material in a production (for example, a slideshow, training
video, or simply as background sound) without the permission of the copyright holder.
You should never sell a copy of anything that youve sampled from a copyrighted source
such as a CD.
You can back up a CD that you have bought to an MP3 file for your personal use (for
example, to make an MP3 file that you can listen to on a portable MP3 player). Current
copyright law allows you to make a single backup of copyrighted software that you have
bought. Its no different than making a copy of a disk on a cassette to play in your car
(although some folks question the legality of making such copies, as long as you only
make one copy and do it for personal use only, you wont have to worry about the copyright police pounding on your door). On the other hand, it is both illegal and immoral to
copy tracks from discs that you have borrowed from a friend or library.
Hardware
The job of the audio circuitry in your computer is to set the air in motion, making sounds
that you can hear to alert you, to entertain you, and to amaze you. Computers have had
sound circuitry from the very beginning. But in the beginning of computers, as in the
beginning of life on earth, things were primitive, about the audio equivalent of amoebic
blobs bobbing along en masse in black swamps. From humble beginnings, however, computer sound systems have evolved to parity with the stuff in all but the best and most
esoteric stereo systems.
From the standpoint of a computer, sound is foreign stuff. Indeed, its something that happens to stuffairwhile the computer deals with the non-stuff of logical thoughts. Video
images are much more akin to computer electronicsat least the photons that you see are
electromagnetic. Sound is purely mechanical, and that makes the computers job of dealing with it tough. To make sound audible, it somehow has to do something mechanical. It
needs a transducer, a device that transmits energy from one system to anotherfrom the
electrical computer to the kinetic world sound.
Basic Sound System

Although the audio capabilities of some computers rival the best stereo systems, the least
common denominator among them is low indeed. The basic sound system that youre
assured of finding in all computers is exactly the same primitive design that IBM bolted
into its original Personal Computer.
Hardware
To be charitable, the basic computer sound system wasnt designed for high fidelity. In
fact, it was conceived as a beeper. Its goal was to generate pure, if harsh, tones to alert you
to events occurring in your computer (for example, the beep code of the BIOS). After all,
in 1981 computers didnt sing opera.
This basic sound system has three componentsa tone generator, an amplifier, and a
loudspeakerall of which must be called rudimentary because theres nothing lower on
the scale. When all worked together, you could make your computer beep as easily as typing Ctrl+G in DOS (this wont work in Windows). The frequency and amplitude of the
tone was predetermined by the designers of the first IBM Personal Computer. You were
lucky to get any noise at all, let alone a choice.
Clever programmers quickly discovered that they could easily alter the tone, even play
short ditties with their data. As programmers got clever, they found they could modulate
the primitive sound system and indeed make the computer sing. Considering the standard
equipment, you can make your computer sound surprisingly good just by adding the right
driver to Windows.
Tone Generator
The fundamental tone-generation circuit is the oscillator, the same as the clock that generates the operating frequency of your computers microprocessor. The only difference is
that the tone generator operates at lower frequencies, those within the range of human
hearing (once they are translated into sounds).
The first computers used one of the channels of their computers 8253 or 8254-2
timer/counter integrated circuit chips as the required oscillator. Modern computers integrate the same functions into their chipsets.
No matter the implementation, the circuits work the same. The timer develops a train of
pulses by turning a voltage on and off. The timing of these cycles determines the frequency of the tone the circuit produces.
The computer timer/counter chip starts with a crystal-controlled fixed frequency of
1.19MHz and divides it down into the audio range. A register in the timer chip stores a
16-bit divisor, by which value the timer reduces the oscillator frequency. Loading the
highest possible value into the divisor register (65,535) generates the lowest possible tone
the basic computer sound system can produce, about 18 Hz, low enough to strain the limits of normal hearing were the computers speaker able to reproduce it. Divisor values
above about 64 produce tones beyond the upper range of human hearing.
Because of the circuit design of the computer, these tones are produced as square waves,
which means they are not pure tones but are rich in overtones or harmonics. Musically
they sound harshexactly what the doctor (or engineer) ordered for warning tones.
905
906
In this basic operating mode, the dynamics of the signal are limited. The output of the
timer/oscillator chip is set at a constant levelthe standard digital signal levelso the
sound level produced by the speaker does not vary. All the sounds produced by the computers motherboard have the same level. Some tones generated by the computer timer
sound louder than others primarily because they are more obnoxious. They are made
from the exact right combination of frequencies to nag at the aesthetic parts of your brain.
Thats about all they were designed to do. Listen long enough and youll agree that the
computers designers succeeded beyond their wildest dreams at creating obnoxious sound.
Using a technique called pulse-width modulation, programmers discovered they could use
even this primitive control system to add dynamics to the sounds they generated. Pulsewidth modulation uses the duty cycle of a high-frequency signal coupled with a low-pass
filter to encode the loudness of an analog signal equivalent. The loudness of a sound corresponds to the length of a signal pulse of a high-frequency carrier wave. A soft sound
gets a brief pulse whereas loud sounds are full-strength square waves. The low-pass filter
eliminates the high carrier frequency from the signal and leaves a variable-strength audio
signal (the modulation).
Amplifier
The output of the tone-generator chip is too weak for sounding a speaker at a listenable
level. The chip simply cannot supply enough current. The standard way to boost signal
strength is with an amplifier. The basic computer sound system uses a simple operational
amplifier. Modern systems incorporate this circuitry into the basic motherboard chipset.
In any case, even with the boost, the signal amounts to only about 100 to 200 milliwatts,
not enough to shake a theater with THX.
The standard computer design also adds a low-pass filter and a current limiting resistor
between the driver and the loudspeaker. The low-pass filter eliminates frequencies higher
than normal hearing range (and some in the upper ranges that you probably can readily
hear). Computers often use higher, inaudible frequencies as pieces that can be put
together to make audible sounds, and the low pass filter prevents artifacts from these high
frequencies from leaking into the speaker. In other words, it smoothes things out.
A resistor (typically about 33 ohms) in series with the loudspeaker prevents the internal
loudspeaker of a computer from drawing too much current and overloading the driver circuit. A resistor also lowers the loudness of the speaker because it absorbs some power as
part of the current-limiting process. Although some circuit-tinkerers bypass this resistor
to make their computers louder, doing so risks damage to the driver circuit.
Loudspeaker
The actual noisemaker in the computers basic sound system is a small, two-to-three-inch
dynamic loudspeaker. To most computer designers, the internal loudspeaker is an obligatory headache. They have to put one somewhere inside the computer no matter how
Hardware
inconvenient. Usually the speaker gets added almost as an afterthought. After all, all you
have to do is hear it. You dont have to hear it well.
Because computer designers make no effort at optimizing the quality of the sound of the
basic speaker, it is usually unbaffled, meaning it is open to the air on both sides (which
ruins whatever ability it has to produce bass notes). Its small size and acoustics prevent it
from generating any appreciable sound at low frequencies, anyway. Its physical mounting
and design limit is high frequency range as well. Outside of replacing the loudspeaker
with something better, you cannot do anything to break through these limits.
In most computers, the speaker connects to the motherboard using a simple, short twowire twisted-pair cable. The motherboard usually, but not always, includes a four-pin
loudspeaker connector. In many computers, one of the pins is removed for keying the
polarity of the speaker connection. One matching hole in the speaker cable connection
often is blocked. Only two of the four pins of the motherboard connector are active, the
two at the ends of the connector. The center one or two pins arent connected to anything. Figure 25.1 shows this connection.
FIGURE 25.1
Basic computer
speaker connections.
Motherboard connector
Speaker connector
In most computers, the loudspeaker is electrically floating. That is, neither of its terminals
is connected to the chassis or any other computer wiring except for the short to which it is
soldered. When the speaker is electrically floating, the polarity of its connection is irrelevant, so the keying of the connection is essentially meaningless. In other words, if the
speaker connector on your computer is not keyedif all four pins are present in the
motherboard connector and none of the holes in the speaker connector are plugged
dont worry. The speaker will operate properly no matter how you plug it in.
Drivers
Other than beeping to indicate errors, the basic sound system in a computer has a life of
leisure. If left to its own devices, it would sit around mute all the while you run your computer. Applications can, however, take direct hardware control and use the basic sound
system for audible effects. For example, text-to-speech converters can use the basic sound
system with pulse-width modulation techniques to simulate human speech.
Under all versions of Windows, only the native tone-generating capabilities of the basic
sound system get used, again just to beep warnings. Microsoft and some others have developed speaker drivers that allow the built-in basic sound system to play more elaborate
907
908
noises, such as the Windows sound when the operating system starts. These drivers let
you play WAV files only. They do not synthesize sounds, so they cannot play MIDI files
(discussed later) nor do they work with most games.
Some business systems still lack high-quality sound (perhaps as a way for managers dissuading employees from playing computer games on company time). A speaker driver can
sidestep the issue, adding some more advanced capabilities to the tiny internal speaker.
The prototype of these drivers is called the Windows speaker driver and was developed by
Microsoft strictly for Windows 3.1. The speaker driver has not been updated for more
recent versions of Windows. In fact, the history of the speaker driver is even more checkeredit was included with the beta test versions of Windows 3.1 but not the release version. During the development of Windows 3.1, Microsoft found that this driver sometimes
misbehaved with some programs. To avoid support headaches, Microsoft elected not to
include the driver in the basic Windows 3.1 package. The driver is included in the
Microsoft Driver Library and Microsoft does license it to developers to include with their
products. It remains available from a number of sources, as are other, similar drivers.
These drivers will work under Windows 95 and some later versions, although Microsoft
offers no explicit support of such operation. The official Microsoft strategy is to require a
real sound system in your computer if you want to hear anything beyond simple beeps. If
youre used to simple beeps, these drivers sound amazingly convincing. They are not,
however, a substitute for a real sound system.
Sound Systems
Clearly the basic sound system in computers is inadequate for the needs of modern multimedia. Getting something better requires additional circuitry. Traditionally, all the
required electronics get packaged on a single expansion card termed a soundboard. Higher
quality audio has become such a necessity in modern computers that most new notebook
machines include all the circuitry of a soundboard on their motherboards or an audio riser
board (which simply extends the motherboard another level). Many new desktop computers and replacement motherboards also make all the functions of a soundboard an integral
part of their designs. The physical location of the circuits is irrelevant to their normal
operation. Electrically and logically, they are equivalent.
To cope with the needs of multimedia software and the demands of human hearing and
expectation, the sound system needs to carry out several audio-related functions using specific hardware features. Foremost is the conversion of digital sound data into the analog
form that speakers can shake into something that you can hear using a digital-to-analog
converter. In addition, most sound systems sample or record sounds for later playback
with a built-in analog-to-digital converter. They also create sounds of their own using
built-in synthesizers. Sound systems include mixer circuits as well that combine together
Hardware
audio from all the sources available to your computertypically a microphone, the output
of the sound systems digital-to-analog converter (which itself combines the synthesizer,
WAV files read from disk, and other digital sources), the analog audio output of your
computers CD player, and an auxiliary input from whatever audio source tickles your
imagination. Finally, the sound system includes an amplifier that takes this aural goulash
and raises it to ear-pleasing volume.
Soundboards may include additional functions, one of which is required by the various
multimedia computer specificationsa MIDI interface. This additional connection lets
you link your computer to electronic musical instrumentsfor example, allowing your
computer to serve as a sequencer or, going the other way, connecting a keyboard to control the sound systems synthesizer. Some makers of soundboards want their products to
act as single-slot multimedia upgrades, so they include CD drive interfaces on their
soundboards. With the trend for incorporating audio onto motherboards, however, MIDI
has been left behind. Few computers with built-in audio offer built-in MIDI as well. In
general, to integrate your computer with external musical keyboards and synthesizers,
youll have to add a dedicated MIDI interface on an expansion board.
Soundboards can be distinguished in several ways. The most important of these divisions
are the three Cs of soundboardscompatibility, connections, and quality. Compatibility
determines the software with which a given soundboard will work. The connections the
board supports determines what you can plug inusually MIDI and CD devices. Quality
influences how satisfied you will be with the results, essentially whether you will be
delighted or dismayed by your foray into multimedia.
Compatibility
Compatibility often is the more important, because if your software cant coax a whimper
from your sound system, you wont hear anything no matter what you plug in or how well
the circuitry on the board might be able to do its job. Compatibility issues arise at two
levels: hardware and software. More practically, you can regard these levels as DOS and
Windows compatibility (or games and Windows, if DOS is foreign to your nomenclature).
For the most part, compatibility refers to the synthesizer capabilities of a sound system.
For your games to make the proper noises, it must be able to tell the sound system exactly
what to do, when and whether to belch and boom. Your software needs to know which
ports to access the functions of the sound system. With todays sound systems, this compatibility is ensured through Windows and the driver software associated with your sound
system.
If you want to play old games that sidestep Windows, compatibility becomes an issue.
Most old games and other software require compliance with two basic de facto industry
standards: Ad Lib and SoundBlaster.
909
910
Beyond synthesis capabilities, the reigning standard for computer sound systems is Audio
Codec 97, a set of specifications published by Intel that documents an audio architecture
originally designed around two separate chipsa controller that handles all digital operations and an analog chip that turns the computer signals into audio. The two are connected by a five-wire serial connection termed the AC link.
The heart of the Audio Codec 97 design are two digital-to-analog converters (DAC)
capable of operating at a 48 kilohertz sampling rate to generate a pair of stereo audio
channels. The specification requires that the controller be able to simultaneously process
four signals: two inputs and two outputs. Each channel must be able to translate the signals to or from at least six sampling rates to the 48 kilohertz basic rate of the system.
These include 8.0, 11.025, 16.0, 22.05, 32.0, and 44.1 kilohertz. To maintain true hi-fi
quality, the specification requires a signal-to-noise ratio of 90 dB.
The DACs are fed by a mixer that accepts inputs from all the digital sources in the computer, including the synthesizer section of the Intel codec chipset. For inputs, the system
includes a pair of analog-to-digital converters that also operate at 48 kilohertz as well as
an optional third, matching channel dedicated as a microphone input. Nearly all computers sold today meet or exceed the requirements of this standard.
Control
Sound systems provide important control functions for music-making and audio playback.
The mixer circuitry in the sound system serves as a volume control for each of the various
sources that it handles.
An audio mixer combines several signals into onefor example, making the distinct signals of two instruments into a duet in one signal. Most audio mixers allow you to individually set the volume levels of each of the signals that they combine.
Mixers work by summing their input signals. Analog mixers work by adding together their
input voltages. For example, an analog mixer would combine a 0.3 volt signal with a 0.4
volt signal to produce a 0.7 volt signal. Digital mixers combine digital audio signals by
adding them together mathematically using their digital values. The results are the same
as in analog mixing, only the type of audio signal differs.
In your computer, however, the difference is significant. The sound system performs
analog mixing in real time in its onboard circuitry. Most computers use their microprocessors to do digital mixing, and they usually dont make the mix in real time. For
example, your computer may mix together the sounds from two files and let you later
play back the combination.
All sound systems incorporate mixer circuitry that lets you combine all the analog signals
the board works with. The resulting audio mixture goes to your speakers and the analogto-digital converter on the board so that you can record it into a file.
Hardware
To control the relative volume levels of the various mixer inputs, most soundboard-makers
provide mixer software that gives you an onscreen slider to adjust each input interactively.
Quality
In sound systems, quality wears more than one face. Every board has designed-in capabilities and, likewise, limits that control what the board might possibly do. Lurking beneath,
however, is the quality that a given board can actually produce. Dreams and reality being
as they are, most sound systems aspire higher than they perform. This difference is hardly
unexpected and would not be an issue were not the difference so great. A sound system
may have specifications that put it beyond Compact Disc quality and yet perform on par
with an ancient AM radio crackling away in a thunderstorm.
In general, most sound systems list their quality capabilities in the range of the digital
signals they manage. That is, a given sound system may support a range of sampling frequencies and bit-depths. Nearly any modern sound system worth considering for your
computer will list capabilities at least as good as the 44.1KHz sampling and 16-bit resolution of the CD medium. Newer boards should accommodate the 48KHz sampling of professional audio and the DVD system.
In terms of digital quality, the CD rates should be good for a flat frequency response from
DC to 15 kilohertz with a signal-to-noise ratio of about 96 decibels. Commercial sound
systems miss both of those marks.
The shortfalls arise not in the digital circuitryafter all, any change in a digital signal,
including those that would degrade sound quality, are errors, and no modern computer
should let errors arise in the data it handles. Rather, the analog circuitry on the sound system teases, tortures, and truncates the signals that travel through it. Both frequency
response and dynamic range suffer.
Most of the time, man-handling audio signals makes little difference. When sounds play
through typical computer loudspeakers, you probably wont hear the deficiencies. Most
inexpensive computer speakers (which means most computer speakers) are so bad that
they mask the signal shortfalls. Listen through headphones or through a good stereo system, however, and the problems become readily apparent.
Many computer sound systems shortchange you on frequency response. Certainly the sampling rate of a digital system automatically imposes a limit on the high frequencies that
the system can handle. At the other end of the sound spectrum, there should be no such
limitation. A digital system should easily be capable of encoding, storing, and reproducing not only the lowest frequencies that you can hear but also sounds lower than you
can hear and even levels of direct current. Less expensive sound systems do not dip so
low, however. In fact, many boards cannot process low-frequency sounds within
the range of human ears, often missing the fundamental frequencies of bass notes and
911
912
the stomach-wrenching rumbles of computer game special effects. These limits arise
in the analog circuitry on the board from the coupling of AC signals through the various amplifier stages.
Most common analog systems require coupling capacitors between amplifier stages (and
in the output) to block both supply voltages and direct current errors from interfering
with the amplified audio signal. The size of the capacitor (its rating in microfarads, a unit
of measure of electrostatic capacity) determines the low-frequency limit of the overall system. Larger capacitors are more expensive and harder to place on compact circuit boards,
so manufacturers of budget products are apt to skimp on capacitor size. Better audio circuits use larger capacitors that pass lower frequencies or direct-coupled designs that
totally eliminate their need (and push response down to DC).
The size of any capacitor in the amplifier output is particularly critical because, as the
power level increases, the required capacity to pass a given frequency increases.
Consequently, the first place the low-frequency limit of a sound system suffers usually is
in its power amplifier, the part that provides a high-level signal suitable for unpowered
speakers. It is not unusual for low-priced sound systems to cut off sounds below 150 Hz,
which means few bass frequencies get through. Because the low frequencies are not present as the speaker jacks of such sound systems, plugging in better amplified speakers or
even your stereo system will do nothing to ameliorate the low frequency deficiencies.
Line-level outputs, as opposed to speaker outputs, are more likely to preserve low frequencies, and they should be preferred for quality connections.
The other signal problem with inexpensive sound systems is noise. The noise level in any
digital system is constrained by the bit-depth, and in the typical 16-bit system, the noise
floor is pushed down below that of many stereo components. Listen critically to the
sounds from many sound systems and CD-ROM drives, however, and youll hear squeaks
and peeps akin to the conversations of Martians, as well as more mundane sounds such as
swooshes and hisses. Most of these sounds are simply extraneous signals intercepted by
the sound system, mixed with the sounds you want to hear and amplified to offend your
ears. Not only do these arcane sounds interfere with your listening enjoyment, but they
may also make their imprint on the sounds you digitize and store, forever preserving the
marginal quality of your poor choice of sound system for posterity. Most sound systems
keep the level of these noises below that which you can hear through inexpensive speakers, but listen to soft music from your CD drive, and you may find your peace shattered
by Martian madness.
Better sound systems incorporate shielding to minimize the pickup of extraneous noises.
For example, the circuit traces on the boards will be shielded by ground planes, extra layers
of circuit traces at ground potential covering the inner traces of multilayer printed circuit
boards. Only the best sound systems can digitize analog signals without raising the overall
noise level well above the 92 to 96 dB promised by good CD drives.
Hardware
If you plan on using a sound system for transferring analog audio to digital form (for
example, to back up your old vinyl phonograph records to CD), you will want a sound
system that guarantees low noise and low-frequency response extending down to 20 Hz.
Digital Audio Output Systems

To avoid the problems inherent in using low-level audio signals inside noisy digital
machines, the makers of computers and audio equipment are increasingly moving the sensitive analog circuitry out of the computer itself. For example, professional recording systems may put both analog input and output circuits in an external box connected through
a digital port. Consumer-level devices such as USB speakers and MP3 players adopt the
same strategy, although for different reasons.
The underlying idea is to eliminate the need for almost any analog circuitry inside your
computer and cut the interference from the digital signals, which may be a thousand times
louder than the audio, thus making crosstalk (the leakage of an undesired signal into the
desired signal) almost inevitable. The result is that the low-level noises you can hear
through your speakers when your hard disk rattles into action (as well as any strange
sounds that appear when your computer engages in heavy-duty processing) disappear.
Your computer becomes a truly hi-fi sound system.
USB Speakers
USB speakers look like ordinary loudspeakers, usually a pair of small desktop speakers or a
three-piece system with a subwoofer and two satellites. The only difference between USB
speakers and conventional speakers is the connection. As you might guess, the USB speakers use an all-digital USB link with your computer instead of an analog connection. The
necessary analog-to-digital conversion circuitry is built in to the speakers.
In theory, USB speakers can eliminate the need for a soundboard or sound system in your
computer, at least if you only want to play back sounds. In addition, the speaker-maker
can tailor the D-to-A converter and amplifier to exactly match the speaker. It also eliminates a confusing jack from the back of your computer and integrates all your computers
wiring into the USB system. Because the signals used by USB speakers remain digital at
all times, both inside your computer and through the cable that transports them into the
speakers, they are not subject to noise or the degradation that analog signals suffer. The
speakers reproduce the sounds encoded in your audio files in pure, unadulterated form.
In theory, USB speakers should sound better than ordinary analog speakers. But the sound
quality of any speaker system mostly depends on the transducers rather than the signals.
Small speakers cannot produce room-filling bass notes, for example.
USB speakers have shortcomings, too. Digital audio eats a big chunk from the available USB
bandwidthas a practical matter, once you account for overhead, the speaker connection
swallows up about one-third of a USB version 1.0 channels bandwidth, limiting the number
913
914
of other devices that can share the USB port. Load the port too heavily, and the sound is
likely to drop out for long, irritating gaps. With the introduction of the USB 2.0 interface
and its wide bandwidth, such problems are unlikely (but both speakers and your computers
USB port should follow the USB 2.0 standard).
Remember, too, the speaker connection is only one function of a sound system. USB speakers do nothing to assume the other sound system functions. For example, they provide no
microphone input. Although you could, in theory, link your microphone to your computer
with a USB connection, too, USB mics are as rare as unicorns or dot-com success stories.
Although USB speakers have strong theoretical advantages over conventional analog
speakers, in current applications the benefits accrue mostly to the computer-maker.
Designing a computer system that relies on USB speakers shifts the costs of the analog
circuitry to the speaker manufacturer.
On the other hand, using the high speed of USB 2.0, some multimedia developers are creating entire external sound systems for computers. These have the advantage of moving
the low-level audio circuitry out of the noisy system unit. In theory, external sound systems can delivery better overall audio quality.
MP3 Players
One of the truly revolutionary products of digital audio technology, the MP3 player is the
ultimate portable music machine. Because it has no moving parts, it holds two key advantages over other portable music systems. No discs or tapes to shake, rattle, or roll ensures
that jogging or even leaping hurdles wont cause your music to warble, skip, or entirely
stop. No motor to move any parts means that an MP3 player doesnt waste energy battling friction and spinning its wheels during every song you play. The MP3 player therefore needs less power for a given playing time.
At heart, a portable MP3 player is little more than a USB speaker with memory. It downloads digital code, typically through a USB port, and decodes the digits into analog format
that you can hear through headphones or speakers. The difference is that USB speakers
play back in real time, but the MP3 player remembers and plays on demand.
An MP3 player requires prodigious memory, enough to hold all the songs you want to
hear. Thats why the MP3 in the name is so important. Without MP3 compression, the
typical player would be able to store little more than a single song. The monotony would
make you want to cut any exercise short. With MP3 technology, however, you can put an
hour of music in the palm of your handor even a days worth, if you opt for a jukebox
player with a built-in hard disk drive.
MP3 players also incorporate a control system that, at an invisible level, allows the player
to manage its memory and keep tabs on the songs it stores. On the visible level, it lets you
organize your music and play the selections you want.
Hardware
Memory and compression determine how much music you can fit into any given player.
Most MP3 players have built-in memory, although many supplement their native endowments with memory cards or Memory Sticks. The minimum in any current player is
32MB of native RAM. At the typical rate of music downloaded from the Internet, 64 kilobits per second, thats enough for about an hour of music67 minutes. More memory
increases playing time commensurately; high data rates for better quality playback cuts
playback time. Table 25.5 is a matrix showing playing time for various combinations of
memory and quality.
TA B L E 2 5 . 5
MP3 Playing Times Versus Memory Capacity and Data Rate
Data Rate
Memory
64Kbps
128Kbps
256Kbps
320Kbps
16MB
33 min
17 min
8 min
7 min
32MB
1 hr 7 min
33 min
17 min
13 min
64MB
2 hr 13 min
1 hr 7 min
33 min
27 min
128MB
4 hr 27 min
2 hr 13 min
1 hr 7 min
53 min
256MB
8 hr 53 min
4 hr 27 min
2 hr 13 min
1 hr 47 min
512MB
17 hr 47 min
8 hr 53 min
4 hr 27 min
3 hr 33 min
1GB
35 hr 33 min
17 hr 47 min
8 hr 53 min
7 hr 7 min
Transducers
The bridge between the electronic world of audio (both analog and digital) and the
mechanical world of sound is the acoustic transducer. The microphone converts sound into
audio, and the loudspeaker converts audio into sound.
Microphones
All sound systems have microphone inputs to enable you to capture your voice into the
digital medium. You can use digital transcriptions of your voice to annotate reports,
spreadsheets, and other files or incorporate them into multimedia presentations. With a
suitable sound system, you can even connect high-quality microphones to your computer
and make digital recordings of music, edit them, and write them to CDs to play in your
stereo system.
The job of the microphone is simple: to translate changes in air pressure into voltage
changes. The accuracy of the microphones translation determines the quality of the sound
that can be recorded. No microphone is perfect. Each subtly distorts the translation, not
making the results unidentifiable but minutely coloring the captured sound. One side of the
915
916
microphone designers art is to make these colorations as pleasing as possible. Another

side of the art is to attempt to make the microphone work more like the human ear, tuning in only to what you want to hear, rejecting unwanted sounds.
Technologies
Engineers can use any of several technologies to build microphones. The microphones
youre most likely to connect to a sound system to capture your voice are dynamic. A
dynamic microphone acts like a small electrical generator or dynamo, using a moving
magnetic field to induce a current in a coil of wire. To detect changes in air pressure, a
dynamic microphone puts a diaphragm into the path of sound waves. The diaphragm is
typically made from lightweight plastic and formed into a domed shape, or something
even more elaborate, to stiffen it. The diaphragm connects to a lightweight coil of wire
called a voice coil thats wrapped around a small, usually cylindrical, permanent magnet.
The voice coil is suspended so that it can move across the magnet as the diaphragm
vibrates. The moving coil in the permanent magnetic field generates a small voltage,
which provides the signal to the microphone input of your sound system.
Most microphones used for recording music today use a different operating principle.
Called condenser microphones (or sometimes, capacitor microphones), they modify an existing
voltage instead of generating a new one. In a classic condenser microphone, the
diaphragm acts as one plate of an electrical capacitor (which, in the days of vacuum tubes
was often called a condenser, hence the name of the microphone). As the diaphragm
vibrates, the diaphragm capacitance changes, which in turn modifies the original voltage.
Directionality
Microphones are often described by the directionality, how they respond to sounds coming
from different directions. An omnidirectional microphone does not discriminate between
sounds, no matter what direction they come from. This type of microphone hears everything the same in a full circle around itself. A unidirectional microphone has one preferred
direction in which it hears best. It partially rejects sounds from other directions. Most
unidirectional microphones are most sensitive to sounds directly in front of them. Sounds
in the preferred direction are called on-axis sounds. Those that are not favored are called
off-axis sounds. The most popular unidirectional microphone is called the cardioid microphone because of the heart-like shape of its pattern of sensitivity (kardia being Greek for
heart). Hypercardioid microphones focus their coverage more narrowly while maintaining
the basic cardioid shape. Bidirectional microphones are, as the name implies, sensitive to
sounds coming from two directions, generally the front and rear of the microphone,
which resembles the numeral 8. Consequently, bidirectional microphones are sometimes
called figure-eight microphones. This design is chiefly used in some special stereophonic
recording techniques. Figure 25.2 illustrates the major types of microphone directional
patterns.
Hardware
FIGURE 25.2
Microphone directional patterns.
Omnidirectional
Bidirectional
Cardioid
unidirectional
The inexpensive microphones that accompany cassette tape recorders and some sound
systems are typically omnidirectional dynamic microphones. If you want to minimize
external noises or office commotion when annotating documents, a better unidirectional
microphone will often make a vast improvement in sound quality.
The signals produced by microphones are measured in several ways. The two most
important characteristics are impedance and signal level.
Microphones are known as low impedance (from 50 to 600 ohms) and high impedance
(50,000 and more ohms). Some microphones have switches that allow you to change their
impedance. Plugging a microphone of one impedance into a circuit meant for another results
in low power transferfaint signals. Nearly all professional microphones and most other
microphones now operate at low impedance, as do most microphone inputs. If your microphone has an impedance switch, youll usually want it set to the low (150 ohm) position.
The signal levels produced by microphones are measured in millivolts or dB (decibels) at a
given sound pressure level. This value is nominal. Loud sounds produce higher voltages.
Most microphones produce signals described as 60 to 40 dBv, and they will work with
most microphone inputs. If you shout into any microphone, particularly one with a higher
output (closer to 40 dB), its output level may be too high for some circuits to process
properly, particularly those in consumer equipmentsay, your computers sound system.
The high level may cause distortion. Adding an attenuator (or switching the microphone
with output level switches to a lower level) will eliminate the distortion.
Microphone signals can be balanced or unbalanced. Balanced signals require two wires
and a ground; unbalanced, one wire and a ground. Balanced signals are more immune to
noise. Unbalanced signals require less sophisticated electronic input circuitry. Most sound
systems use unbalanced signals. Most professional microphones produce balanced signals.
You can often convert a balanced signal into an unbalanced one (so you can connect a professional microphone to your sound system) by tying together one of the two signal wires
of the balanced circuit with the ground. The ground and the other signal wire then act as
an unbalanced circuit.
917
918
Connectors
Both inexpensive microphones and sound systems use the same kind of connector, known
as a miniature phone plug. Better quality professional microphones with balanced signals
typically use XLR connectors (named after the model designation of one of the original
designs) with three pins for their two signals and ground. In these connectors, pin 1 is
always ground. In balanced circuits, pin 2 carries the positive signal; pin 3, the negative.
When used in unbalanced circuits, pins 1 and 3 are usually connected together.
Phone plugs have two or three connections. The end of the plug is called the tip, and the
shaft of the connector is called the sleeve. Some connectors have a third contact in the
form of a thin ring between the tip and the sleeve. This ring is called the ring. Figure 25.3
illustrates a typical phone plug.
FIGURE 25.3
Ring
Components of a
typical phone plug.
Tip
Sleeve
In unbalanced audio circuits, the tip is always connected to the hot or positive signal wire,
and the sleeve is connected to the shield or ground. With balanced signals, positive still
connects to the tip, negative connects to the ring, and the shield or ground goes to the
sleeve. In stereo connections, the left channel goes to the tip, the right channel to the
ring, and the common ground or shield goes to the sleeve.
Loudspeakers
From the standpoint of a computer, moving air is a challenge as great as bringing together
distant worlds, the electronic and the mechanical. To make audible sounds, the computer
must somehow do mechanical work. It needs a transducer, a device that transmits energy
from one system to anotherfrom the electrical computer to the kinetic world of sound.
The device of choice is the dynamic loudspeaker, invented in 1921 by Kellogg Rice.
The dynamic loudspeaker reverses the dynamic microphone design. An electrical current
activates a voice-coil (a solenoid or coil of wire that gives the speaker its voice) that acts as
an electromagnet, which is wrapped around a permanent magnet. The changing current
in the voice-coil changes its magnetic field, which changes its attraction and repulsion of
the permanent magnet, which makes the voice-coil move in proportion to the current
change. A diaphragm called the speaker cone is connected to the voice-coil and moves with
the voice-coil to create the pressure waves of sound. The entire assembly of voice-coil,
cone, and supporting frame is called a speaker driver.
The art of loudspeaker design only begins with the driver. The range of human hearing
far exceeds the ability of any driver to reproduce sound uniformly. Accurately reproducing
Hardware
the full range of frequencies that you can hear requires either massive electronic compensation (called equalization by audio engineers) or using multiple speaker drivers, with each
driver restricted to a limited frequency range.
Commercial speaker systems split the full audible frequency range into two or three
ranges to produce two-way and three-way speaker systems. Modern systems may use more
than one driver in each range, so a three-way system may actually have five drivers.
Woofers operate at the lowest frequencies, which mostly involve bass notes, usually at frequencies of 150 hertz and lower. Tweeters handle the high frequencies associated with the
treble control, frequencies that start somewhere in the range of 2000 to 5000 hertz and
wander off to the limits of human hearing. Midrange speaker drivers take care of the range
in between. A crossover divides the full range of sound into the individual ranges required
by the specialized speaker drivers.
The term subwoofer is also used to describe a special, auxiliary baffled speaker system
meant to enhance the sound of ordinary speakers by extending their low-frequency range.
Because the human ear cannot localize low-frequency sounds, you can place this subwoofer anywhere in a listening room without much effect on stereophonic imaging. The
other, smaller speakers are often termed satellite speakers.
Baffles and Enclosures
The low-frequency range is particularly difficult for speaker systems to reproduce. The
physics of sound require that more air move at lower frequencies to achieve the same
pressure changes or loudness, so larger speakers do a better job generating low frequency
sounds. But the packaging of the speaker also influences its low-frequency reproduction.
At low frequencies, the pressure waves created by a loudspeaker can travel a substantial
distance in the time it takes the speaker cone to move in and out. In fact, when frequencies are low enough, the air has time to travel from the high pressure area in front of the
speaker to the low pressure area behind an outward-moving speaker cone. The moving air
cancels out the air pressure changes and the sound. At low frequenciestypically those
below about 150 Hza loudspeaker in free air has little sound output. The small size and
free-air mounting of the loudspeakers inside computers severely constrain their ability to
reproduce low frequencies.
To extend the low-frequency range of loudspeakers, designers may install the driver in a
cabinet that blocks the air flow from the front to the back of the speaker. The cabinet of a
speaker system is often termed a baffle or enclosure. Strictly speaking, the two terms are not
equivalent. A baffle controls the flow of sound, whereas an enclosure is a cabinet that encircles the rear of the speaker.
Not just any cabinet will do. The design of the cabinet influences the ultimate range of the
system as well as its ability to deliver uniform frequency response. As with any enclosed
volume, the speaker enclosure has a particular resonance. By tuning the resonance of the
919
920
enclosure, speaker system designers can extend the frequency range of their products.
Larger enclosures have lower resonances, which helps accentuate the lowest frequencies
speaker drivers can produce.
Most speaker enclosures use one of two designs. Acoustic suspension speaker systems seal
the low-frequency driver in a cabinet, using the confined air to act as a spring (which in
effect suspends the speaker cone in its resting position). A ducted speaker or tuned-port
speaker or bass reflex speaker puts a vent or hole in the cabinet. The vent both lowers the
resonance of the enclosure and, when properly designed, allows the sound escaping from
the vent to reinforce that produced by the speaker driver. Ducted speakers are consequently more efficient, producing louder sounds for a given power, and they can be
smaller for a given frequency range.
Although tuning a speaker cabinet can extend its frequency range downward, it cant work
magic. The laws of physics stand in the way of allowing a speaker of a size that would fit
on your desk or bookshelf or inside a monitor from reproducing bass notes at levels you
can hear. For most business applications, reproducing low frequencies isnt necessary and
may even be bothersome to coworkers in adjacent cubicles when you start blasting foreign
agents and aliens.
Subwoofers and Satellites
A subwoofer extends the low-frequency capabilities of your computers sound system for
systems that need or require it. The distinguishing characteristic of the subwoofer is that
it is designed to supplement other speaker systems and reproduce only the lowest audible
frequencies, typically those from 20 to 100 hertz. Because these low frequencies are essentially nondirectional (your ear cannot tell where they are coming from), a single subwoofer suffices in stereo and multichannel sound systems.
The classic speaker system puts all the drivers for various frequency ranges in a single cabinet to produce a full-range speaker system. One major trend in speaker design is to abandon the full-range layout and split the cabinetry. Designers put the midrange speakers and
tweeters into small cabinets called satellite speakers and rely on one or two subwoofers to
produce the low frequencies. Practical considerations underlie this design. The small size
of the satellites allows you to place them where convenient for the best stereo imaging,
and you can hide the nondirectional subwoofers out of sight. Figure 25.4 shows a satellitesubwoofer combination.
Hardware
FIGURE 25.4
Satellite speakers
combined with a
subwoofer.
Passive and Active Systems

The speakers you add to your computer typically come in one of two types: active or
passive. The difference is that active speakers have built-in amplifiers whereas passive
speakers do not.
Most speaker systems designed for stereo systems are passive. They rely on your receiver
or amplifier to provide them with the power they need to make sound. Most computer
speakers are active, with integral amplifiers designed to boost the weak output of the typical sound system to the level required for filling a room with sound.
The amplifiers in active speakers are like any other audio amplifiers, with output power
measured in watts (and in theory matched to the speakers) and quality measured in terms
of frequency response and distortion. The big difference is that most active speakers, originally designed for portable stereos, operate from battery power. If you plan to plug active
speakers into your desktop computer, ensure that you get a battery eliminator power supply so you can plug them into a wall outlet. Otherwise, if youre serious about multimedia,
youll be single-handedly supporting the entire battery industry.
Most sound systems produce sufficient power to operate small passive speaker systems.
Their outputs are almost uniformly about four watts because all use similar circuitry to
generate the power. This level is enough even for many large stereo-style passive speaker
systems. Most active speakers work with these higher-powered sound systems and, in
many cases, deliver better (if just louder!) sound through their own amplifiers.
921
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
26
Printers
Printing, reduced to its basics, is the art of moving ink from one place to
another. Although that definition likely will please only a college instructor lost in his own vagueness, any more precise description fails in the
face of a reality laced with printouts of a thousand fonts and far fewer
thoughts. A modern computer printer takes ink from a reservoir and
deposits it on paper or some other medium in patterns determined by
your ideas and computer. In other words, a printer makes your thoughts
visible.
Behind this worthy goal is one of the broadest arrays of technology in
data processing, including processes akin to hammers, squirt guns, and
flashlights. The range of performance is wider than with any other
peripheral. Various printers operate at speeds from lethargic to lightning-like, from slower than an arthritic typist with one hand tied behind
his back to faster than Speedy Gonzales having just munched tacos laced
with amphetamines. They are packaged as everything from one-pound
toteables to truss-stressing monsters and look like anything from
Neolithic bricks to Batmans nightmares. Some personal-size printers dot
paper with text quality that rivals that of a professional publisher and
chart out graphics with speed and sharpness that puts a plotter to shame.
Some make a two-year-olds handiwork look elegant.
Printer technology predates the video display system. In fact, printing is
far older than the personal computer, computers in general, or even electronics. It can trace its inky roots back at least as far as Johannes
Gutenberg, who first slathered ink on slugs of lead and squeezed it onto
paper over 500 years ago (he printed his famous 42-Line Bible in Mainz,
in what is now Germany, in 1455).
924
Gutenberg didnt invent the ink-transfer process. His insight was to subdivide the woodblocks used for making page-size prints of pictures into reusable pieces. Instead of wood,
with its limited life, he cast his characters out of rot-free metal. Each alphabetic letter
became its own metal printing block. In effect, Gutenberg invented character-based technology that served the first generation of computer printers so well.
The latest printing technologies simply take Gutenberg to the extreme, subdividing the
printed character and producing the ink-equivalent of videos all-points-addressable
graphics. The technology divides each character into a matrix of minuscule dots and
prints them individually. The arrangement of dots makes each characteror forms the
bright and dark parts of a graphic image. The dots have gotten smaller and the detail produced greater as computer technology and the ability to control the placement of individual dots has improved.
Thirty years ago, about the best printing technology had mastered the ability to place an
entire character as a single piece on paper, squeezing ink out all at once, much like
Gutenberg, as the fully formed character printer. Today, a character made by a laser printer
may comprise 10,000 dots, each one individually controlled.
At one time, printing had two faces. Commercial printing was Gutenbergs offspring, the
process of making multiple copies by mass production. Computer printing was more concerned with making individual copies. Computer printers, like the first typewriters (raise
your hand if you remember what a typewriter is), used mechanical means to make each
page look more readable, translating electronic data into ink form.
Today, however, the lines are blurring. The personal printers you attach to your computer
are just as apt to be publishing engines that crank out pamphlets, fliers, and newsletters as
they are individual page-makers. Commercial printing presses are going electronic and are
just as capable of printing single pages as any desktop machine. Commercial printing
presses may, in fact, use the same technologies as the desktop printer, though on a more
massive scale.
Print Engines
The challenge faced by any printer is how to get ink (or something that looks like ink but
isnt) onto paperand keep it there. The actual mechanism that handles this job and
forms an image on paper is called the print engine. Engineers have developed a number of
technologies for print engines, and each uses a somewhat different physical principle to
form an image on paper. Although each technology has its strengths, weaknesses, and
idiosyncrasies, you might not be able to tell the difference between the pages they print.
Careful attention to detail has pushed quality up to a level where the paper rather than the
printer is the chief limit on resolution, and color comes close to photographic, falling
short only on the depth that only a thick gelatin coating makes possible.
CHAPTER 26 PRINTERS
Print Engines
In making those images, however, the various print engine technologies work differently
at different speeds, at different noise levels, and with different requirements. These differences can make one style of print engine a better choice for your particular application
than the others.
Impact Dot-Matrix
Todays computer printer evolved from a brash, noisy creation called the impact dot-matrix
printer. Although only a few models persist on the market, impact printers once were the
mainstay of the industry, and their method of forming characters, from a matrix of dots
one line at a time, survives to this day in nearly all low-cost printers.
The heart of the dot-matrix printer is a mechanical printhead that shuttles back and forth
across the width of the paper. A number of thin print wires act as the hammers that
squeeze ink from a fabric or Mylar ribbon to paper. Once the mainstay of computer printing, the classic dot-matrix printer is now an endangered species. It remains noteworthy,
however, as the progenitor of the computer printer.
In most impact dot-matrix printers, a seemingly complex but efficient mechanism controls
each of the print wires. The print wire normally is held away from the ribbon and paper,
and against the force of a spring, by a strong permanent magnet. The magnet is wrapped
with a coil of wire that forms an electromagnet, wound so that its polarity is the opposite
of that of the permanent magnet. To fire the print wire against the ribbon and paper, this
electromagnet is energized (under computer control, of course), and its field neutralizes
that of the permanent magnet. Without the force of the permanent magnet holding the
print wire back, the spring forcefully jabs the print wire out against the ribbon, squeezing
ink onto the paper. After the print wire makes its dot, the electromagnet is de-energized
and the permanent magnet pulls the print wire back to its idle position, ready to fire
again. Figure 26.1 shows a conceptual view of the mechanism associated with one printhead wire.
FIGURE 26.1
Conceptual view of
the impact dot-matrix
printhead mechanism.
Electromagnet coil
Permanent magnet
Print wire
Spring
The two-magnets-and-spring approach is designed with one primary purposeto hold

the print wire away from the paper (and out of harms way) when no power is supplied to
925
926
the printer and the printhead. The complexity is justified by the protection it affords the
delicate print wires.
The printhead of a dot-matrix printer is made from a number of these print wire mechanisms. Most first-generation personal computer printers and many current machines use
nine wires arrayed in a vertical column. To produce high quality, the second generation of
these machines increased the number of print wires to 18 or 24. These often are arranged
in parallel rows with the print wires vertically staggered, although some machines use different arrangements. Because the larger number of print wires fit into the same space (and
print at the same character height), they can pack more detail into what they print.
Because they are often finer than the print wires of lesser endowed machines, the multitude of print wires also promises higher resolution.
No matter the number of print wires, the printhead moves horizontally as a unit across
the paper to print a line of characters or graphics. Each wire fires as necessary to form the
individual characters or the appropriate dots for the graphic image. The impact of each
wire is precisely timed so that it falls on exactly the right position in the matrix. The wires
fire on the flythe printhead never pauses until it reaches the other side of the paper.
A major factor in determining the printing speed of dot-matrix machines is the time
required between successive strikes of each print wire. Physical laws of motion limit the
acceleration that each print wire can achieve in ramming toward the paper and back.
Therefore, the time needed to retract and reactuate each print wire puts a physical limit
on how rapidly the printhead can travel across the paper. It cannot sweep past the next dot
position before each of the print wires inside it is ready to fire. If the printhead travels too
fast, dot-positioning (and character shapes) would become rather haphazard.
Adding color to an impact dot-matrix printer is relatively straightforward. The color that
the impact printer actually prints is governed by the ink in or on its ribbon. Although
some manufacturers build color impact printers using multiple ribbons, the most successful (and least expensive) design uses special multicolored ribbons lined with three or four
bands corresponding to the primary colors. To change colors, the printer shifts the ribbon
vertically so that a differently hued band lies in front of the print wires. Most of the time
the printer will render a row in one color, shift ribbon colors, and then go across the same
row in a different color. The extra mechanism required is simple and inexpensive, costing
as little as $50 extra. (Of course, the color ribbon costs more and does not last as long as
its monochrome equivalent.)
Although the ribbons used by most of these color printers are soaked with three or four
colors of ink, they can achieve seven colors on paper by combining color pairs. For example, laying a layer of blue over a layer of yellow results in an approximation of green.
Most impact printers can spread their output across any medium that ink has an affinity
for, including any paper you might have lying around your home, from onion skin to thin
CHAPTER 26 PRINTERS
Print Engines
cardstock. Although both impact and non-impact technologies have been developed to the
point that either can produce high-quality or high-speed output, impact technology takes
the lead when you share one of the most common business needs: making multipart
forms. Impact printers can hammer an impression not just through a ribbon but through
several sheets of paper as well. Slide a carbon between the sheets, or better yet, treat the
paper for noncarbon duplicates, and you get multiple, guaranteed-identical copies with a
single pass through the mechanism. For a number of business applicationsfor example,
the generation of charge receiptsexact carbon copies are a necessity and impact printing
is an absolute requirement.
But impact printers have fallen from favor for several reasons. The primary one is, as always,
cost. That marvelous printhead is a complicated mechanism with many parts to manufacturer and many areas to develop problems. Todays leading printer technologies have cut the
number of moving parts in a printer (and thus its cost) dramatically. Moreover, because the
wires are mechanical constructions, making them smaller is a tough engineering problem.
Compared to other technologies, the size of the dots made by impact printers is huge and
the quality is low. Impact printouts look like, well, old-fashioned computer printouts.
Impact printers reveal their mechanical heritage in other ways. The hammer bashing
against the ribbon and paper makes noise, a sharp staccato rattle that is high in amplitude
and rich in high frequency components, penetrating and bothersome as a dental drill or
angry horde of giant, hungry mosquitoes. Typically, the impact printer rattles and prattles
louder than most normal conversational tones, and it is more obnoxious than an argument. The higher speed the impact printer, the higher the pitch of the noise and the more
penetrating it becomes. Whats more, printheads wear out. Nothing can take a constant
beating without suffering, and tiny printhead wires are no different.
Engineers discovered how to duplicate the work of the impact dot-matrix printer, only do
it better, sharper, and with fewer moving parts and noise. The result was the inkjet
printer, todays low-cost, mass-market leader.
Inkjets
Todays most popular personal printers use inkjet print engines. The odd name, inkjet,
actually describes the printing technology. If it conjures up images of the Nautilus and
giant squid or a B-52 spraying out blue fluid instead of a fluffy white contrail, your mind
is on the right track. Inkjet printers are electronic squids that squirt out ink like miniature
jet engines fueled in full color. Although this technology sounds unlikelya printer that
sprays droplets of ink onto paperit works well enough to deliver image sharpness on par
with most other output technologies.
In essence, the inkjet printer is a line printer, little more than a dot-matrix printer with
the hammer impact removed. Instead of a hammer pounding ink onto paper, the inkjet
927
928
flings it into place from tiny nozzles, each one corresponding to a print wire of the impact
dot-matrix printer. The motive force can be an electromagnet or, as is more likely today, a
piezoelectric crystal (a thin crystal that bends when electricity is applied across it). A
sharp, digital pulse of electricity causes the crystal to twitch and force ink through the
nozzle in its flight to paper.
Todays inkjet printers are able to make sharper images than impact dot-matrix technology because they do not use ribbons, which would blur their images. The on-paper quality of an inkjet can equal and often better that of more expensive laser printers. Even
inexpensive models claim resolution as high or higher than laser printers, say about 1200
or 1440 dots per inch.
Another advantage of the inkjet is color. Adding color is another simple elaboration. Most
color impact printers race their printheads across each line several times, shifting between
different ribbon colors on each passfor example, printing a yellow row, then magenta,
then cyan, and finally black. Inkjet printers typically handle three or four colors in a single
pass of the printhead, although the height of colored columns often is shorter.
The liquid ink of inkjet printers can be a virtue when it comes to color. The inks remain
fluid enough even after they have been sprayed on paper to physically blend together.
This gives color inkjet printers the ability to actually mix their primary colors together to
create intermediary tones. The range of color quality from inkjet printers is wide. The
best yield some of the brightest, most saturated colors available from any technology. The
vast majority, however, cannot quite produce a True Color palette.
Because inkjets are non-impact printers, they are much quieter than ordinary dot-matrix
engines. Without hammers pounding ink onto paper like a bunch of myopic carpenters
chasing elusive nails, inkjet printers sound almost serene in their everyday work. The tiny
droplets of ink rustle so little air they make not a whisper. About the only sound you hear
from them is the carriage coursing back and forth.
As mechanical line printers, however, inkjet engines have an inherent speed disadvantage
when compared to page printers. Although they deliver comparable speeds on text when
they use only black ink, color printing slows them considerably, to one-third speed or less.
The underlying reason for this slowdown is that most color inkjets dont treat colors
equally and favor black. After all, youll likely print black more often than any color or
blend. A common Lexmark color inkjet printhead illustrates the point. It prints columns
of color only 16 dots high while printing black columns 56 dots high (see Figure 26.2).
Printing a line of color the same height as one in black requires multiple passes, even
though the printer can spray all three colors with each pass.
CHAPTER 26 PRINTERS
Print Engines
FIGURE 26.2
Printheads from a
color inkjet printer
showing nozzle
placement.
Black ink nozzle plate
Color ink nozzle plate
Inkjet technology also has disadvantages. Although for general use you can consider them
to be plain-paper printers, able to make satisfactory images on any kind of stock that will
feed through the mechanism, to yield their highest quality inkjets require special paper
with controlled absorbency. Although plain paper produces printouts adequate for business letters and other public disclosures, inkjet paper delivers the last iota of sharpness.
You also have to be careful to print on the correct side of the paper because most paper
stocks are treated for absorption only on one side. If you try to get by using cheap paper
that is too porous, the inks wick away into a blur. If the paper is too glossy, the wet ink
can smudge.
Early inkjet printers also had the reputation, often deserved, of clogging regularly. To
avoid such problems, better inkjets have built-in routines that clean the nozzles with each
use. These cleaning procedures do, however, waste expensive ink. Most nozzles now are
self-sealing, so when they are not used, air cannot get to the ink. Some manufacturers
even combine the inkjet and ink supply into one easily changeable module. If, however,
you pack an inkjet away without properly purging and cleaning it first, it is not likely to
work when you resurrect it months later.
Inkjet printers commonly use two different technologies, thermal and piezo-electric. (A
third inkjet technology, phase-change, is distinct enough to have entirely different printing qualities.) At heart, the basic technology of both kinds of inkjets is the same. The
machines rely on the combination of the small orifice in the nozzle and the surface tension of liquid ink to prevent a constant dribble from the jets. Instead of oozing out, the
ink puckers around the hole in the inkjet the same way that droplets of water bead up on a
waxy surface. The tiny ink droplets scrunch together rather than spread out or flow out
the nozzle, because the attraction of the molecules in the ink (or water) is stronger than
the force of gravity. The inkjet engine needs to apply some force to break the surface tension and force the ink out, and thats where the differences in inkjet technologies arise.
929
930
Thermal Inkjets
The most common inkjet technology is called thermal because it uses heat inside its printhead to boil a tiny quantity of water-based ink. Boiling produces tiny bubbles of steam
that can balloon out from the nozzle orifices of the printhead. The thermal mechanism
carefully controls the bubble formation. It can hold the temperature in the nozzle at just
the right point to keep the ink bubble from bursting. Then, when it needs to make a dot
on the paper, the printhead warms the nozzle, the bubble bursts, and the ink sprays from
the nozzle to the paper to make a dot. Because the bubbles are so tiny, little heat or time
is required to make and burst the bubblesthe printhead can do it hundreds of times in a
second.
This obscure process was discovered by a research specialist at Canon way back in 1977,
but developing it into a practical printer took about seven years. The first mass-marketed
computer inkjet printer was the Hewlett-Packard ThinkJet, introduced in May, 1984,
which used the thermal inkjet process (which HP traces back to a 1979 discovery by HP
researcher John Vaught). This single-color printer delivered 96 dot per inch resolution at
a speed of 150 characters per second, about on par with the impact dot-matrix printers
available at the same time. The technologynot to mention the speed and resolution
has improved substantially since then. The proprietary name, BubbleJet, used by Canon
for its inkjet printer derives from this technology, although thermal-bubble design is also
used in printers manufactured by Hewlett-Packard, Lexmark, and Texas Instruments.
The heat that makes the bubbles is the primary disadvantage of the thermal inkjet system.
It slowly wears out the printhead, requiring that you periodically replace it to keep the
printer working at its best. Some manufacturers minimize this problem by combining
their printers nozzles with their ink cartridges so that when you add more ink you automatically replace the nozzles. With this design you never have to replace the nozzles, at
least independently, because you do it every time you add more ink.
Because nozzles ordinarily last much longer than the supply in any reasonable inkjet reservoir, other manufacturers make the nozzles a separately replaceable part. The principal
difference between these two systems amounts to nothing more than how you do the
maintenance. Although the combined nozzles-and-ink approach would seem to be more
expensive, the difference in the ultimate cost of using either system is negligible.
Piezoelectric Inkjets
The alternative inkjet design uses the squirt gun approachmechanical pressure to
squeeze the ink from the printhead nozzles. Instead of a plunger pump, however, these
printers generally use special nozzles that squash down and squeeze out the ink. These
nozzles are made from a piezoelectric crystal, a material that bends when a voltage is applied
across it. When the printer zaps the piezoelectric nozzle with a voltage jolt, the entire
CHAPTER 26 PRINTERS
Print Engines
nozzle flexes inward, squeezing the ink from inside and out the nozzle, spraying it out to
the paper. This piezoelectric nozzle mechanism is used primarily by Epson in its Stylus
line of inkjet printers.
The chief benefit of this design, according to Epson, is a longer-lived printhead. The
company also claims it yields cleaner dots on paper. Bursting bubbles may make halos of
ink splatter, whereas the liquid droplets from a piezoelectric printer form more solid dots.
Phase-Change Inkjets
Closely related to inkjet machines are phase-change printers. These printers are actually a
derivation on inkjet technology that concentrates on the ink more than its motion. Instead
of using solvent-based inks that are fixed (that is, that dry) by evaporation or adsorption
into the print medium, the phase-change printer uses inks that harden, changing phase
from liquid to solid. Scientifically speaking, the hardening process is a change in the state
or phase of the ink, hence the name of the technology.
The ink of the phase-change printer starts as solid sticks or chunks of specially dyed wax.
The printhead melts the ink into a thin liquid that is retained in a reservoir inside the
printhead. The nozzles mechanically force out the liquid and spray it on paper. The tiny
droplets, no longer heated, rapidly cool on the medium, returning to its solid state.
Because of the use of solid ink, this kind of printer is sometimes called a solid inkjet printer.
The first printer to use phase-change technology was the Howtek Pixelmaster in the late
1980s. Marketed mostly as a specialty machine, the Howtek made little impression in the
industry. Phase-change technology received its major push from Tektronix with its introduction of its Phaser III PXi in 1991. Tektronix, which was acquired by Xerox in 2001,
refined phase-change technology to achieve smoother images and operation. Whereas the
Pixelmaster used plastic-based inks that left little lumps on paper and sometimes clogged
the printhead, the Phaser III used wax-based inks and a final processing stepa cold fuser
that flattened the cold ink droplets with a steel roller as the paper rolled out of the printer.
Phase-change printers are less sensitive to media than ordinary inkjet printers. They are
also renown for the bright, saturated colors of their wax-based inks.
Laser
The one revolution that has changed the faces of both offices and forests around the
world is the photocopier. Trees plummet by the millions to provide fodder for the duplicate, triplicate, megaplicate. Todays non-impact, bit-image laser printer owes its life to
this technology.
At heart, the laser printer principle is simple. Some materials react to light in strange
ways. Selenium and some complex organic compounds modify their electrical conductivity
931
932
in response to exposure to light. Both copiers and laser printers capitalize on this photoelectric effect by focusing an optical image on a photoconductive drum that has been
given a static electrical charge. The charge drains away from the conductive areas that
have been struck by light but persists in the dark areas. A special pigment called a toner is
then spread across the drum, and the toner sticks to the charged areas. A roller squeezes
paper against the drum to transfer the pigment to the paper. The pigment gets bonded to
the paper by heating or fusing it.
The laser printer actually evolved from the photocopier. Rather than the familiar electrostatic Xerox machine, however, the true ancestor of the laser printer was a similar competing process called electrophotography, which used a bright light to capture an image and
make it visible with a fine carbon-based toner. The process was developed during the
1960s by Keizo Yamaji at Canon. The first commercial application of the technology,
called New Process to distinguish it from the old process (xerography), was a Canon photocopier released in 1968.
The first true laser printer was a demonstration unit made by Canon in 1975 based on a
modified photocopier. The first commercial computer laser printer came in 1984 when
Hewlett-Packard introduced its first LaserJet, which was based on the Canon CX engine.
At heart, it and all later laser printers use the same process, a kind of heat-set, lightinspired offset printing.
The magic in a laser printer is forming the image by making a laser beam scan back and
forth across the imaging drum. The trick, well known to stage magicians, is to use mirrors. A small, rotating mirror reflects the laser across the drum, tracing each scan line
across it. The drum rotates to advance to the next scan line, synchronized to the flying
beam of laser light. To make the light-and-dark pattern of the image, the laser beam is
modulated on and off. Its rapidly switched on for light areas, off for dark areas, one
minuscule dot at a time to form a bit-image.
The major variations on laser printing differ only in the light beam and how it is modulated. LCD-shutter printers, for example, put an electronic shutter (or an array of them)
between a constant light source (which need not be a laser) and the imaging drum to
modulate the beam. LED printers modulate ordinary light-emitting diodes (LEDs) as their
optical source. In any case, these machines rely on the same electrophotographic process
as the laser printer to carry out the actual printing process.
The basic laser printer mechanism requires more than just a beam and a drum. In fact, it
involves several drums or rollers, as many as six in a single-color printer and more in color
machines. Each has a specific role in the printing process. Figure 26.3 shows the layout of
the various rollers.
CHAPTER 26 PRINTERS
Print Engines
FIGURE 26.3
Conceptual view of
the laser printer
mechanism.
Clea
ning
Toner
Charging
Roller
blad
Optical Photo Conductor

Drum
Fuser
Backup
Roller
Developer
Roller
Paper
Transfer
Roller
The imaging drum, often termed the OPC for optical photoconductor, first must be
charged before it will accept an image. A special roller called the charging roller applies the
electrostatic charge uniformly across the OPC.
After the full width of an area of the OPC gets its charge, it rotates in front of the modulated light beam. As the beam scans across the OPC drum and the drum turns, the system
creates an electrostatic replica of the page to be printed.
To form a visible image, a developing roller then dusts the OPC drum with particles of
toner. The light-struck areas with an electrostatic charge attract and hold the toner
against the drum. The unexposed parts of the drum do not.
The printer rolls the paper between the OPC drum and a transfer roller, which has a
strong electrostatic charge that attracts the toner off the drum. Because the paper is in
between the transfer roller and the OPC drum, the toner collects on the paper in accord
with the pattern that was formed by the modulated laser. At this point, only a slight electrostatic charge holds the toner to the paper.
To make the image permanent, the printer squeezes the paper between a fuser and backup
roller. As the paper passes through, the printer heats the fuser to a high temperatureon
the order of 350 degrees Fahrenheit (200 degrees Celsius). The heat of the fuser and the
pressure from the backup roller melt the toner and stick it permanently on the paper. The
completed page rolls out the printer.
Meanwhile, the OPC drum continues to spin, wiping the already-printed area against a
cleaning blade that scrapes any leftover toner from it. As the drum rotates around to the
charging roller again, the process repeats. Early lasers used OPC drums large enough to
hold an entire single-page image. Modern machines use smaller rollers that form the
image as a continuous process.
933
934
Although individual manufacturers may alter this basic layout to fit a particular package or
to refine the process, the technology used by all laser machines is essentially the same. At
a given resolution level, the results produced by most mechanisms is about the same, too.
You need an eye loupe to see the differences. The major difference is that manufacturers
have progressively refined both the mechanism and electronics to produce higher resolutions. Basic laser printer resolution starts at 300 dots per inch. The mainstream is now at
the 600 dpi level. The best computer-oriented laser printers boast 1200 dpi resolution.
In most laser printers, the resolution level is fixed primarily by the electronics inside the
printer. The most important part of the control circuitry is the raster image processor, also
known as the RIP. The job of the RIP is to translate the string of characters or other
printing commands into the bit-image that the printer transfers to paper. In effect, the
RIP works like a video board, interpreting drawing commands (a single letter in a print
stream is actually a drawing command to print that letter), computing the position of each
dot on the page and pushing the appropriate value into the printers memory. The memory of the printer is arranged in a raster just like the raster of a video screen, and one
memory cella single bit in the typical black-and-white laser printercorresponds to
each dot position on paper.
The RIP itself may, by design, limit a laser printer to a given resolution. Some early laser
printers made this constraint into an advantage, allowing resolution upgrades through
after-market products that replaced the printers internal RIP and controlled the printer
and its laser through a video input. The video input earns its name because its signal is
applied directly to the light source in the laser in raster scanned form (like a television
image), bypassing most of the printers electronics. The add-in processor can modulate the
laser at higher rates to create higher resolutions.
Moving from 300 dpi to 600 dpi and 1200 dpi means more than changing the RIP and
adding memory, however. The higher resolutions also demand improved toner formulations because, at high resolutions, the size of toner particles limits sharpness, much as the
size of print wires limits impact dot-matrix resolution. With higher-resolution laser printers, it becomes increasingly important to get the right toner, particularly if you have toner
cartridges refilled. The wrong toner limits resolution just as a fuzzy ribbon limits the
quality of impact printer output.
Adding color to a laser printer is more than dumping a few more colors of toner. The
laser must separately image each of its three or four primary colors and transfer the toner
corresponding to each to the paper. The imaging process for each color requires forming
an entire image by passing it past the OPC drum. Forming a complete image consequently requires three or four passes of the drum.
Exactly what constitutes a pass varies among manufacturers. Most color laser printers use
three or four distinct passes of each sheet of paper. The paper rolls around the drum and
CHAPTER 26 PRINTERS
Print Engines
makes four complete turns. Each color gets images separately on the drum, then separately transferred to the sheet. The printer wipes the drum clean between passes.
So-called one-pass printing, pioneered by Hewlett-Packard, still requires the drum to
make four complete passes as each color gets separately scanned on the drum and toner is
dusted on the drum separately for each color. The paper, however, only passes once
through the machine to accept the full-color image at once and then to have all four
colors fused together onto the paper. The first three colors merely transfer to the drum.
After the last colorblackgets coated on the drum, the printer runs the paper through
and transfers the toner to it. The paper thus makes a single pass through the printer,
hence the one-pass name.
This single-pass laser technology yields no real speed advantage. The photoconductor
drum still spins around the same number of times as a four-pass printer. The speed at
which the drum turns and the number of turns it makes determines engine speed, so the
one-pass process doesnt make a significant performance increase.
The advantage to one-pass color laser printing comes in the registration of the separate
color images. With conventional color laser systems, the alignment of the paper must be
critically maintained for all four passes for all the colors to properly line up. With the
one-pass system, paper alignment is not a problem. Only the drum needs to maintain its
alignment, which is easy to do because it is part of the mechanism rather than an interloper from the outside world.
No matter the number of passes, adding color in laser printing subtracts speed. In general,
color laser speed falls to one-quarter the monochrome speed of a similar engine because of
the requirement of four passes. (With three-pass printing, speed falls to one-third the
monochrome rate). For example, a printer rated at 12 pages per minute in monochrome
will deliver about 3 ppm in color. Even allowing for this slowdown, however, color lasers
are usually faster than other color page printers. They are also often quieter. Compared to
thermal wax transfer printers, a popular high-quality color technology, they are also economical because a laser uses toner only for the actual printed image. Thermal wax machines
need a full page of ink for each page printed regardless of the density of the image.
Direct Thermal
A printer that works on the same principle as a wood-burning set might seem better for a
Boy Scout than an on-the-go executive, but todays easiest-to-tote printers do exactly
thatthe equivalent of charring an image on paper. Thermal printers use the same electrical heating of the wood-burner, a resistance that heats up with the flow of current. In
the case of the thermal printer, however, the resistance element is tiny and heats and cools
quickly, in a fraction of a second. As with inkjets, the thermal printhead is equivalent to
that of a dot-matrix printer, except that it heats rather than hits.
935
936
Thermal printers do not, however, actually char the paper on which they print. Getting
paper that hot would be dangerous, precariously close to combustion (although it might
let the printer do double-duty as a cigarette lighter). Instead, thermal printers use special,
thermally sensitive paper that turns from white to near-black at a moderate temperature.
Thermal technology is ideal for portable printers because few moving parts are involved
only the printhead moves, nothing inside it. No springs and wires means no jamming.
The tiny, resistive elements require little power to heat, actually less than is needed to fire
a wire in an impact printer. Thermal printers can be lightweight, quiet, and reliable. They
can even run on batteries.
The special paper they require is one drawback. Not only is it costly (because it is, after
all, special paper), but it feels funny and it is prone to discoloration if it is inadvertently
heated to too high a temperature. Paper cannot tell the difference between a hot printhead and a cozy corner in the sun.
Thermal printers have become special application machines. Inkjets have many of the same
virtues and more reasonable paper; therefore, low-cost inkjets dominate the printer market.
Thermal Transfer
Engineers have made thermal technology more independent of the paper or printing
medium by moving the image-forming substance from the paper to a carrier or ribbon.
Instead of changing a characteristic of the paper, these machines transfer pigment or dyes
from the carrier to the paper. The heat from the printhead melts the binder holding the
ink to the carrier, allowing the ink to transfer to the paper. On the cool paper, the binder
again binds the ink in place. In that the binder is often a wax, these machines are often
called thermal wax transfer printers.
These machines produce the richest, purest, most even and saturated color of any color
print technology. Because the thermal elements have no moving parts, they can be made
almost arbitrarily small to yield high resolutions. Current thermal wax print engines
achieve resolutions similar to those of laser printers. However, due to exigencies of printhead designs, the top resolution of these printers extends only in one dimension (vertical).
Top thermal wax printers achieve 300 dots per inch horizontally and 600 dots per inch
vertically.
Compared to other technologies, however, thermal wax engines are slow and wasteful.
They are slow because the thermal printing elements must have a chance to cool off
before advancing the 1/300th of an inch to the next line on the paper. And they are wasteful because they use wide ink transfer sheets, pure colors supported in a wax-based
medium clinging to a plastic film basesort of like a Mylar typewriter ribbon with a
gland condition. Each of the primary colors to be printed on each page requires a swath of
inked transfer sheet as large as the sheet of paper to be printedthat is, nearly four feet of
CHAPTER 26 PRINTERS
Print Engines
transfer sheet for one page. Consequently, printing a full-color page can be expensive,
typically measured in dollars rather than cents per page.
Because thermal wax printers are not a mass market item and each manufacturer uses its
own designs for both mechanism and supplies, you usually are restricted to one source for
ink sheetsthe printer manufacturer. Although that helps ensure quality (printer-makers
pride themselves on the color and saturation of their inks), it also keeps prices higher than
they might be in a more directly competitive environment.
For color work, some thermal wax printers give you the choice of three- or four-pass
transfer sheets and printing. A three-pass transfer sheet holds the three primary colors of
inkred, yellow, and bluewhereas a four-color sheet adds black. Although black can be
made by overlaying the three primary colors, a separate black ink gives richer, deeper
tones. It also imposes a higher cost and extends printing time by one-third.
From these three primary colors, thermal wax printers claim to be able to make anywhere
from seven to nearly 17 million colors. That prestidigitation requires a mixture of transparent inks, dithering, and ingenuity. Because the inks used by thermal wax printers are
transparent, they can be laid one atop another to create simple secondary colors. They do
not, however, actually mix.
Expanding the thermal wax palette further requires pointillistic mixing, laying different
color dots next to each other and relying on them to visually blend together in a distant
blur. Instead of each dot of ink constituting a picture element, a group of several dots
effectively forms a super pixel of an intermediate color.
The penalty for this wider palette is a loss of resolution. For example, super pixels measuring five by five dots would trim the resolution of a thermal wax printer to 60 dots per
inch. Image quality looks like a color halftonea magazine reproductionrather than a
real photograph. Although the quality is shy of perfection, it is certainly good enough for
a proof of what is going to a film recorder or the service bureau to be made into color
separations.
A variation of the thermal wax design combines the sharpness available from the technology with a versatility and cost more in line with ordinary dot-matrix printers. Instead of
using a page-wide printhead and equally wide transfer sheets, some thermal wax machines
use a line-high printhead and a thin transfer sheet that resembles a Mylar typewriter ribbon. These machines print one sharp line of text or graphics at a time, usually in one
colorblack. They are quiet as inkjets but produce sharper, darker images.
Dye-Diffusion
For true photo-quality output from a printer, todays stellar technology is the thermal dyediffusion process, sometimes called thermal dye-sublimation. Using a mechanism similar to
937
938
that of the thermal wax process, dye-diffusion printers are designed to use penetrating
dyes rather than inks. Instead of a dot merely being present or absent, as in the case of a
thermal wax printer, diffusion allows the depth of the color of each dot to vary. The diffusion of the dyes can be carefully controlled by the printhead. Because each of the three
primary colors can have a huge number of intensities (most makers claim 256), the palette
of the dye-diffusion printer is essentially unlimited.
What is limited is the size of the printed area in some printers. The output of most dyediffusion printers looks like photographs in size as well as color. In fact, dye-diffusion
printing doesnt merely look like a photograph, many of your photos are actually made
using a dye-diffusion printer. Kodak, for example, uses dye-diffusion for the color prints
that it processes.
Characteristics
On paper, you may be challenged to determine what printer technology created an image.
Although an old dot-matrix printout is distinctive, modern printers have converged on a
uniform, high-quality level of output. Thats not to say that printer technologies are interchangeable. Laser printers excel at everyday black-and-white business work, producing
pages at high speed and low cost. Inkjets trade their low acquisition cost for a higher cost
per page (mostly because of the absurdly high cost of ink) with the ancillary benefit of
color. They are consequently the top choice for home printing.
But at a deeper level, printers differ in a number of aspects. These include printing speed,
on-paper quality, color capabilities, the print engine, media-handling, and the cost of various consumables. Although many of these issues seem obvious, even trivial, current
printer technologiesand hypeadd some strange twists. For example, speed ratings may
be given in units that cant quite be compared. Quality is more than a matter of dots per
inch. And color can take on an added dimensionblack.
Speed
No one wants to wait. When you finish creating a report or editing a picture, you want
your hardcopy immediatelyyouve finished, and so should your computer. The ideal
printer would be one that produces its work as soon as you give the print commandall
fifty thousand pages of your monthly report in one big belch.
No printer yet has achieved instantaneous operation. In fact, practical machines span a
wide range of printing speeds.
Engineers divide computer printers into two basic types: line printers and page printers. A
line printer, as its name implies, works on text one line at a time. It usually has a printhead
that scans across the paper one line of characters at a time. These printers earn their name
CHAPTER 26 PRINTERS
Characteristics
because they think of printing in terms of the line. Most line printers start printing each
line as soon as all the character data to appear in that line is received. Inkjet printers are
todays preeminent line printers. On the other hand, a page printer rasterizes a full page
image in its own internal memory and prints it one line of dots at a time. It must receive all
the data for a full page before it begins to print the page. Laser printers are page printers.
Line printers and page printers produce equivalent results on paper. For engineers, the
chief difference is what gets held in the printers memory. For you, the chief difference is
how the engineers specify the speed of the two kinds of printer.
Measuring Units
The two most common measurements of printer speed are characters per second and
pages per minute. Both should be straightforward measures. The first represents the number of characters a printer can peck out every second, and the second represents the number of completed pages that roll into the output tray every minute. In printer specification
sheets, however, both are theoretical measures that may have little bearing on how fast a
given job gets printed.
Line printer speed is usually measured in characters per second. Most printer manufacturers derive this figure theoretically. They then take the time the printhead requires to
move from the left side of a page to the right side and divide it into the number of characters that might print on the line. This speed is consequently dependent on the width of
the characters, and manufacturers often choose the most favorable value.
The highest speed does not always result from using the narrowest characters, however.
The rate at which the printhead can spray dots of ink (or hammer print wires) is often
fixed, so to print narrower characters the printer slows its printhead.
Characters per second does not directly translate into pages per minute, but you can
determine a rough correspondence. On a standard sheet of paper, most line printers render 80 characters per line and 60 lines per page, a total of 4800 characters per page.
Because there are 60 seconds in every minute, each page per minute of speed translates
into 80 characters per second. Or you can divide the number of characters per second by
80 to get an approximate page per minute rating.
This conversion can never be exact, particularly on real-world printing chores. Few documents you print will fill every line from left to right with dense text. A line printer uses
time only to work on lines that are actually printed. Modern printers have sufficient builtin intelligence to recognize areas that will appear blank on the printed sheet and dont
bother moving their printheads over these empty spaces. Page printers, on the other hand,
must scan an entire page even if it only has a single line of text on it. A line printer, on the
other hand, dispenses with a single-line page in a few seconds.
939
940
Engine Speed Versus Throughput

Even within a given family of printers, ratings do not reflect real-world performance. The
characters per second or pages per minute rating usually given for a printer does not indicate the rate at which you can expect printed sheets to dribble into the output tray. These
speed measurements indicate the engine speed, the absolute fastest the mechanism of the
printer allows paper to flow through its path. A number of outside factors slow the actual
throughput of a printer to a rate loweroften substantially sothan its engine speed.
With line printers, the speed ratings can come close to actual throughput. The major slowdowns for line printers occur only when lines are short and the printhead changes direction
often and when the printhead travels long distances down the sheet without printing.
With page printers, the difference between theory and reality can be dramatic. Because of
their high resolutions, line printers require huge amounts of data to make bit-image graphics. The transfer time alone for this information can be substantial. Page printers suffer this
penalty most severely when your computer rasterizes the image and sends the entire page as
a bit-image to the printer. On the other hand, if the printer rasterizes the image, the processing time for the rasterization process adds to the print time. In either case it is rare,
indeed, for pages to be prepared as quickly as the engine can print them when graphics are
involved. Instead of pages per minute, throughput may shift to minutes per page.
Modes
Some printers operate in a number of different modes that trade off print quality for
speed. These modes vary with the technology used by the printer.
Older impact printers often had several other modes. In draft mode, they blackened only
every other dot. The thinly laid dots gave text and graphics a characteristic gray look.
Near letter quality mode slowed the printhead so that text characters could be rendered
without the machine-gun separated-dots look of draft mode. Because the dot density is
higher, characters appear fully black and are easier to read. Letter quality mode slowed
printing further, often using two or more passes to give as much detail as possible to each
individual character.
Inkjet printers arent bothered by the mechanical limitations of impact printers, so you
need not worry so much about dot density at higher speeds. Nevertheless, the time
required to form each jet of ink they print constrains the speed of their printheads. Most
inkjet printers operate at the maximum speed the jet-forming process allows at the time.
However, some offer the choice of higher quality modes keyed to different paper types.
An inkjet printer may slow down to render characters with greater quality on better paper.
Although it may not indicate separate printing modes, you can often see a speed difference when you shift between types of paper and resolution levels.
When it comes to choosing between color and black-and-white printing, speed differences
can be substantial. Most color inkjet designs have fewer nozzles for color ink than they do
CHAPTER 26 PRINTERS
Characteristics
for black ink. Typically, each of the three primary colors will have one-third the number
of nozzles as black. As a result, the printhead must make three times as many passes to
render color in the same detail as black, so color mode is often one-third the speed as
black-and-white printing.
The speed relationship between color and black-and-white printing varies widely, however. In comparing the speeds of two printers, you must be careful to compare the same
mode. The most relevant mode is the one youre likely to use most. If you do mostly text,
then black-and-white speed should be the most important measure to you. If you plan to
do extensive color printing, compare color speeds.
Although many bit-image printers dont allow you to directly alter their resolution, you
can accelerate printing by making judicious choices through software. A lower resolution
requires less time for rendering the individual dots, so in graphics mode, choosing a lower
resolution can dramatically accelerate print speed. The speed gain isnt only mechanical.
When you select a lower resolution, you also reduce rendering time. Fewer dots means
less data to manipulate. At low resolutions, graphics printing speed can approach engine
speed. The downside is, of course, you might not like the rough look of what you print.
Quality
The look of what you get on paper isnt completely in your control. By selecting the resolution your printer uses, you can raise speed and lower quality. But every printer faces a
limit as to the maximum quality it can produce. This limit is enforced by the design of the
printer and its mechanical construction. The cause and measurement of these constraints
vary with the printer technology.
As with computer displays, the resolution and addressability of any kind of printer are
often confused. Resolution indicates the reality of what you see on paper; addressability indicates the more abstract notion of dot placement. When resolution is mentioned, particularly with impact dot-matrix printers, most of the time addressability is intended. A
printer may be able to address any position on the paper with an accuracy of, say, 1/120th
of an inch. If an impact print wire is larger than 1/120th of an inch in diameter, however,
the machine never is able to render detail as small as 1/120th of an inch. Inkjet printer
mechanisms do a good job of matching addressability and resolution, but those efforts easily get undone when you use the wrong printing medium. If inkjet ink gets absorbed into
paper fibers, it spreads out and obscures the excellent resolution many of these machines
can produce.
Getting addressability to approach resolution was often a challenge for the designers of the
impact dot-matrix printer (and one of the many reasons this technology has fallen from
favor). The big dots made by the wide print wires blurred out the detail. Better quality
impact dot-matrix printers had more print wires, and they were smaller. Also, the ribbon that
941
942
is inserted between the wires and paper blurred each dot hammered out by an impact dotmatrix printer. Mechanical limits also constrained the on-paper resolution of impact machines.
With non-impact bit-image printers, resolution and addressability usually are the same,
although some use techniques to improve apparent resolution without altering the number
of dots they put in a given area.
Resolution Enhancement Technology (ReT) improves the apparent quality of on-paper printing within the limits of resolutionit can make printing look sharper than would ordinarily be possible. The enhancement technology, introduced by Hewlett-Packard in March,
1990, with its LaserJet III line of printers, works by altering the size of toner dots at the
edges of characters and diagonal lines to reduce the jagged steps inherent in any matrix
bit-image printing technique. Using ReT, the actual on-paper resolution remains at the
rated value of the print enginefor example, 300 or 600 dpibut the optimized dot size
makes the printing appear sharper.
Increasing resolution is more than a matter of refining the design of print engine mechanics.
The printers electronics must be adapted to match, including adding more memory
substantially more. Memory requirements increase as the square of the linear dot density.
Doubling the number of dots per inch quadruples memory needs. At high resolutions, the
memory needs for rasterizing the image can become prodigiousabout 14MB for a 1200 dpi
image. Table 26.1 lists the raster memory needs for common monochrome printer resolutions.
TA B L E 2 6 . 1
Raster Memory Size for Monochrome Printer Resolutions
Resolution
Dots
Bytes
75 dpi
450,000
56,250
150 dpi
1,800,000
225,000
300 dpi
7,200,000
900,000
360 dpi
10,368,000
1,296,000
600 dpi
28,800,000
3,600,000
720 dpi
41,472,000
5,184,000
1200 dpi
115,200,000
14,400,000
Adding color, of course, increases the memory requirements. Fortunately, the color bit-depth
used by common printer technologies doesnt impose the same extreme demands as monitors.
A printer has only a few colors corresponding to the hues of its inks and, except for continuous-tone technologies such as dye-diffusion, the range of each color usually is limited to on or
off. Thankfully, color resolutions are generally substantially lower than monochrome, defined
by the size of the color super-pixels rather than individual dots. In any case, the raster memory
requirements of a color printer are substantially higher than monochrome.
CHAPTER 26 PRINTERS
Characteristics
Note that when printing text, page printers may operate in a character-mapped mode, so
memory usage is not as great. Even with minimal memory, a printer can store a full-page
image in ASCII or a similar code, one byte per letter, as well as the definitions for the
characters of several fonts. In this mode, it generates the individual dots of each character
as the page is scanned through the printer.
Moving to higher resolutions makes other demands on a printer as well. For example, in
laser printers, finer resolutions require improved toner formulations because, at high resolutions, the size of toner particles limits sharpness, much as the size of print wires limits
impact dot-matrix resolution. With higher-resolution laser printers, it becomes increasingly
important to get the right toner, particularly if you have toner cartridges refilled. The wrong
toner limits resolution just as a fuzzy ribbon limits the quality of impact printer output.
Color
Printers start with the primaries when it comes to color. They start with inks corresponding
to the three primary colorsred, yellow, and blue. If you want anything beyond those, the
printer must find some way of mixing them together. This mixing can be physical or optical.
The physical mixing of colors requires that two or more colors of ink actually mix
together while they are wet. Printer inks are, however, designed to dry rapidly, so the
colors to be mixed must be applied simultaneously or in quick succession. Few printers
rely on the physical mixing of inks to increase the number of colors they produce.
Optical mixing takes place in either of two ways. One color of ink can be applied over
another (that has already dried) or the colors can be applied adjacent to one another.
Appling multiple layers of color requires that the inks be to some degree transparent,
because a truly opaque ink would obscure the first color to be applied. Most modern
printer inks are transparent, which allows them to be used on transparencies for overhead
projection as well as on paper. The exact hue of a transparent ink is, of course, dependent
on the color of the medium it is applied to.
Optical mixing also takes place when dots of two or more colors are intermixed. If the
dots are so close together that the eye cannot individually resolve each one, their colors
blend together on the retina, mixing the individual hues together. Most computer color
printers take advantage of this kind of optical mixing by dithering.
Three-, Four-, and Six-Color Printers

Color primaries in printing arent so simple as the familiar threesome. To achieve better
color reproduction, printers use a skewed set of primary colorsmagenta instead of red,
cyan instead of blue, and plain-old ordinary yellow. Even this mix is so far from perfect
that, when all are combined, they yield something thats often far from black.
Consequently, better printers include black in their primary colors.
943
944
Black, in fact, may play two roles in a color printer. A few inkjet printers allow you to
choose between black-and-white and color operation as simply as swapping ink cartridges.
In these machines, black is treated as a separate hue that cannot be mixed in blends with
the three color primaries. These three-color printers render colors only from the three primaries even though some machines can print pure black when using a suitable black-only
ink cartridge. The approximation of black made from the three primaries is termed
composite black and often has a off-color cast. Four-color printers put black on the same
footing as the three primary hues and mix with all four together. This four-color printing
technique gives superior blacks, purer grays, and greater depth to all darker shades.
To further increase the range of pure colors possible with a printer, manufacturers are
adding more colors of ink. The most common addition is two extra ink colors beyond the
standard four, generally orange and violet, which extend the spectrum that the printer can
render. This greater range in primaries translates into more realistic reproduction of photographs with less need for other color-enhancing techniques, such as dithering.
Dithering
Color televisions do an excellent job with their three primaries and paint a nearly infinite
spectrum. But the television tube has a luxury most printers lack. The television can modulate its electron beam and change its intensity. Most printers are stuck with a single
intensity for each color. As a result, the basic range of most printers is four pure colors
and seven when using mixtures, blending magenta and blue to make violet, magenta and
yellow for orange, and blue and yellow for green. Count the background color of the
paper being printed upon, and the basic range of most color printers is eight hues.
Commercial color printing faces the same problem of trying to render a wide spectrum
from four primary colors. To extend the range of printing presses, graphic artists make
color halftones. They break an image into dots photographically using a screen. Using
special photographic techniques (or, more often today, a computer), they can vary the size
of the dot with the intensity of the color.
Most computer printers cannot vary the size of their dots. To achieve a halftone effect,
they use dithering. In dithering, colors beyond the range of pure hues of which a printer is
capable are rendered in patterns of primary-colored dots. Instead of each printed dot representing a single pixel of an image, dithering uses a small array of dots to make a single
pixel. These multiple-dot pixels are termed super-pixels. By varying the number of dots
that actually get printed with a given color of ink in the super-pixel matrix, the printer can
vary the perceived intensity of the color.
The problem with dithering is that it degrades the perceived resolution of the color image.
The resolution is limited by the size of the super-pixels rather than the individual dots. For
example, to attempt to render an image in True Color (eight bits per primary), the printer
must use super-pixels measuring eight by eight dots. The resolution falls by an equivalent
factor. A printer with 600 dpi resolution yields a color image with 75 dpi resolution.
CHAPTER 26 PRINTERS
Characteristics
Drivers
Getting good color with dithering is more art than science. The choice of dithering pattern
determines how smooth colors can be rendered. A bad choice of dithering pattern often
results in a moir pattern overlaid on your printed images or wide gaps between super-pixels. Moreover, colors dont mix the same on screen and on paper. The two media often use
entirely different color spaces (RGB for your monitor, CYMK for your printer), thus requiring a translation step between them. Inks only aspire to be pure colors. The primary colors
may land far from the mark, and a color blended from them may be strange, indeed.
Your printer driver can adjust for all these issues. How well the programmer charged with
writing the driver does his job is the final determinant in the color quality your printer
produces. A good driver can create photo-quality images from an inkjet printer, whereas a
bad driver can make deplorable pictures even when using the same underlying print
engine. Unfortunately, the quality of a printers driver isnt quantified on the specifications
sheet. You can only judge it by looking at the output of a printer. For highest quality,
however, youll always want driver software written for your particular model of printer,
not one that your printer emulates. Moreover, youll want to get the latest driver. You may
want to periodically cruise the Web site of your printer-maker to catch driver updates as
they come out.
Permanence
Most inks fade. The ultraviolet energy in daylight attacks many pigments, breaking the
molecular bonds that give a dye or pigment its color. With enough light, the color
changes, shifting in hue, or fades until it is invisible on paper. Light is not the only culprit.
Airborne chemicals (such as pollution) also react with dyes and pigments and can change
their colors. Pigments sometimes even react with each other when they are mixeda
problem for painters thats rarely encountered in printing.
Of all ink colors, black is usually the most permanent, at least when it is made using carbon as a pigment. This black does not fade in light and, except for fire, rarely reacts with
chemicals in the air. The black toner used by most laser printers is therefore among the
most color-permanent of printing. The conventional commercial printing ink thats used
for books usually is based on carbon pigments.
Colors, however, vary in their permanence. Many of the brightest hues fade quickly, while
a few hues are immutable as black. The color inks used by most computer printers are
designed to last only for a few years. For most business documents, this impermanence is
no problem. It may even save you a trip to the shredder. But if you want to print artwork
or a family photo, having it turn green or fade to white after a few months of display is far
from ideal. To help prevent problems with fading, some printer manufacturers (particularly those offering photo printers) have begun to manufacture permanent inks rated for
more than 100 years of display.
945
946
Permanence has another aspect. Some inks are water soluble. When a printout made
using soluble ink get soaked or even damp, the colors can run like a bad tie-dye job. The
somewhat psychedelic effect can be interesting, but it also interferes with legibility. Most
inkjet printer inks are water soluble. For more permanent printouts, some manufacturers
offer nonsoluble inks and special papers that help fix the soluble inks and prevent their
running.
Paper-Handling
Key to the design of all printers is that their imaging systems operate in only one dimension, one line at a time, be it the text-like line of the line printer or the single raster line
of the page printer. To create a full two-dimensional image, all printers require that the
printing mediumtypically papermove past the print mechanism. With page printers,
this motion must be smooth and continuous. With line printers, the paper must cog forward, hold its position, and then cog forward to the next line. Achieving high resolutions
without distortion requires precision paper movement with a tolerance of variation far
smaller than the number of dots per inch the printer is to produce.
Adding further complexity, the printer must be able to move paper or other printing
media in and out of its mechanism. Most modern printers use sheet feeders that can pull a
single page from a stack, route it past the imaging mechanism, and stack it in an output
tray. Older printers and some of the highest speed machines use continuous-form paper,
which trades a simplified printer mechanism for your trouble in tearing sheets apart. Each
paper-handling method has its own complications and refinements.
The basic unit of computer printing is the page, a single sheet of paper, so it is only natural
for you to want your printer to work with individual sheets. The computer printing process,
however, is one that works with volumenot pages but print jobs, not sheets but reams.
The individual sheet poses problems in printing. To make your ideas, onscreen images,
and printed hardcopy agree, each sheet must get properly aligned so that its images
appear at the proper place and at the proper angle on every sheet. Getting perfect alignment can be vexing for both human and mechanical hands. Getting thousands into alignment is a project that might please only the Master of the Inquisition. Yet every laser
printer, most inkjet printers, and a variety of other machines face that challenge every
time you start a print job.
To cope with this hardcopy torture, the printer requires a complex mechanism called the
cut-sheet feeder, or simply the sheet feeder. Youll find considerable variation in the designs
of the sheet feeders of printers. All are complicated designs involving cogs, gears, rods,
and rollers, and every engineer appears to have his own favorite arrangement. The inner
complexity of these machines is something to marvel at but not dissect, unless you have
CHAPTER 26 PRINTERS
Characteristics
too much time on your hands. Differences in the designs of these mechanisms do have a
number of practical effects: the capacity of the printer, which relates to how long it can
run without your attention; the kinds and sizes of stock that roll through; how sheets are
collated and whether you have to spend half an afternoon to get all the pages in order; and
duplex printing, which automatically covers both sides of each sheet.
No matter the details, however, all sheet feeders can trace their heritage back to one progenitor designthe basic friction feed mechanism.
Friction Feed
When you load a single sheet of paper into a modern inkjet printer, it reverts to friction
feed, which uses the same technology as yesteryears mechanical typewriter. It moves paper
through its mechanism by squeezing the paper against a rubber drive roller. The first
inkjet printers actually used the large rubber roller of the typewriter, called a platen, and
smaller drive rollers. Friction between the rubber and the paper or other printing media
gives the system a positive grip that prevents slipping and ensures each sheet gets where
its suppose to. This friction also gives the technology its name.
In modern printers, the friction feed mechanism often goes by other names, such as cutsheet feeder and bin feed mechanism. All rely on the friction between paper and rubber rollers.
The mechanism also relies on the lack of friction to operate properly. To feed individual
sheets, the friction feed mechanism must grasp only a single sheet, which must slide
against all the others. If theres too much friction between individual sheets of paper, they
may stick together. Several may go through the printer at once, or you may seem to print
on every other sheet. Many factors in the production and packaging of paper affect
whether sheets stick to one another. In addition, the environment exerts its own effect
according to Hewlett-Packard, humidity can cause paper jams. When paper is too damp,
sheets can stick together. When it is too dry, paper may develop a static charge, which also
makes sheets stick together. Most laser printers have some means of draining the static
from paper, but sometimes it is not successful.
If you cannot change the humidity around your printer and its paper, you can ward off the
effects of sticky sheets by choosing a different kind or brand of paper. Some people recommend riffling through the paper before you load it into your printer to separate the
sheets. Riffling each ream also helps avoid the other causes of sheets sticking together.
Modern printers integrate the feed mechanism with the rest of the printer drive system.
You load cut sheets into a bin or removable tray, and the printer takes over from there.
The mechanism is reduced to a number of rollers chained, belted, or geared together that
pull the paper smoothly through the printer. This integrated design reduces complexity,
increases reliability, and often trims versatility. Its chief limitations are in the areas of
capacity and stock-handling.
947
948
Capacity
The most obvious difference between sheet feed mechanisms of the printer is capacity.
Some machines are made only for light, personal use and have modestly sized paper bins
that hold 50 or fewer sheets. In practical terms, this means that every 10 to 15 minutes,
you must attend to the needs of the printer, loading and removing the wads of paper that
course through it. Larger trays require less intervention. Printers designed for heavy network use may hold several thousand sheets at a time.
The chief enemy of capacity is size. A compact printer must necessarily devote less
spaceand therefore less capacityto stocking paper. A tray large enough to accommodate a ream (500 sheets) of paper would double the overall volume of some inkjet printers.
In addition, larger tray capacities make building the feed mechanism more difficult. The
printer must deal with a larger overall variation in the height of the paper stack, which can
challenge both the mechanism and its designer.
A printer needs at least two trays or binsone to hold blank stock waiting to be printed
upon and one to hold the results of the printing. These need not be, and often are not,
the same size. Most print jobs range from a few to a few dozen sheets, and you will usually
want to grab the results as soon as the printing finishes. An output bin large enough to
accommodate your typical print job usually is sufficient for a personal printer. The input
tray usually holds more so that you need to load it less frequentlyyou certainly dont
want to deal with the chore every time you make a printout.
Media-Handling
Most printers are designed to handle a range of printing media, from paper stock and cardboard to transparency acetates. Not all printers handle all types of media. Part of the limitation is in the print engine itself. Many constraints arise from the feed mechanism, however.
With any cut-sheet mechanism, size is an important issue. All printers impose minimum
size requirements on the media you feed them. The length of each sheet must be long
enough so that one set of drive rollers can push it to the next. When too short sheets slide
between rollers, nothing except your intervention can move them out. Similarly, each sheet
must be wide enough that the drive rollers can get a proper grip. The maximum width is
dictated by the width of the paper path through the printer. The maximum length is
enforced by the size of paper trays and the imaging capabilities of the printer engine.
In any case, when selecting a printer you must be certain that it can handle the size of
media you want to use. Most modern printers are designed primarily for standard lettersize sheets; some, but not all, accommodate legal-size sheets. If you want to use other
sizes, take a close look at the specifications. Table 26.2 lists the dimensions of common
sizes of paper.
CHAPTER 26 PRINTERS
Characteristics
TA B L E 2 6 . 2
Dimensions of Common Paper Sizes
Designation
Height
Millimeters
Width
Millimeters
Height
Inches
Width
Inches
A9
37
52
1.5
2.1
B9
45
64
1.8
2.5
A8
52
74
2.1
2.9
B8
64
91
2.5
3.6
A7
74
105
2.9
4.1
B7
91
128
3.6
5.0
A6
105
148
4.1
5.8
B6
128
182
5.0
7.2
A5
148
210
5.8
8.3
Octavo
152
229
B5
182
256
7.2
10.1
Executive
184
267
7.25
10.5
A4
210
297
8.3
11.7
Letter
216
279
8.5
11
Legal
216
356
8.5
14
Quarto
241
309
9.5
12
B4
257
364
10.1
14.3
Tabloid
279
432
11
17
A3
297
420
11.7
16.5
Folio
309
508
12
20
Foolscap
343
432
13.5
17
B3
364
515
14.3
20.3
A2
420
594
16.5
23.4
B2
515
728
20.3
28.7
A1
594
841
23.4
33.1
B1
728
1030
28.7
40.6
A0
841
1189
33.1
46.8
B0
1030
1456
40.6
57.3
949
950
Most sheet-fed printers cannot print to the edges of any sheet. The actual image area is
smaller because drive mechanisms may reserve a space to grip the medium, and the engine
may be smaller than the sheet to minimize costs. If you want to print to the edge of a
sheet, you often need a printer capable of handling larger media and then must trim each
page when it is done. Printing to (and beyond) the edge of a sheet is termed full-bleed
printing. Only a few sheet-fed printers are capable of managing the task.
Printing media also differ in weight, which roughly corresponds to the thickness of paper.
In general, laser printers are the most critical in regard to media weight. The capabilities
of a given printer are listed as a range of paper weights the printer can handlein the case
of laser printers, typically from 16 to 24 pounds (most business stationery uses 20- or 24pound stock). If you want to print heavier covers for reports, your printer needs to be able
to handle 70-pound paper. Similarly, printer specifications will reveal whether the mechanism can deal with transparency media and label sheets.
Laser printers impose an additional specification on paper stockmoisture content. The
moisture content of paper affects more than whether sheets stick together. It also changes
the conductivity of the paper. The laser printing process is based on carefully controlled
static charges, including applying a charge to the paper to make toner stick to it. If paper
is too moist or conductive, the charge and the toner may drain away before the image is
fused to the sheet. In fact, high humidity around a laser printer can affect the quality of its
printoutspale printouts or those with broken characters can often be traced to paper
containing too much moisture or operating the printer in a high-humidity environment
(which in turn makes the paper moist).
Most modern printers readily accommodate envelopes, again with specific enforced size
restrictions. As with paper, envelopes come in standard sizes, the most common of which
are listed in Table 26.3.
TA B L E 2 6 . 3
Common Envelope Sizes (Flap Folded)
Designation
Height in
Millimeters
Width in
Millimeters
Height in
Inches
Width in
Inches
6 3/4
91.4
165
3.6
6.5
Monarch
98.4
190.5
3.875
7.5
Com-10
195
241
4.125
9.5
DL
110
220
4.33
8.66
C5
165
229
6.5
9.01
CHAPTER 26 PRINTERS
Consumables
With a modern computer printer, you should expect to load envelopes in the normal
paper tray. Be wary of printers that require some special handling of envelopesyou may
find it more vexing than you want to deal with.
Collation
The process of getting the sheets of a multipage print job in proper order is called
collation. A printer may take care of the process automatically or leave the chore to you.
When sheet-fed printers disgorge their output, it can fall into the output tray in one of
two waysface up or face down. Although it might be nice to see what horrors you have
spread on paper immediately rather than saving up for one massive heart attack, face
down is the better choice. When sheets pile on top of one another, face down means you
do not need to sort through the stack to put everything in proper order.
Most printers now automatically collate by stacking sheets face down. A few have selectable paths that give you the choice of face-up or face-down output.
Duplex Operation
A duplex printer is one that automatically prints on both sides of each sheet when you
want it to. The chief advantage of double-sided printing is, of course, you use half as
much paper, although you usually need thicker, more expensive stock so that one side does
not show through to the other.
You can easily simulate duplex printing by printing on one side, turning each sheet over,
and printing on the other. When you have a multipage print job, however, it can be
daunting to keep the proper pages together. A single jam can ruin the entire job.
With laser printers, you should never try to print on both sides of a sheet except when
using a duplex printer. When printing on the second side, the heat of the second fusing
process can melt the toner from the first pass. This toner may stick to the fuser and contaminate later pages. With sufficient build-up, the printer may jam. Duplex printers eliminate the problem by fusing both sides of the sheet at once.
Consumables
Consumables are those things that your printer uses up, wears out, or burns through as it
does its work. Paper is the primary consumable, and the need for it is obvious with any
printer. Other consumables are less obvious, sometimes even devious in the way they can
eat into your budget.
You probably think you are familiar with the cost of these consumables. A couple months
after the old dot-matrix ribbon starts printing too faintly to read, you finally get around to
951
952
ordering a new $5 ribbon to hold you through for the rest of the decade. But if you buy
one of todays top-quality printerslaser, thermal wax, or dye-diffusionyou may be in
for a surprise. When the toner or transfer sheet runs out, the replacement may cost as
much as your old dot-matrix printer.
The modern trend in printer design and marketing is to follow the razor blade principle. The marketers of razors (the nonelectric shavers) discovered they could make greater
profits selling razor blades by offering the razors that use them at a low price, even a loss.
After all, once you sell the razor you lock in repeat customers for the blades.
Similarly, the prices of many inkjet and laser printers have tumbled while the consumables
remain infuriatingly expensive, often a good fraction (such as one-third) of the cost of the
printer itself. This odd situation results from the magic of marketing. By yourself you
cant do anything about it, but you must be aware of it to buy wisely.
If you truly want to make the best buy in getting a new printer, you must consider its
overall cost of ownership. This total cost includes not only the purchase price but the cost of
consumables and service over the life of the printer. Take this approach and youll discover
a more expensive printer is often less expensive in the long run.
When you have a small budget, however, the initial price of a printer becomes paramount
because it dictates what you can afford. Even in this situation, however, you should still
take a close look at the price of consumables. Two similarly priced printers may have
widely varying costs for consumables.
Cartridges
Laser printers use up a bit of their mechanism with every page they print. The organic
photoconductor drum on which images are made gradually wears out. (A new drum material, silicon, is supposed to last for the life of the printer, but few printer models currently
use silicon drums.) In addition, the charging corona or other parts may also need to be
periodically replaced. And, of course, you need toner.
Laser printer manufacturers have taken various approaches to replacing these consumables. Hewlett-Packards LaserJet printers, for example, are designed with one-piece cartridges that contain both the drum and toner. The whole assembly is replaced as a single
unit when the toner runs out. Other laser printers are designed so that the toner, drum,
and sometimes the fuser, can be replaced individually.
The makers of the latter style of printer contend that the drum lasts for many times more
copies than a single shot of toner, so dumping the drum before its time is wasteful. On the
other hand, the all-in-one cartridge folks contend that they design their drums to last only
as long as the toner.
CHAPTER 26 PRINTERS
Consumables
Surprisingly, from a cost standpoint, the choice of technology does not appear to make a
difference. (From an ecology standpoint, however, the individual replacement scheme still
makes more sense.)
A similar situation reigns among inkjet printers. Some designs incorporate the printhead
nozzles into the ink cartridge. Others make the nozzles a separately replaceable item.
Although the latter should have a cost advantage and a convenience disadvantage, as a
practical matter the differences are not significant.
A more important issue to consider with inkjets is single versus separate cartridges for ink
colors. Many printerstypically the less expensive modelsuse a single cartridge for all
three primary ink colors. If you use all three colors equally, this is a convenient arrangement. Most of the time, however, one color will run out before another and force you to
scrap a cartridge still holding a supply of two colors of rather expensive ink. If you are frugal or simply appalled at the price of inkjet ink, youll want a separate cartridge for each
ink color.
Refilling
One way to tiptoe around the high cost of laser printer consumables is to get toner or ink
cartridges refilled. Most manufacturers do not recommend thisbecause they have no
control over the quality of the toner, they cant guarantee that someone elses replacement
works right in their machines. Besides, they miss the profits in selling the toner or ink.
Quality really can be an issue, however. The Resolution Enhancement Technology of the
HP LaserJet III-series, for example, requires toner with a particle size much smaller than
that of toner used by other printers. You cannot tell the difference in toner just by looking
at itbut you can when blotchy gray pages pour out of the printer. When you get cartridges refilled, you must be sure to get the proper toner quality.
Paper
When comparing the costs of using different printer technologies, do not forget to make
allowances for machines that require special paper. In most cases, approved media for such
printers is available only from the machines manufacturer. You must pay the price the
manufacturer asks, which, because of the controlled distribution and special formulation,
is sure to be substantially higher than buying bond paper at the office supply store.
With inkjet printers, paper is another profit area for machine-makers. Getting the highest
quality from an inkjet requires special paper. Inkjet ink tends to blur (which reduces both
sharpness and color contrast) because it dries at least partly by absorption into paper.
Most inkjet printers work with almost any paper stock but produce the best results
sharpest and most colorfulwith specially coated paper that has controlled ink absorption. On nonabsorbent media (for example, projection acetates), the ink must dry solely by
953
954
evaporation, and the output is subject to smudging until the drying process completes. Of
course, the treated paper is substantially more expensive, particularly if you restrict yourself to buying paper branded by the printer-maker. For printing photos, gelatin-coated
papers yield glossy images with great color depthat a price that can reach 50 cents to a
dollar a sheet. Fortunately, inkjet printing has become so pervasive that every paper-maker
offers tailored inkjet papers and competition is driving down prices to the range of ordinary copier paper.
Interfacing
The hardware connection between your computer and printer may be the easiest to manage among all computer peripherals. Most printers, still, use the vintage printer port. A
few (and soon, more than a few) printers use a USB connection. Both of these connection
systems are covered in Chapter 11, Ports. In either case, you need do little more than
plug in the printer. All the details of the linkup are automatically taken care ofor passed
off to software.
And with the software, the interface gets interesting. The interface doesnt just have to get
signals to the printer (the hardware does take care of that), but it must ensure that the
right signals, those that the printer understands, get there. Modern software considers
every dot on the page individually and has to tell the printer what to do with it. Your computer, through its operating system and printer driver, describes what the printer should
put on paper using a command language or by sending a bit-image of the entire page to
print.
At one time the control language used by a printer determined what applications could
take advantage of all the features and the quality the printer had to offer. You had to
match your software to the command set the printer understood.
Modern operating systems have eliminated this concern. Instead, the chief issue is the
printer driver. The driver matches the printers native commands to the operating system
and all the applications that run under it. Once you install a driver, the printers command
set doesnt matter.
That said, printers usually fit three classes: Windows GDI, PCL, or PostScript.
Windows GDI printers have drivers that directly mate with the Windows Graphic Device
Interface. They use proprietary commands for eliciting each printers features. In most
cases, graphics and even the entire page may be sent through the GDI to the printer as a
bit-image. In most cases, the GDI is the fastest way of communicating with a printer. The
computer, with its powerful processor, does all the work of rasterizing the image. Most
Windows printer drivers now use the GDI.
CHAPTER 26 PRINTERS
Interfacing
PCL is the abbreviation for Printer Control Language, a command system originally developed by Hewlett-Packard for its inkjet printers and later adapted to laser machines as
well. The commands in PCL tell a specific printer what to do to make an image on a page
of paper. PCL thus focuses on how to draw the image on the page.
Although PCL is now a common language used by a wide variety of printers, it is a devicedependent language. That is, the driver for a printer may send out somewhat different
codes for another output device, even if it is another printer.
Through the years, PCL has gone through six distinct versions (four that apply to laser
printers), each building on the features of the earlier versions. PCL ships the information
to be printed as a combination of characters and graphics commands, relying on a raster
image processor in the printer.
PostScript is a page-description language developed by Adobe Systems, now in its third
major revision (PostScript 3). It describes every dot that appears on a printed page. The
language is intended to be device independent. That is, the PostScript code that yields a
printed page will be the same for all devices that understand PostScript, and not necessarily only printers. PostScript focuses on the final output, what the image on paper looks
like. How that image gets drawn doesnt matter to PostScript.
This device independence is a blessing when you need to preview something on a draftquality machine before sending it out to a production houseboth the draft printer and
the production houses typesetting equipment work from the same code. At the same time,
PostScript attaches a penalty. It requires substantial processing to convert the language
commands into a raster image of a page. Using PostScript usually results in slower
throughput from the printer.
955
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
27
Fax
Fax, short for facsimile transmissions, gives the power of Star Treks transporter system (usually without the aliens and pyrotechnics) to anyone
who needs to get a document somewhere else in the world at the speed
of light. Although fax doesnt quite dematerialize paper, it does move the
images and information a document contains across continents and
reconstructs them at the end of its near-instantaneous travels. The recipient gets to hold in his own hands a nearly exact duplicate of the original,
the infamous reasonable facsimile.
From that angle, fax is a telecopiera Xerox machine with a thousand
miles of wire between where you slide the original in and the duplicate
falls out. In fact, a product called the Telecopier was once offered by
Xerox Corporation and was the progenitor of todays fax machines.
From a computer perspective, however, fax operates as a teleprinter rather
than a telecopier, a printer that operates by remote control miles away
from your computer. You create documents using Word or some other
word-processing program and, instead of printing them on paper, send
them half a world away. In fact, if you dont have a real printer installed
in your computer, Windows assumes youll use fax as your primary output system. You can imagine fax working like a 1200-mile printer cable.
For Windows, fax is just another way of printing. Just as the Windows
GDI rasterizes images bound for laser printers, fax drivers (much like
printer drivers) rasterize them for fax transmission. Windows, using fax
drivers, calculates every dot that will appear on a sheet of paper and then
sends them outbut, instead of using the control code for a printer and
the printer interface, Windows uses standardized fax codes and your telephone line.
958
At one time, computer-based fax was a hot new technology bound to change the world.
Although the technology has cooledor at least the excitement has died downcomputer-based fax remains alive and well, still carrying out its job, only now far from the
limelight, hidden in the shadow of the Internet. Although manufacturers no longer need
fax out datasheets when they are posted on the Web, contracts and updates to blueprints
still need the fast, remote replication that only fax can provide. Although even a $500
computer is not cost competitive with a $50 standalone fax machine, no mere fax machine
has the programmable features of a computer to handle mailing lists and get the word out
with a fax blitz. Nor can a mere fax machine save and edit fax pages.
Fax and computers arent just friendstheyre coworkers. Together they expand the capabilities of both fax and computer technologies.
Technology
In a classic fax system, you start using fax by dialing up a distant fax system using a touch
pad on your fax machine, just as you would any other telephone. You slide a sheet of paper
into the faxs scanner, and the page curls around a drum in front of a photodetector. Much
as a television picture is broken into numerous scan lines, a fax machine scans images as a
series of lines, takes them one at a time, and strings all of the lines scanned from the document into a continuous stream of information. The fax machine converts the data stream
into a series of modulated tones for transmission over the telephone line. After a connection is made at the receiving end, another fax machine converts the data stream into black
and white dots representing the original image, much as a television set reconstructs a TV
image. A printer puts the results on paper using either thermal or laser printer technology.
Computer-based fax systems can do away with the paper. Fax software can take the allelectronic images you draw or paint with your graphics software and convert them into
the standard format thats used for fax transmissions. A fax modem in your computer can
then send that data to a standard fax machine, which converts the data into hard-copy
form. Alternatively, your computer fax system can receive a transmission from a standard
fax machine and capture the image into a graphics file. You can then convert the file into
another graphic format using conversion software, edit the image with your favorite painting program, or turn its text contents into ASCII form using optical character recognition
(OCR) software. You can even turn your computer into the equivalent of a standard fax
machine by adding a scanner to capture images from paper. Your printer will turn fax
reception into hard copy, although at a fraction of the speed of a standalone fax machine.
Computer-based fax beats standalone fax with its management capabilities. Computer fax
software can broadcast fax messages to as wide a mailing list as you can accommodate on
your hard disk, waiting until early morning hours when long-distance rates are cheapest to
make the calls. You can easily manage the mailing list as you would any other computer
database.
CHAPTER 27 FAX
Hardware
The concept of facsimile transmissions is not new. As early as 1842, Alexander Bain
patented an electromechanical device that could translate wire-based signals into marks on
paper. Newspaper wire photos, which are based on the same principles, have been used
for generations.
Hardware
The first fax machines looked much like current models only biggerand usually
smellier. Each had a telephone handset and a touch-tone keypad for dialing a distant
machine. Each had a slot to slide in documents and another where received pages slid
out, often a bit damp from the primitive chemical-based printing technology. Today, the
fax machine combines several other office functionsa fax modem with a scanner,
printer, and telephone setthat would tempt you to tie in your computer, but the
standalone fax machine puts all its goodies out of the reach of your computer and its
applications.
Today, standalone fax machines survive because, compared to computers, they are easier
to usethey have no need for the hassle of clicking through windows or even booting
up a computer to simply dial the telephone. Moreover, most fax machines are designed
asand perceived as by office workersordinary business machines, nothing so exotic
or esoteric as a computer. They are seen more as a telephone with a paper slot than a
brain-draining thinking machine. Even unskilled office workers are unfazed by a standalone fax machine.
A document whisking its way across the country is nothing more than digital data. Once
that simple fact registered on computer engineers, they figured out ways of linking that
data with the days best data manipulatorthe personal computer. The solution was the
fax modem. Briefly, a small industry flourished making modems that could give your computer fax capabilities. Then the technology went mainstream, and the fax modem became
a de rigueur part of every computer.
Fax machinemakers got another idea that they could give you the best of both worlds
with standalone fax machines that accepted data such as telephone lists through a cable
connection with your computer. Modern technology has done those simple connections
one bettermaybe four better. Multifunction printers often start with an ordinary fax
machine but give your computer direct access to individual fax functions. The one multifunction printer can also serve as both a printer and a scanner for your computerat
least if you dont demand graphic arts-quality scanning. Or you can slide a sheet of
paper in, press a button, and have a digital copier. At a more mundane level, you can
pick up the receiver and use the multifunction printer as an ordinary telephone. One
manufacturer has even added an answering machine to its fax printers.
959
960
Modems
In the computer realm, the term fax modem means an expansion board that slides into a
vacant slot inside your computer to give it the capability of sending and receiving fax
transmissions. Although at one time it took a specialized product to handle faxes, even the
least expensive of modern modems include fax capabilities.
The fax modem converts rasterized documents and page scans into a form compatible
with the international telephone system. Nearly every high-speed modem sold today has
built-in fax capabilities. This bonus results from the huge demand for fax in the business
world, coupled with the trivial cost of adding fax capabilities to chips used to build modern modems. Everything thats necessary for fax comes built in to the same chipsets that
make normal high-speed modem communications possible.
Multifunction Printers
The fax sections of multifunction printers are little more than ordinary standalone fax
machines. Nearly all commercial products use the same chipsets as fax modems. They differ chiefly in the capability of their standalone functions: how many numbers they remember and organize.
The printer sections of multifunction machines vary widely, nearly as widely as ordinary
printer technology. The least expensive multifunction machines use inkjet printers.
Although at one time the bargain-basement machines got away with including only monochrome printing, nearly all inkjet-based multifunction printers now have color capabilities.
That doesnt mean the machines can send and receive color images. The color serves only
the computer printer function, so you have the full spectrum available to you when want
you want to print, for example, a page from the Web.
Nothing precludes a multifunction printer from using other technologies, and a few more
expensive machines do use laser engines. You get all the advantages of a laser printer
fast, sharp images on plain paperalong with the ability to send fax directly from the
printer or through your computer. As with other general-purpose laser printers, youre
restricted to monochrome printing, both for fax and general computer output. (No manufacturer has adapted color laser technology to fax printing.) Fax standards limit the resolution of documents received by fax to 200 dots per inch. Scans and copies are limited by
the resolution of the scanner section. Output from your computer, however, prints at the
hardware resolution of the print engine.
Most multifunction printers use drum scanner technology. It is less expensive to move
paper than it is to move an internal scanner. Drum technology also permits more compact
machines and makes adding document feeders for multipage faxes much easier. A few multifunction printers use flatbed scanners. Although these are not as useful for faxing (document feeders for them are expensive add-ons), they add versatility to the scanning function.
CHAPTER 27 FAX
Group Standards
As with other flatbeds, they allow the use of small and odd-shaped documents as well as
three-dimensional objects (such as books) that would be off limits to a drum scanner.
The scanners in multifunction printers are optimized for fax use and do not pretend to
compete with dedicated scanners for producing high-quality graphics. They often have
limited resolution (some as low as 200 dpi), limited dynamic range, and sometimes no
color capabilities at all. Fax doesnt need color. The output is, however, sufficient for document processing and works well with OCR software on your computer.
To make copies, the multifunction printer manufacturers simply link the fax scanner to
the printer. The connection is indirect, detouring through the microprocessor and memory, a scenic route that adds versatility. You can make multiple copies (up to 99 on each
machine) and, with most machines, alter the size of the image. On the downside, quality is
limited by scanner resolution (no chance of counterfeiting $20 bills with a multifunction
machine), and speed is constrained by processing and the printer engine. Some machines
take nearly a minute to copy a single page, although speeds are increasing.
Group Standards
The widespread use of fax in business is a more recent phenomenon, and its growth parallels that of the computer for much the same underlying reason. Desktop computers did
not take off until the industry found a standard to followthe IBM computer. Similarly,
the explosive growth of fax began only after the CCITT adopted standards for the transmission of facsimile data.
Analog
The original system, now termed Group 1, was based on analog technology and used frequency shift keying, much as 300 baud modems do, to transmit a page of information in
six minutes. Group 2 improved upon that analog technology and doubled the speed of
transmission, up to three minutes per page.
Group 3
The big break with the past was the CCITTs adoption in 1980 of the Group 3 fax protocol, which is entirely digitally based. Using data compression and modems that operate at
up to 14,400 bits per second, full-page documents can be transmitted in 20 to 60 seconds
using the Group 3 protocol. New transmission standards promise to pump up the basic
Group 3 data rate to 28,800 bits per second.
Resolution
Under the original Group 3 standard, two degrees of resolution or on-paper sharpness are
possible: standard, which allows 1728 dots horizontally across the page (about 200 dots
961
962
per inch) and 100 dots per inch vertically, and fine, which doubles the vertical resolution
to achieve 200 by 200 dpi and requires about twice the transmission time. Fine resolution
also approximately doubles the time required to transmit a fax page because it doubles the
data that must be moved.
Revisions to the Group 3 standard have added more possible resolutions. Two new resolutions compensate for the slight elongation that creeps into fax documents when generated
and transmitted in purely electronic form. New fax products may optionally send and
receive at resolutions of 204 by 98 pixels per inch in standard mode or 204 by 196 pixels
per inch in fine mode. Two new high-resolution modes of 300 by 300 pixels per inch and
400 by 400 pixels per inch were also established. The 300 by 300 mode enables fax
machines, laser printers, and scanners to share the same resolution levels for higher quality when transferring images between them. To take advantage of these resolutions, both
sending and receiving fax equipment must support the new modes.
Data Rates
The basic speed of a Group 3 fax transmission depends on the underlying communications standard that the fax product follows. These standards are similar to data modem
standards. With the exception of V.34, data and fax modems operate under different standards, even when using the same data rates. Consequently, data and fax modems are not
interchangeable, and a modem that provides high-speed fax capabilities (say, 9600 bps)
may operate more slowly in data mode (say, 2400 bps).
The Group 3 protocol does not define a single speed for fax transmissions but allows the
use of any of a variety of transmission standards. At data rates of 2400 and 4800 bits per
second, fax modems operate under the V.27 ter standard (note that ter stands for tertiary).
At 7200 and 9600 bits per second, they follow V.29 (or V.17, which incorporates these
V.29 modes). At 12,000 and 14,400 bits per second, fax modems follow V.17. The V.34
standard will take both fax and data modems up to 28,800 bits per second. New standards
will allow the use of the Group 3 fax protocol over ISDN and other future digital telephone services.
Fax modems are typically described by the communications standards they support or by
the maximum data rate at which they can operate. Most modern fax modems follow the
V.17 standard, which incorporates the lower V.29 speeds. Most will also fall back to V.27
ter to accommodate older, slower fax products.
Compression
In a typical fax machine, you slide a page into the machine, place the call, and the machine
calls a distant number. Once the connection is negotiated, the fax machine scans the page
with a photodetector inside the machine, which detects the black and white patterns on
the page one line at a time at a resolution of 200 dots per inch. The result is a series of
bits with the digital ones (1) and zeros (0) corresponding to the black and white samples
CHAPTER 27 FAX
Group Standards
each 1/200th of an inch. The fax machine compresses this raw data stream to increase the
apparent data rate and shorten transmission times.
Data compression makes the true speed of transmitting a page dependent on the amount
of detail that each page contains. In operation, the data-compression algorithm reduces
the amount of data that must be transferred by a factor of five to ten. On the other hand,
a bad phone connection can slow fax transmissions, as fax modems automatically fall back
to lower speeds to cope with poor line quality.
Group 3 fax products may use any of three levels of data compression, designated as MH,
MR, and MMR. The typical Group 3 fax product includes only MH compression. The
others are optional, and MMR is particularly rare. To be sure that a given fax product uses
MR or MMR, you will need to check its specifications.
MH stands for Modified Huffman encoding, which is also known as one-dimensional
encoding. MH was built in to the Group 3 standard in 1980 so that a fax machine could
send a full page in less than one minute using a standard V.27 ter modem that operated at
4800 bits per second. With 9600 bps modems, that time is cut nearly in half.
MR, or Modified Read encoding, was added as an option shortly after MH encoding was
adopted. MR starts with standard MH encoding for the first line of the transmission but
then encodes the second line as differences from the first line. Because with fine images,
line data changes little between adjacent lines, usually little change in information is
required. To prevent errors from rippling through an entire document, at the third line,
MR starts over with a plain MH scan. In other words, odd-numbered scan lines are MH
and even lines contain only difference information from the previous line. If a full line is
lost in transmission, MR limits the damage to, at most, two lines. Overall, the transmission time savings in advancing from MH to MR amounts to 15 to 20 percent, the exact
figure depending on message contents.
MMR, or Modified Modified Read encoding, foregoes the safety of the MR technique and
records the entire page as difference data. Using MMR, the first line serves as a reference
and is all white. Every subsequent line is encoded as the difference from the preceding
line until the end of a page. However, an error in any one line will repeat in every subsequent line, so losing one line can garble an entire page. To help prevent such problems,
MMR can incorporate its own error-correction mode (ECM) through which the receiving
fax system can request the retransmission of any lines received in error. Only the bad lines
are updated, and the rest of the page is reconstructed from the new data. MMR with
ECM is the most efficient scheme used for compressing fax transmissions and can cut the
time needed for a page transmission with MH in half.
Instead of individual dots, under MH (and therefore MR and MMR) the bit-pattern of
each scan line on the page is coded as short line segments, and the code indicates the
number of dots in each segment. The fax machine sends this run-length coded data to
963
964
the remote fax machine. Included in the transmitted signal is a rudimentary form of error
protection, but missed bits are not reproduced when the receiving fax machine reconstructs the original page.
The exact code used by MH under Group 3 fax uses four code groupstwo for sequences
of white dots and two for sequences of black dots. Sequences from 0 to 63 dots long are
coded using terminating codes, which express the exact number of dots of the given color in
the segment. If the segment of like-color dots scanned from the paper is longer than 63
dots, MH codes it as two code groupsa terminating code and a make-up code. The
make-up code value indicates the number of 64-dot blocks in the single-color segment.
Binary File Transfer

More than just following the same modem standard, the capabilities of fax service are
merging with those of standard data communications. New fax modems, for example,
incorporate Binary File Transfer (BFT) capabilities, which enable them to ship BFT files
from one fax system to another as easily as document pages. You could, for example, send
a file from your computer to a printer for a remote printout or to a computer where it
could be received automatically. The receiving fax modem picks up the line, makes the
connection, and records the file as dutifully as it would an ordinary fax pagewithout
anyone standing around to control the modem.
Group 4
In 1984, the CCITT approved a super-performance facsimile standard called Group 4,
which allows resolutions of up to 400 by 400 dpi as well as higher-speed transmissions of
lower resolutions. Although not quite typeset quality (phototypesetters are capable of resolutions of about 1200 dpi), the best of Group 4 is about equal to the resolving capability
of the human eye at normal reading distance. However, todays Group 4 fax machines
require high-speed dedicated lines and do not operate as dial-up devices. Group 3 equipment using new, higher-resolution standards and coupled to digital services offers a lowercost alternative to Group 4.
Interface Classes
As with data modems, fax modems must link up with your computer and its software.
Unlike data modems, which were blessed with a standard since early on (the Hayes command set), fax modems lacked a single standard. In recent years, however, the Electronics
Industry Association and the Telecommunications Industry Association have created a
standard that is essentially an extension to the Hayes AT command set. The standard
embraces two classes for support of Group 3 fax communications: Class 1 and Class 2.
CHAPTER 27 FAX
Interface Classes
Class 1 is the earlier standard. Under the Class 1 standard, most of the processing of fax
documents is performed by computer software. The resulting fax data is sent to the
modem for direct transmission. It includes requirements for autodialing; a GSTN interface; V-series signal conversion; HDLC data framing, transparency, and error detection;
control commands and responses; and data commands and reception.
Class 2 shifts the work of preparing the fax document for transmission to the fax modem
itself. The modem hardware handles the data compression and error control for the transmission. The Class 2 standard also incorporates additional flow-control and station-identification features, including T.30 protocol implementation, session status reporting, phase
C data transfer, padding for minimum scan line time, quality check on received data, and
packet protocol for the DTE/DCE interface.
These classes hint at the most significant difference between computer-based fax systems,
which is software. Fax modem hardware determines the connections that can be made, but
the software determines the ultimate capabilities of the system. A fax modem that adheres
to various standards (classes as well as protocols) will open for you the widest selection of
software and the widest range of features.
965
PART 6
Infrastructure
Imagine someone asking you if youd like a Rolex for free. As soon as you nod your head and
hold out your hands, he empties a box of tiny gears, levers, cams, and a pair of watch hands
into your palms.
Its kind of a kit, he laughs, like a mad hatter after a tea party.
Thats where weve left your understanding of computer hardware. Weve talked about all the
parts but havent bothered to put them all together. Worse, we havent even discussed what
holds them all together and gives them what they need to work.
Its not jam or jelly. In this book, well call it your computers infrastructure, because it holds
everything together and supplies what you need, much like the infrastructure of a citythe
roads, utility wires, and water and sewer pipes that link you to civilization and civilized life.
More than that, its also the foundation that your computer is built upon, the metal and plastic that make up the computer that you can touch.
In the early chapters of this book, we looked at the pieces that make up the electronic circuits
that actually make your computer work. We discussed printed circuit boards in the abstract
what they do and how they are madebut stopped short of the prevailing standards that dictate how big they are and how they plug together. Thats because your computer would work
no matter how engineers fit the pieces together. As long as they take care of their assigned
functions, the computer will work.
But if they are not put together, if they lack what were calling infrastructure, they cannot
work, or at least they cannot work together. This infrastructure stuff is an afterthought, but
one thats necessary to making things work.
The most important piece of this infrastructure is the motherboard. It holds all the essential
circuitry of the computer. On it youll find the microprocessor, memory, ports, and often the
audio and often the video circuits of your computer. The motherboard is the centerpiece
where everything comes together.
The circuits that dont fit on the motherboard reside on expansion boards, which plug into
slots on the motherboard. The slots are the physical embodiment of the expansion bus. In
addition, the motherboard provides the link between the power supply and the circuits of
your computer.
The power supply is an essential element to any electronic device. It provides a steady,
refined source of electricity for every component in your computer.
Holding all this together is your computers case. Although seemingly the least-technical part
of your high-tech computer, the simple case embodies a remarkable number of functions and
technologies of its own. Not only is it tailored to fit the essential circuitry of your computer
and its peripherals, it also matches your environment and how you use your computer.
Whats more, it protects your computers circuits from that environmentand protects you
from its circuits.
Put it all together, and you have a complete computer.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
28
Principles: Hardware
Open your computer and youll see the motherboard inside. It usually
lines the bottom of a desktop system, covers one side of a tower or minitower system, or fills the center of a notebook machine like the meat of a
sandwich or ham of a burger. Its the biggest circuit board inside the
computer, likely the biggest circuit board in any electronic device you
have around your home or office. Typically, the motherboard takes the
form of a thick green sheet about the size of a piece of notebook paper
and is decorated with an array of electronic components.
Although all motherboards look much the same, that similarity belies
many differences in technologies and approaches to computer design.
Some computer-makers strive to cram as much circuitry as possible on
the motherboard. Others put as little as possible there. The difference
affects both the initial cost of your computer and its future use and
expansion.
Design Approaches
Nothing about computers requires a motherboard. You could build a
computer without oneat least if you had sufficient knowledge of digital
circuits and electronic fabrication, not to mention patience that would
make Job seem a member of the television generation. Building a computer around a single centralized circuit board seems obvious, even natural, only because of its nearly universal use. Engineers designed the very
first mass-market computers around a big green motherboard layout,
and this design persists to this day.
970
Motherboards exist from more than force of habit, however. For the computer manufacturer, the motherboard design approach has immediate allure. Building a computer with a
single large motherboard is often the most economical way to go, at least if your aim is
soldering together systems and pushing them out the loading dock. There are alternatives,
however, that can be more versatile and are suited to some applications. The more modular approach used in some of these alternatives allows you more freedom in putting
together or upgrading a system to try to keep up with the race of technology.
The motherboard-centered design of most computers is actually a compromise approach
between two diametrically opposed design philosophies. One approach aims at diversity,
adaptability, and expandability by putting the individual functional elements (microprocessor, memory, and input/output circuitry) on separate boards that plug into connectors that
link them together through a circuit bus. You can change the power and personality of
such a computer as easily as swapping boards. Such machines are known as bus-oriented
computers because everything connects through a bus, akin to an expansion bus. The alternative concentrates on economy and simplicity by uniting all the essential components of
the computer on a single large board, thus making a single-board computer. Each of these
designs has its strengths and weaknesses.
Bus-Oriented Computers
At the time the computer was developed, the bus-oriented design was the conservative
approach. A true bus-oriented design seems the exact opposite of the motherboard.
Instead of centralizing all circuitry, the bus-oriented design spreads it among multiple circuit boards. Its sort of the Los Angeles approach to computer designit sprawls out all
over without a distinct downtown. Only a freeway system links everything together to
make a working community. In the bus-oriented computer, that freeway system is the bus.
The bus approach enabled each computer to be custom-configured for its particular purpose and business. You attached whatever components the computer application required
to the bus. When you needed them, you could plug larger, more powerful processors,
even multiple processors, into the bus. This modular design enabled the system to expand
as business needs expanded. It also allowed for easier service. Any individual board that
failed could be quickly removed and replaced without circuit-level surgery.
Actually, among smaller computers that preceded the introduction of the first IBM PC in
1981, the bus-oriented design originated as a matter of necessity simply because all the
components required to make a computer would not fit on a circuit board of practical
size. The overflowing circuitry had to be spread among multiple boards, and the bus was
the easiest way to link them all. Although miniaturization has nearly eliminated such
needs for board space, the bus-oriented design still occasionally resurfaces. Youll sometimes find special-purpose computers, such as numerical control systems and network
servers, that use the bus-oriented approach for the sake of its modularity.
CHAPTER 28 PRINCIPLES: HARDWARE
Design Approaches
Single-Board Computers
The advent of integrated circuits, microprocessors, and miniaturized assemblies that put
multiple electronic circuit components into a single package, often as small as a fingernail, greatly reduced the amount of circuit board required for building a computer. By
the end of the 1970s, putting an entire digital computer on a single circuit board became
practical.
Reducing a computer to a single circuit board was also desirable for a number of reasons.
Primary among them was cost. Fewer boards means less fabrication expense and lower
materials cost. Not only can the board be made smaller, but the circuitry thats necessary
to match each board to the bus can be eliminated. Moreover, single-board computers have
an advantage in reliability. Connectors are the most failure-prone part of any computer
system. The single-board design eliminates the bus connectors as a potential source for
system failure.
On the downside, however, the single-board computer design is decidedly less flexible
than the bus-oriented approach. The single board has its capabilities forever fixed the
moment it is soldered together at the factory. It can never become more powerful or transcend its original design. It cannot adapt to new technologic developments.
Modern technology has made most of us accept the inevitable obsolescence of computers,
so the sting of the single-board approach now is much milder than with the first
machines. Computers have almost reached the point of being throwaway devices. Locking
a machine to a given level of technology is no hardship.
Notebook and sub-notebook computers all follow the single-board design for the sake of
compactness. Squeezing everything onto a single board nearly eliminates the need for
expansion boards sprouting up everywhere. By minimizing (or eliminating) connections
between boards, the single-board design also improves the reliability of notebook
machinestheres less to jostle and work itself loose.
But the single-board approach is also becoming more prevalent on the desktop. The motivation is mostly price. Putting everything on a single motherboard allows computermakers to save the cost of extra boards and connectors, which helps pare down the retail
price of machines. Many of todays computers are true single-board designs but still offer
expansion slots for later adding options.
Some manufacturers have tried using todays high-speed ports in lieu of an expansion bus
to make sealed-box computers, designed to make maintenance costs for businesses essentially zero. With nothing to service or upgrade inside, these machines have no service
costs during their operating lifetimes. So far, however, such an approach has not been a
marketplace success.
971
972
Compromise Designs
Until computers became true mass-market products, most followed a design compromise.
Rather than strictly following either the single-board or bus-oriented approach, computer-makers brought the two philosophies together, mixing the best features of the single-board computer and the bus-oriented design in one box. This was the design that
IBM chose for its first Personal Computer, which set the design for almost two decades of
desktop computers. In it, one large board hosts the essential circuitry that defines the
computer, but it relies on additional circuits on secondary circuit boards and provides further space for expansion and adaptability.
Throughout the history of the computer, functions have migrated from auxiliary circuit
boards to the motherboard to bring desktop systems closer to the ideal single-board computer design. Early computers required extra circuit boards for their serial and parallel
ports. Modern computers pack those ports and USB (and even FireWire) ports on the
motherboard. At one time, most system memory, mass-storage interfaces, high-quality
sound circuitry, network adapters, and video circuitry all required additional boards. Now
many systems incorporate all these functions on their motherboards.
At least three motivations underlie this migration: expectations, cost, and capability. As the
power and potential of personal computers have increased, people expect more from their
computers. The basic requirements for a personal computer have risen so that features
that were once options and afterthoughts are now required. To broaden the market for
personal computers, manufacturers have striven to push prices down. Putting the basics
required in a computer on the main circuit board lowers the overall cost of the system for
exactly the same reasons that a single-board computer is cheaper to make than the equivalent bus-oriented machine. Moreover, using the most modern technologies, manufacturers
simply can fit more features on a single circuit board. The original computer had hardly a
spare square inch for additional functions. Today, all the features of a computer hundreds
of times more powerful than the original IBM PC will fit into a couple of chips.
As with any trend, however, aberrant counter-trends in computer design appear and disappear occasionally. Some system designers have chosen to complicate their systems to make
them explicitly upgradable by pulling essential features, such as the microprocessor, off
the main board. The rationale underlying this more modular design is that it gives the
manufacturer (and your dealer) more flexibility. The computer-makers can introduce new
models as fast as they can slide a new expansion board into a boxmotherboard support
circuitry need not be reengineered. Dealers can minimize their inventories. Instead of
stocking several models, the dealer (and manufacturer) need only keep a single box on the
shelf, shuffling the appropriate microprocessor module into it as the demand arises. For
you, as the computer purchaser, these modular systems also promise upgradability, which
is a concept thats desirable in the abstract (your computer need never become obsolete)
but often impractical (upgrading is rarely a cost-effective strategy).
Nomenclature
Today, the compromise design survives in mainstream machines, although most motherboards delegate only their video circuitry to an expansion boardvideo being the one
place where manufacturers are constantly adding improvements that benefit (mostly)
game players. Most machines retain the capability for adding additional circuit boards,
even though most people are not apt to bother.
Blade Servers
In the computer rooms of todays businesses, a further variation on motherboard design is
making itself prominent. Called the blade server, this new design puts an entire computeractually a powerful server computeron a single board thats often the size of an
expansion board (see Expansion Boards, later). These motherboards earn their name
because the boards are wide and flat like a knife blade. The blades slide into a large board
in a rack-mounted chassis like component boards would slide into a bus-oriented computer. The bus, however, only provides power and, sometimes, a network connection
between the blades.
Blade servers have become popular because they allow several computers to fit where only
one used to. They lower costs because each server requires no cabinet or power supply of
its own. Also, individual servers can be replaced almost instantly by sliding the bad board
out and a replacement in should there be a failure.
Despite the new technology, each blade is simply a single-board computer. The only
change is that microelectronics have made its circuits more compact.
Nomenclature
More confusing than board designs is the terminology used to describe the motherboard.
The industry is rife with words that mean the same thing as motherboard or have subtle
variations from that definition. Here are some of the most common:
System boards. This term is little more than a desexed version of motherboard,
introduced to the realm of personal computer by IBM when it introduced its first
machine. At the time America was preoccupied with issues of sexual equality, and
well-meaning but linguistically naive people confused issues of gender with sex. The
term system board stands in its own right, however, because it indicates the role of the
board as the centerpiece of the entire computer system. IBM continues to use the
term to mean motherboard.
Planar boards. Although planar board may seem simply another desexed word for
motherboard, it bears a distinction. Although all modern circuit boards are planar in
the sense they take the form of a flat plane, the planar board in a computer forms
the mounting plane of the entire system.
973
974
Baseboards. Intel often refers to the motherboard as the baseboard in many of its
technical manuals. The company is not consistent about its usagefor example, the
manual dated May, 1996, for Intel VS440FX lists the product as a motherboard,
whereas the manual for the Performance/AU, dated December 1995, terms the
product a baseboard. Again, there is a subtle distinction. A motherboard goes into
a computer; a baseboard decorates the junction of a wall and the floor.
Main board. Apparently contributed by offshore motherboard-makers, the term
main board may be a result of translation, but it is actually a particularly appropriate
term. The main board is the largest circuit board inside and the foundation for the
computer system and, hence, it is the main board in a computers case.
Logic board. In the realm of the Apple Macintosh, the term logic board often refers
to the equivalent of a computers motherboardnotwithstanding, every printed circuit board inside a computer contains digital logic.
Backplane. Another term sometimes used to describe the motherboard in computers is backplane. The name is a carryover from bus-oriented computers. In early busoriented design, all the expansion connectors in the machine were linked by a single
circuit board. The expansion boards slid through the front panel of the computer
and plugged into the expansion connectors in the motherboard at the rear. Because
the board was necessarily planar and at the rear of the computer, the term backplane
was perfectly descriptive. With later designs, the backplane found itself lining the
bottom of the computer case.
Backplanes are described as active if, as in the computer design, they hold active
logic circuitry. A passive backplane is nothing more than expansion connectors linked
by wires or printed circuitry. The system boards of most personal computers could
be described as active backplanes, although most engineers reserve the term backplane for bus-oriented computers in which the microprocessor plugs into the backplane rather than residing on it. The active circuitry on an active backplane under
such a limited definition would comprise bus control logic that facilitates the communication between boards.
Expansion Boards
You know an expansion board as soon as you see one. It has an edge connector that fits
only one place: on the motherboard made to accommodate it. But the reality of the
expansion board is confused by issues of nomenclature and standards.
Nomenclature
Even the name expansion board holds a bit of confusion. Some people call them expansion
cards. Is it a card or a board? In fact, the computer industry uses the two terms
Expansion Boards
indiscriminately. This confused usage has a good precedentlong before computers,

the electronics industry used the same terms for printed circuit assemblies. If there was
any distinction, board was the generalized term (as in printed circuit board). Cards typically were smaller and usually plugged into a connector. Boards were usually screwed or
bolted in place. No one cared, because both terms were generally understood.
However, the expansion boards used by notebook computers truly are cards, even though
their printed circuit cards (or boards) are safely sealed inside their metal cases. They are
called cards because of their similarity to credit cards. The names of the standards for
them explicitly make them cards: PC Card and CardBus. These names and the similarity
to credit cards were not accidentalthe little expansion boards were designed to be a
familiar and friendly credit-card size.
In the computer industry, several terms describe the boards used for expanding computers, sometimes with subtle differences in design or technology. Some of these terms
include the following:
Expansion boards. The smaller printed circuit boards that plug into your computers motherboard are most often termed expansion boards because they provide you
with the means of expanding the capabilities of your computer. As noted before, the
expansion board is distinct from the expansion slot, the space inside the computer
chassis the board occupies (or potentially occupies), and the expansion connector
into which you plug the board.
Expansion boards are often distinguished by the standard followed by their interface
or the connector at the bottom of the board. For example, an ISA board follows the
Industry Standard Architecture bus standard, and a PCI board follows the Peripheral
Component Interconnect standard. Well discuss these standards a bit later.
Option boards. Some computer-makers prefer to describe expansion boards as
option boards. You plug them into your system to add an optional feature. Strictly
speaking, then, a standard equipment expansion boardfor example, a graphics
adapterwould not be an option board, but for consistencys sake (or maybe inconsistency), most manufacturers include such standard equipment among their options
boards, perhaps to give you the idea youre getting options for freejust like that
lunch.
Daughter boards. Strictly speaking, any board that plugs into a motherboard
should be a daughter board, but in the realm of the computer, the family relationship
is not so straightforward. Many boards that plug into the motherboard of a computer have special names of their ownmemory modules, microprocessor cartridges, and expansion boards are all daughter boards. However, most computer
hardwaremakers reserve the term daughter board for add-on circuit boards that
attach as a second layer to their expansion boards.
975
976
This two-story form of packaging was prevalent when all the circuitry needed to
build an expansion board just wouldnt fit in the space available in a single slot. The
daughter board bought added square inches for circuitry. Todays circuits are so
compact that this form of construction is rarely used. Most manufacturers dont even
use the entire allowable size for their expansion board products.
Riser boards. As noted previously, low-profile computers reduce their size by providing horizontal slots for their expansion boards. To connect these boards to the motherboard, most use a special board called a riser board. As with an ordinary expansion
board, the riser board plugs into the motherboard, but its circuit endowment comprises little more than a set of connectors to accommodate your expansion boards.
Although computer expansion boards can all be considered daughter boards, not all
daughter boards are expansion boards. For example, some computer expansion
boards can themselves be expanded by plugging a daughter board onto them.
Because such boards plug only into their host board, they are not true computer
expansion boards. Most people call the circuit boards that plug into the motherboard the systems expansion boards. Circuit boards that plug into expansion boards
are daughter boards. That convention at least relieves us of adding another generation and creating the granddaughter board.
Construction
As a concept, an expansion board might be almost anything as long as it can fulfill its purpose of enhancing the capabilities of the motherboard. As a practical matter, however, the
expansion board has evolved to become a printed circuit board that fits into a connector in
the motherboard and a space inside the computer chassis that makes up an expansion slot.
This size and shape of an expansion board is entirely arbitrary. The expansion board could
be as large as all creation or as small as a single chip.
Of course, expansion boards are not. Standards have been set on the size of expansion
boards to make them interchangeable so that boards from different manufacturers will fit
in as many computers as possible.
These standards are not entirely arbitrary. Several factors have influenced the choice of
size. For example, an expansion board cannot be larger than the space provided by the
expansion slot; otherwise, it would not fit and could hardly fulfill its expansion function.
The board has to be large enough to hold the circuitry it needs to do what it has to do.
Manufacturers prefer smaller boards because they cost less to make. However, the board
cant be too small or it cannot hold the required expansion connector.
Several components define the physical reality of the standard computer expansion board.
The board proper is an ordinary printed circuit board fabricated with pin-in-hole or surface-mount technology or a combination of both. An expansion connector connects the
Expansion Boards
board to the electronic circuitry of your computer. A retaining bracket secures the board
inside your computer and provides a place to put peripheral connectors.
Substrate
The substrate is the board itself, a slice of glass-epoxy upon which the various circuit
components and connection traces are bonded. Computer-makers fabricate expansion
boards using exactly the same technologies as motherboards.
The original design for expansion boards envisioned one end of each board sliding into a
card guide, a thin slot at one end of the expansion slot, to stabilize the board inside the
computer and keep it from bending or flapping in the breeze. These expansion boards
that stretch from one end of the slot to the other are often called full-length expansion
boards.
Most modern expansion board designs dont require all the area allowed for the substrate
in the computer and are classed as short cards. Because of their diminutive dimensions and
low mass, they are adequately secured in your computer by the expansion connector and
their retaining brackets.
Retaining Bracket
Nearly all expansion boards have an L-shaped bracket attached at one end. Manufacturers
use a number of terms for this bracket, perhaps the most colorful being ORB, an
acronym for option retaining bracket. The current trend is to refer to this bracket as the
bracket.
In a computer, the bracket serves two functions. It secures and stabilizes the expansion
board in its slot. It provides a mounting space for port connectors that may be required
for connecting peripherals to the expansion board, and it helps shield your computer,
keeping electrical interference inside your computers case by plugging up the hole at the
end of the expansion slot.
In most computers, a screw secures the bracket to the computers chassis. When installing
an expansion board, you should always ensure that this screw tightly holds each expansion
board in place. Properly installing each board with a screw will prevent you from accidentally pushing the board out of the expansion connector when you plug into the connector
on the board. (Tilting the expansion board can cause the contacts on its edge connector to
bridge across several pins of the expansion connector, thus shorting them out and possibly
crashing or even damaging your computer.) In addition, firmly screw-mounting the
bracket ensures electrical continuity between the bracket and computer chassis.
Connector
The card-edge connector on each expansion board is little more than an extension of the
etched copper traces of the printed circuits on the board substrate. The chief difference is
977
978
that the connector pads are gold plated during the fabrication of the expansion board.
The gold does not tarnish or oxidize, so it ensures that the edge connector will make a
clean contact with the expansion connector on the motherboard.
The chief current expansion board standard uses the placement of pad areas and slots to
key the board so that expansion boards fit only in slots designed for them.
Nearly all expansion standards for desktop computers use edge connectors for one very good
reason: They are cheap. The connectors contacts get etched onto the board at the same
time as the rest of its circuit traces. The only extra expense is the thin gold plating on the
contact area to stave off the oxidation of the copper or lead-and-tin-coated traces.
After the pragmatic choice of an edge connector, the creator of an expansion standard still
has a variety of choices. Most important is the spacing between contacts in the connector.
The spacing, along with the number of contacts, determines the size. In a dream world
the one in which many designers operatethe size of the connector is no concern. In the
real world, however, it has two dramatic effects. It determines how much space must be
given up to connectors on the motherboard, and it governs the insertion force of an
expansion board into the connector.
In true bus-style computers, the board space given up to the expansion bus is immaterial.
The computer chassis is nothing but the bus, so there is no problem in devoting the
whole back or bottom of the machine to the expansion bus. In the traditional computer
design, however, the bus takes up space on the motherboard, which also has to provide the
basic circuitry of the computer. The more motherboard space taken up by the bus, the less
is available for building the basic computer. Consequently, the bus area must be as compact as possible to yield the largest possible circuit space on the motherboard.
The larger the connector, the more area of the expansion board that rubs against the contacts inside the socket when you plug the board in. To ensure a reliable electrical connection, the socket contacts must press forcefully against the contact tabs on the circuit
board. Sliding an expansion board into a socket requires enough force to squeeze the
board contacts between the socket contacts. The more area devoted to contacts, the
greater the required insertion force. When a connector is too long, the insertion force may
be greater than some people can comfortably apply, even greater than the automatic insertion machinery used by computer manufacturers can apply. Worse, if the insertion force is
high enough, sliding in an expansion board may overly stress the motherboard, potentially
cracking it or one of the conductive traces and putting it out of action.
Making the contact smaller shortens the connector, cutting down on the motherboard
space required for the bus and reducing insertion force. It also requires greater precision
in the manufacture of expansion boards. Nevertheless, newer expansion board standards
are marked by closer spacing of their edge connector contactsjust compare an ISA
board to a PCI board inside your computer.
Cases
More specialized bus standards, such as those for notebook computers and industrial computers, rely on pin connectors. These necessarily cost more because they add another part
that must be soldered to each expansion board. But because the connector is a separate
part, it can be manufactured with greater precision, and the greater precision allows for
smaller, more compact connectors. Moreover, pin connectors can use more than two contact rows to further reduce the space that must be devoted to the bus. Pin connectors are
also more reliable because they allow their contacts to mate on multiple sides. Pin connectors are also easier to shield, making them more desirable as concerns about emissions
increase along with bus speeds.
Power
The infrastructure that makes a city includes not only highways and mass transportation
but also utilities, which supply businesses and homes with what they need in the modern
world: water, sewers, gas, and electricity.
Computers, too, need utility supplies to keep them going. As mere electronic beasts, however, they need neither water nor gas, and whatever sewage they create usually gets stored
on their hard disks rather than carted away. Electricity is the sole necessity of the computer.
Not just any electricity will do. Computers have a definite and strict requirement for the
power they use, which fortunately for us corresponds to ordinary utility power that you
get at your wall outlet. The circuits inside your computer would disappear in a flash if
utility power got to them, however. They require a special diet, one that the infrastructure
of your computer must create for them.
The power supply inside your computer converts utility power into a form thats safe for
computer circuitry. The infrastructure of your computer distributes this power, both to
the motherboard and to the various disk drives you install inside your computer.
Cases
The final aspect of your computers infrastructure is what truly holds everything
togetherits case. Although it appears to be nothing more than a box, the simple case has
several complex functions. It is both the foundation for your computer and a protective
shell that guards against both physical and invisible dangers. More importantly, the case is
the one part of your computer that you see all the time. It is your pride and joy. The case
is the physical embodiment of your computer.
In fact, the case is the body of your computer. Its a housing, vessel, and shield that provides the delicate electronics of the computer a secure environment in which to work.
Cases come in various sizes, shapes, and effectiveness at their protective tasks to match
your computer and the way you plan to use it.
979
980
In its protective role, the case guards the delicate circuitry inside your computer. Part of
that protection is physical, guarding against physical dangersforces that might act
against its circuit boards, bending, stressing, even breaking them with deleterious results
to their operation. It also prevents electrical short circuits that may be caused by the infall
of the foreign objects that typically inhabit the officepaper clips, staples, letter openers,
beer cans, and errant bridgework. The case also guards against invisible dangers, principally strong electrical fields that could induce noise that would interfere with the data
handling of your system, potentially inducing errors that would crash your system.
The protective shield of the case works both ways. It also keeps whats inside your computer inside your computer. Among the wonders of the workings of a computer, two in
particular pose problems for the outside world. The electrical voltages inside the computer can be a shocking discovery if you accidentally encounter them. And the high-frequency electrical signals that course through the computers circuits can radiate like radio
broadcasts and interfere with the reception of other transmissionsincluding everything
from television to aircraft navigational beacons.
Your computers case also has a more mundane role. Its physical presence gives you a place
to put the things that you want to connect to your computer. Drive bays allow you to put
mass-storage devices within ready reach of your computers logic circuits while affording
the cases protection to your peripherals. In addition, your computers case provides the
physical embodiment of the expansion slot, affording the boards that slide into the connectors of your computers expansion bus protection with the same mechanical and electrical
shelter as the rest of the system.
The case can play a more mundane role, too. It also can serve as the worlds most expensive monitor stand, raising your screen to an appropriate viewing angle, elevated high
above the clutter and confusion of your desktop. Or a tall, desk-side computer can be an
impromptu stand for your coffee or cola cup or a more permanent residence for the
papers that spillover from your desk. Appropriately chosen, the right notebook computer
can be a gimmick pick-up device that can win you datesproviding you hang with the
right (or wrong, depending on your perspective) crowd.
Compounding the function of your computers case is the need to be selective. Some of
whats inside your computer needs to get outheat, for instance. And some of whats outside needs to get insuch as signals from the keyboard and power from your electrical
outlets. In addition, the computer case must form a solid foundation upon which your system can be built. It must give disk drives a firm base and hold electrical assemblies out of
harms way. Overall, the simple case may not be as simple as you think.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
29
Motherboards
This is where it all comes together. Everything weve talked about so far
in this book is either part of, connects to, or comes from the motherboard. Without a motherboard, you would not have a computer. Youd
just have a box, a few leftover parts, and a high degree of frustration. You
wouldnt even have a jack to plug into.
Not only does the motherboard make your computer, but it also defines
it. The components and circuits on the motherboard determine exactly
what your computer can do and how fast it runs.
Nearly all computers and compatible computers share one common feature: They are built with a single, large printed circuit board as their
foundation. In many cases, the big boardusually called the motherboardessentially is the entire computer. Almost completely self-contained, the one board holds the most vital electronic components that
define the computer: its microprocessor, support circuitry, memory, and
often video and audio functions. Anything you want to add to your
computer plugs into the expansion bus thats part of the motherboard.
As such a basic element, the motherboard defines both the computer
and its capabilities. The circuitry it contains determines the overall performance of your system. Without the motherboard, you wouldnt have
a computer.
In a modern computer, the motherboard is the big green centerpiece inside
the case. Each computer-maker essentially builds the rest of its computer
around the motherboard. On the motherboard, the computer-makers put
all the most important electrical circuits that make up the computer. The
expansion bus on the motherboard provides a foundation for future expansion, adding new features and capabilities to your computer.
984
Physical Requirements
Besides holding the essential circuitry of a computer, the motherboard of a computer must
accommodate some form of expansion. In desktop computers, the motherboard is home
to special electrical jacks called expansion connectors that allow you to plug in additional
printed circuit boards. The space potentially occupied by an expansion board is an expansion slot, usually referred to simply as a slot.
Notebook computers and their ilk also incorporate expansion boards but in a different
form. A protective shell, usually sheathed with aluminum, encases their circuit boards to
make a near monolithic assembly thats termed a PCMCIA card or PC Card. PCMCIA
stands for the organization that sets the standard for these cards, the Personal Computer
Memory Card International Association. PC Card is actually a term of art that more particularly describes one of the interface standards used by these cards. Most people call the
space potentially occupied by the cards a PCMCIA slot or simply a card slot.
Manufacturers cannot designate expansion areas on their motherboards willy-nilly. There
is order and reason behind their designsall backed by standards.
Beyond standardization, other concerns include the number, size, and arrangement of the
slots. These considerations determine how expandable a given computer really is. In addition, the spacing of the bus connectors in the slots is a concern. Put them too close
together, and youll limit the designers choice of circuit components to those that are
short or force the designer to put the components flat against the board, thus wasting its
expensive space. Even the number of expansion connectors may be set or limited by the
bus standard.
Bus Standard
The basic characterization of motherboards is by the standard they follow for the physical
and electrical characteristics of their expansion slots. The choice dictates which boards
plug into the motherboardin other words, which products you can use to upgrade your
computer. Although the choice is no longer wide, variations remain.
Expansion standards for desktop computer motherboards have gone through a lengthy
evolution, with a number of interesting but now essentially irrelevant side trips. Today,
however, one standard (PCI) dominates, and the only option is whether a given motherboard or computer condescends to accommodate legacy (old) expansion boards that follow
the ISA standard. The situation is simpler among notebook computers because their evolution has been shorter and they missed the chaotic early days of computer development.
Youll find a complete discussion about the history of expansion standards in Chapter 9,
Expansion Buses.
CHAPTER 29 MOTHERBOARDS
Physical Requirements
Slot Number
The number of expansion slots in a given computer is a compromise with multiple considerations. More is better, but a computer with an infinite number of slots would be infinitely wide. The practical dimensions of the motherboard and case limit the space
available for slots.
Electrical considerations also keep the slot count modest. The high-speed signals used by
todays expansion buses are hard to control. The higher the speed, the fewer slots that are
practical. Even at 33MHz, todays most popular expansion bus standard allows for only
three slots per controller.
Adding more slots requires additional control circuitry, which adds to the price of the
computer (as does the cost of the expansion connectors themselves). With manufacturers
slugging it out over every penny of their prices, the tendency is to constrain costs and slot
count.
How many expansion slots you need on a motherboard depends on how many functions
the motherboard-maker has integrated into the board. Most practical computers fill one
or more of their slots with standard equipment (such as a video board and modem). Over
the three-to-five-year life of the typical computer, youre likely to add more than one new
accessory to it, possibly something you might not have conceived of when you bought the
machine. Therefore, you need to plan in advance for the need for expansion.
Notebook computers get away with two or fewer slots because they encapsulate all the
normal functions of a computer on their motherboards. Nevertheless, two is still better
than one, and youre likely to use whats available. Note, too, that super-thin computers,
which are most likely to have single slots, are the ones that will more likely need two. For
example, these machines may require a slot to run a CD drive that would otherwise be
built in to a thicker system.
Slot Spacing
Expansion boards are three-dimensional objects (as is everything else in the real world). In
addition to length and width, they also have thickness. PCMCIA explicitly gives all three
dimensions of conforming cards. The thickness of desktop computer expansion boards is
more implicit, determined by the spacing of the expansion slots. If slot spacing is too narrow, boards simply wont fit. Worse, one board might touch another and short-circuit a
computer signal, leading to erratic operation or complete non-operation of the computer.
If slot spacing is too wide, fewer will fit in a computer of reasonable size, thus limiting
your expansion options.
A printed circuit board itself is quite thin, about one-eighth inch, and can be made even
thinner when board dimensions become small. The thickness of most expansion boards
985
986
and the requirement for adequate slot spacing arise from the components installed on the
boardwhich may include another printed circuit board clinging to the first like a
remora.
Surface-mounted components make thinner boards practical, but taller components
remain prevalent enough that the spacing of slots has remained the same since 1982. On
all motherboards, expansion connectors are located on 0.8-inch centers.
Not only does this standard set the maximum thickness of any expansion board, it also
dictates the number of slots that may be available for a given size of motherboard. The
board must be as least wide enough to accommodate all the slots it is to hold. Some motherboards are hardly wider than that.
In any case, the spacing of expansion slots was originally set arbitrarily. It represents what
the developers of the first computers thought was the optimum compromise between
compact layout and adequate allowance for the height of circuit components. The choice
was wise enough that it still reigns today.
Slot Layout
In desktop computers, the layout of the slots is mostly a concern of the system designer. As
long as your boards fit, you shouldnt have to worry. The designer, however, must fret about
the electrical characteristics of signals that are close to invading the territory of microwave
ovens. Some are purely electronic issues that cause high-speed buses to operate erratically
and interact detrimentally with other electronic circuits by generating interference.
Expansion buses pose more engineering problems than other circuits because they require
the balancing of several conflicting needs. The expansion bus must operate as fast as possible to achieve the highest possible transfer rate. The expansion bus also must provide a
number of connectors for attaching peripherals, and these connectors must be spread apart
to allow a reasonable thickness for each peripheral. In other words, the bus itself must
stretch over several inches. The common solution is to use several short buses instead of
one long one. Although the connectors on the motherboard may appear to link all expansion boards into a single bus, if your system has more than three high-speed expansion slots,
it likely has multiple buses, each with its own controller, powering the slots.
Theres another design consideration in desktop and tower computerswhether expansion boards slide into their slots vertically or horizontally. In full-size systems, expansion
boards plug into the motherboard perpendicularly. The result is that the main plane of
each expansion board is vertical in desktop computers and horizontal in tower-style
machines. A few compact desktop computers, termed low-profile designs, align expansion
boards parallel to the motherboard, horizontal in desktop systems.
Standards
Although alignment of the expansion slots appears to be only an aesthetic consideration, it

also has a practical importance. Vertical boards cool naturally through convection. Air
currents can rise across each board and cool its innermost components. Horizontal boards
defeat convective cooling because the boards themselves block vertical air currents. In
general, computers with horizontal expansion boards should have some kind of active
cooling systemtranslated from technical language, that means they need a fan to blow a
cooling breeze across them.
Slot layout is more complicated in notebook computers in which the layout of PCMCIA
slots may determine what cards you can plug into your computer. Most notebook computers have two slots, and they are arranged like the barrels of a shotguneither in a side-byside or over-and-under fashion. These options are not functionally equivalent. PC Cards
come in any of three thicknesses (discussed later), and most slots accommodate the two
slimmer cards. An over-and-under arrangement of two slots will hold either two slim
cards or one of the thickest sort. Although the over-and-under configuration gives you the
largest number of expansion options, many of todays ultra-thin notebooks can only
accommodate side-by-side slot mountings.
Standards
Modern motherboards can be any size or shape that suits the design of the computer.
Only the need for standard-size expansion slot connectors limits the freedom of design.
Many computers, however, are built around motherboards of a few standard sizes. This
standardization is a matter of convenience. It allows a computer manufacturer flexibility in
the choice of suppliersstandardized dimensions make motherboards interchangeable.
For you as the purchaser of a new computer, motherboard standardization has its downside. You face the problem of the computer manufacturer selling systems equipped with a
motherboard du jour, whatever OEM motherboard was available cheapest on the day the
computer was put together. On the other hand, a computer built around a standard-sized
motherboard gives you upgrade freedom. Should you become dissatisfied in the performance of your computer, you can replace a standard-size motherboard with a more powerful one.
The earliest motherboard standards followed the leads set by IBM. They duplicated the
physical dimensions of the motherboards used by the most popular IBM machines. Even
when they lopped off vast areas of board to trim costs, most manufacturers retained compatibility with IBMs designs, keeping mounting holes in the same locations so that one
board could be substituted for another. This heritage continues even in some of the latest
designs.
987
988
After IBM ceded its influence as the setter of the standard dimensions of motherboards,
the industry was essentially adrift. Major manufacturers developed their own designs without regard to older products, while smaller manufacturers clung to the old board layouts.
The situation is changing, though, with new motherboards standards now promulgated by
Intel. The most recent of these, the ATX motherboard design, goes further than ever before
and specifies not only dimensions and mounting holes but also connector placement and
even connector designs. To standardize the motherboards of more powerful computer workstations, the computer industry recently developed and then dropped the WTX specification.
Motherboard design divergence first arose among makers of small-footprint computers,
machines designed with smaller dimensions to cover less of your desktop. These machines
compromised expansion by reducing the number of expansion slots and drive bays to gain
their more modest measurements. Many manufacturers developed low-profile computers
that reduced system height by turning expansion boards on their sides. These designs
necessitated changes from the more standardized motherboard layout. The computer
industry has rallied around the low-profile design and produced two standards that support the concept. First came the LPX motherboard, and then came a smaller derivative, the
Mini-LPX. To accommodate the needs of new microprocessor and memory technologies
(and to add a wealth of other features), the industry adopted the NLX motherboard.
In general, smaller manufacturers are more likely to use standard-size motherboards.
Larger manufacturers are better able to afford the price of custom-designing cases and
motherboards to match. Even large computer-makers have moved to standard-size boards
at least for their offerings that take advantage of the latest microprocessors. In truth, systems that use the latest microprocessors often all have exactly the same motherboard
design inside, using a motherboard designed and manufactured by Intel. Even large manufacturers may rely on Intel motherboards until their engineers develop a familiarity with
new chips. For example, all but a handful of the first Pentium Pro computer models uniformly used standard-size Intel-manufactured motherboards.
ATX
To bring a degree of uniformity to motherboard design, the computer industry created a
new motherboard standard that roughly conforms to the Mini-AT board size but with a
few design twists that result in lower-cost engineering. Called ATX, the standard is promulgated by Intel but is openly published. Intel released the most recent version, 2.1, in
February 1996, fine-tuning the design based on industry feedback.
The ATX standard defines the number and position of the motherboard mounting holes
and offers recommendations as to component, expansion board, and port connector placement. Although the standard does not demand any particular slot type or configuration,
its aimed primarily at ISA, PCI, and ISA/PCI combination designs. It also allows for both
5.0- and 3.3-volt system operation (or both simultaneously).
Standards
The odd orientation of the board facilitates port placement. It provides the maximum
space for expansion boards and port connectors at the rear of the host computer chassis.
The design also envisions that the microprocessor will be located near the right edge of
the board, where it will be in close proximity to both the power supply and the cooling
fan. In the recommended configuration, memory sockets can be readily accessed between
the microprocessor and expansion slots.
The ATX board itself measures a maximum of 12 by 9.6 inches (305 by 244 millimeters).
This size is not a random choice but, according to Intel, was selected to allow manufacturers to cut two boards from a standard-size 18-by-24-inch raw printed circuit panel. It provides sufficient space for about seven expansion slots, which are spaced at the conventional
0.8 inches apart. It incorporates nine mandatory and one optional mounting holes, most
of which are in the same positions as the holes in a Mini-AT motherboard. Figure 29.1
shows the dimensions and mounting hole placement for an ATX motherboard.
FIGURE 29.1
12 inches
Dimensions and hole

placement of an ATX
motherboard.
9.6 inches
Optional
screw hole
Microprocessor
Expansion slots
Memory sockets
The ATX specification goes further than simply indicating mechanical board dimensions.
The standard also embraces the PS/2 size of power supply and specifies a new motherboard power connector (see Chapter 31, Power).
Besides uniformity, the ATX design aims at trimming costs for computer-makers. By
putting port connectors on the motherboard, even in multiple layers, the cost of connecting cables as well as the labor required for assembly is eliminated. Eliminating cables also
helps minimize potential radio frequency interference. In its recommended configuration,
the ATX layout also allows the use of shorter floppy and hard disk connecting cables, with
similar benefits. The power supply choice and location also trims cost for the computer
manufacturer as well as helps the computer run cooler and even quieter.
989
Mini-ATX
The designers of the ATX board realized that the one certainty in computer circuit design
is that functions get combined and made more compact. As more and more of the functions of a computer squeezed into one or two chips, the ATX designers imagined that
soon much of the ATX real estate would be superfluous. In that one of the primary goals
in the design of ATX was trimming costs, they figured that trimming motherboards to a
size smaller than ATX as the technology permitted would reap savings in materials cost.
Consequently, they included a standard size for Mini-ATX motherboards in the original
ATX specification.
The Mini-ATX design chops the ATX motherboard down to 11.2 by 8.2 inches (284 by
208 millimeters). When installed in a computer, the Mini-ATX motherboard still sits at
the rear edge of the chassis so that port connectors can be mounted directly to it without
cables. In most chassis, the left edge still aligns with the left side of the case to allow space
for a full complement of expansion boards. Because of this placement, the smaller size of
the Mini-ATX board cuts off one row of ATX mounting holes. As a result, the lower row
of mounting holes is displaced on the Mini-ATX design, as shown in Figure 29.2.
FIGURE 29.2
11.2 inches
Mini-ATX motherboard dimensions

and layout.
Optional
mounting hole
8.2 inches
990
Microprocessor
Expansion slots
Memory sockets
microATX
The Mini-ATX design has one chief benefit. It reduces the materials costs for a motherboard by about 30 percent when compared to a full-size ATX board. But that wasnt
enough for the creators of the ATX specifications. They created another small-size motherboard called microATX, which took over the role of Mini-ATX. The current incarnation
of the standard, version 1.0, was published in 1997.
Standards
The microATX specification represents more rethinking than a simple extra shrinking.
The new design is actually a bit larger than Mini-ATX. A microATX motherboard measures 9.6 inches (244 millimeters) square.
The chief difference between Mini-ATX and microATX is how the two mate with the
full-size ATX standard. Mini-ATX motherboards pay little heed to the larger standard;
microATX boards fit neatly inside it. Although microATX motherboards are narrower
than full-size ATX designs, they use nearly the same mounting scheme, so a microATX
board will fit into a full-size chassis. Expansion slots appear in the same position, except
microATX allows for fewer slots because of its reduced sizea maximum of four slots,
typically three PCI and one AGP. (ISA slots are also allowed.) In addition, most of the
screw holes used for mounting a microATX motherboard match the positions of those on
a full-size ATX motherboard.
To mate with the conventional ATX chassis, the microATX motherboard reserves the
same space for port connectors as does the full-size board, and the two boards use exactly
the same power connectors. Figure 29.3 illustrates the microATX motherboard and compares it to its full-size sibling.
FIGURE 29.3
The microATX motherboard layout.
Full-size ATX (12 inches)

microATX (9.6 inches)
FlexATX
If small is good, even smaller is better. In 1999, Intels engineers unveiled yet another
motherboard format, FlexATX, which shaves inches off the microATX design. The name
reflects the design goal: FlexATX is meant to be flexible and allows for the development
of devices beyond the traditional personal computer (but based on the traditional personal
computer motherboard). In the classification system of its promoters, FlexATX is not a
standalone standard but an addendum to the microATX specification.
991
992
A FlexATX motherboard measures no more than 9 inches wide and 7.5 inches deep. It
retains the same area (in both size and location) for ports and connectors as that used by
ATX and microATX motherboards.
Seeing a wide application of compact motherboards, however, the FlexATX design fits
inside the conventional ATX chassis. The placement of its mounting holes matches those
of an ATX motherboard, although (as with microATX) two added holes provide mounting
security on the edge adjacent to where excess glass-epoxy was pared off.
The FlexATX design does not specify the number and location of expansion slots because
the creators of the standard envisioned its use in systems that provide no board-style
expansion (although the standard does permit engineers to use their imagination and add
slots to the design).
Table 29.1 lists the dimensions for ATX motherboards and those derived from that standard. The complete ATX specifications and those related to it are available on the Web at
www.formfactor.org.
TA B L E 2 9 . 1
Dimensions of Modern Motherboards
Motherboard
Width
(in)
Length
(mm)
Area
(in)
Area
(mm)
Area
(Square in)
Area
(Square cm)
ATX
12.0
305
9.6
244
115.2
744.2
Mini-ATX
11.2
284
8.2
208
91.8
590.7
microATX
9.6
244
9.6
244
92.2
595.4
FlexATX
9.0
229
7.5
191
67.5
437.4
NLX
Announced in September 1996, by Intel, the NLX motherboard design was a cooperative
effort among several manufacturers, including IBM. At the time of its introduction, 12
computer-makers and one motherboard-only manufacturer (ASUStek) had announced
support of the specification. The computer-makers included AST, Digital Equipment
Corporation (now part of Compaq), Fujitsu, Gateway 2000, Hewlett-Packard, IBM, ICL
Personal Computers, Micron Electronics, NEC Computer Systems, Sony, Toshiba, and
Tulip Computers.
NLX is an improved low-profile layout. It was created specifically to overcome some
shortcomings of earlier, no longer supported designs (LPX and its reduced-size version,
Mini-LPX), in that those designs interfered with adapting to the latest technologies. NLX
is meant to support all current Intel microprocessor designs and memory technologies as
well as the advanced graphics port (AGP) for high-speed interconnections with video
Standards
boards. In addition, the NLX design enhances the physical packaging of systems to allow
greater mechanical integrity, better cooling, and more space for peripheral ports.
Key to the NLX design is the riser board. Although its similar to the riser boards of LPX systems, because you plug expansion boards into it, under the NLX design the riser takes on a
greater role. In effect, it operates as a backplane and the NLX motherboard itself is a glorified
processor board. The riser board attaches permanently to the chassis of its host computer
while the NLX motherboard readily slides in and out of the case like a steroid-enhanced
expansion board. In purest form of the NLX implementation, all cables in the systemincluding the power supplyattach to the riser, and none attach to the motherboard. Typically the
cables for the floppy disk and hard disk drives in a computer will plug into the riser.
Although the design of the NLX motherboard is fixed by the specification, the riser board
is not. Computer-makers can customize its design to accommodate different system
designs. Although the typical riser board has four expansion connectors, the NLX specification allows a great deal of freedom to accommodate not only low-profile computers but
also tower-style systems. The specification describes signals for up to five PCI expansion
slots and an unlimited number of ISA slots on a single riser. All boards slide into the riser
parallel to the motherboard.
The one aberration in the NLX motherboard/riser design is its accommodation for AGP
video boards. The specification reserves a special area on the left side of the motherboard,
opposite the riser board, for a single AGP slot. The AGP board slides into the motherboard slot parallel to the riser board. Microprocessors (the NLX design accommodates up
to two) and memory reside on the right side of the motherboard so that they do not interfere with expansion boards. Figure 29.4 shows the basic layout of this two-board system.
FIGURE 29.4
NLX motherboard
layout.
993
994
An L-shaped integral rear panel of the board provides space for peripheral connectors.
The higher part of the panel allows designers to stack multiple connectors, one over the
other, on the right side of the chassis, away from expansion boards. The NLX design
envisions the motherboard to be readily removable without removing an expansion board
(except the AGP board) from inside the computer case. In a typical computer, the motherboard slides out the side of the chassis, guided by rails at the bottom of the chassis. Four
screws inside the computer secure the board to the chassis and electrically ground the two
together. A latch thats part of the chassis and under the motherboard holds the board
horizontally in place and also serves as a board ejector to aid in removing the motherboard. Spring-like contact fingers around the periphery of the rear panel shield electrically
integrate the board with the rest of the chassis.
The mechanical specification of NLX motherboards allows for a small degree of freedom
in their size. The specs allow any width between eight and nine inches (inclusive), except
for systems integrating AGP video boards. These must be nine inches wide because the
space reserved for the AGP connector is on the last inch of the board. NLX motherboards
may be between 10.0 and 13.6 inches long. The smallest NLX motherboard is approximately the size of a mini-LPX motherboard; the largest about the size of an LPX board.
The NLX specifications provide for three different mounting screw patterns, depending
on the length of the board. An NLX motherboard may use any of the three patterns, but
each NLX chassis design must accommodate all three.
NLX uses a 340-pin connector between the motherboard and riser. This single connector
carries signals for the ISA and PCI buses, IDE drives, and miscellaneous system functions.
In addition, the specification reserves space for a larger connector to accommodate future,
wider expansion buses and an optional connector for other system features.
WTX
Workstations require greater power and design flexibility from ordinary general-purpose
computers, so in September, 1998, Intel introduced a new specification aimed at standardizing many aspects of workstation design. Under the specification, motherboards could be
up to 14 by 16.75 inches or any size smaller. This design proved superfluous, however,
and although the WTX specification is still published on the Web at www.wtx.org, its promoters no longer advocate its use.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
30
Expansion Boards
The essence and key to success of expansion boards is standardization.
By rigidly defining every dimension and contacts, expansion board standards allow you to plug almost any expansion board into almost any
computer. But standardization is not a single issue. Expansion boards
must be physically, electrically, and logically compatible with the motherboards with which they mate.
First of all, a board must fit into the computer for which it is designed.
Consequently, todays expansion board standards dictate the physical size
of expansion boards. In addition, the board must be able to send its signals
to the motherboard and listen to those that the motherboard sends out. It
must have some kind of electrical connection with the motherboard.
Consequently, the physical standardization of expansion boards extends to
their electrical connectors. Before signals can hope to get from one board
to another, the connectors must mate together. They must match as to
size, style, and placement so that they can properly fit together.
Although the physical size of an expansion board has no effect on its electrical compatibility, it is a major issue in compatibility. After all, an expansion board has to fit in the computer it is meant to expand. All expansion
standards define the size of the circuit boards they use either as exact
dimensions or as a set of maximum and, often, minimum dimensions.
Given his druthersor even someone elses druthersa circuit board
designer would rather have more board space, acres upon acres of it
stretching from here to the horizon to the limits of time and the universe. More space to work with makes the job easier. Unfortunately for
998
the designer, practical concerns such as keeping the size of the computer manageable
restrains his ambitions and board space.
On the other hand, smaller boards help manufacturers cut costs, at least after the price for
developing new designs is paid. They require less in the way of the glass-epoxy base materials from which the boards are built. Consequently, setting the dimensions for an expansion board standard is always a compromise.
The prototype of all expansion boards was the card that fit into the original IBM computer of 1981. The bus and physical form of this board lurks behind all computer expansion to this day, evolving into Industry Standard Architecture and legacy expansion
boards, now obsolete thanks to more modern designs.
The reigning standard is the physical size of the PCI bus standard, which specifies the slot
and board size as well as other physical and electrical characteristics of expansion boards.
Even though PCI electrically and logically differs markedly from the original legacy
design, PCI still shows its heritage in several of its physical characteristics. The maximum
board length and thickness both go back to the first days of the computer. The multipart
basic design, too, harkens back to the first computer expansion cards. The same physical
design also embraces the offshoots of PCI technologyPCI-X and PCI-Express.
Other expansion standards youre likely to encounter include the PC Card and CardBus
standards, which from a physical perspective are identical (despite the electrical differences). Your desktop computer may also have a riser board for some of its features (such as
sound and networking), which you may be able to upgrade or replace, much like an
expansion board. Servers may use expansion boards based on the InfiniBand standard.
PCI Boards
The PCI specification defines several variations on the basic expansion board. The standard defines two official sizes of board, each with three connector arrangements (5 volt,
3.3 volt, and dual voltage). The dimensions of the two board sizes may seem arbitrary, but
they owe their heritage to traditional ISA expansion boards. The similarity arose because
early PCI-based computers also hosted ISA expansion boards for compatibility reasons.
The ISA boards defined the size of the computer, and PCI boards were tailored to fit the
space.
Dimensions
A full-size PCI expansion board measures 12.283 inches (312 mm) long. The main body
of the board is about 3.875 inches high, although the expansion edge connector and a
short skirt extend the width of the board to 4.2 inches (106.68 mm). The critical reference
dimension is the centerline of the notch in the expansion connector, which is displaced
CHAPTER 30 EXPANSION BOARDS
PCI Boards
4.113 inches (104.47 mm) from the back edge (retaining bracket side) of the board.
Figure 30.1 shows the dimensions of a full-size PCI board in 5-volt configuration. Cards
designed for 3.3 volts or universal operation will differ in contact number and placement
but not overall size.
FIGURE 30.1
12.283"
4.2"
3.875"
Primary dimensions of
a full-size 5-volt PCI
expansion board.
PCI also defines a short board, about half the length of a full-size board at 6.875 inches
(174.63 mm) front to back. All other vital dimensions, including the distance from rear
edge to the registration notch in the expansion connector, are identical to a full-size
board. Figure 30.2 shows the dimensions of a PCI short card.
FIGURE 30.2
6.875"
3.875"
4.2"
Primary dimensions of
a 5-volt PCI short card.
No matter the length of the board, PCI specifies a maximum thickness. The components
on a PCI board can rise no more than 0.570 inches above the substrate. On the other side
of the board, components can protrude no more than 0.105 inches. Add those heights to
the nominal 0.100-inch thickness of the substrate, and PCI provides for 0.025 inches of
safety gap between board components, with standard slot spacing of 0.800 inches.
Connectors
To accommodate the development of low-voltage green computers, PCI specifies two
connector types and three different connector regimes: a 5-volt connector for todays prevailing circuit designs, a 3.3-volt connector for low-power designs, and the capability to
combine both connectors on a single expansion board for a smooth transition between
designs. A key on 5-volt sockets (blocking pins 50 and 51) prevents the insertion of 3.3volt boards. (Five-volt boards have a slot corresponding to the key.) A key on 3.3-volt
999
1000
sockets (at pins 12 and 13) restricts the insertion to correspondingly slotted 3.3-volt
boards. Boards capable of discriminating the two voltage regimes have slots in both places,
as shown in Figure 30.3.
FIGURE 30.3
Edge connectors for
5 and 3.3 volt PCI
cards (32 bit).
3.3 Volt
5 Volt
The 64-bit implementation of PCI extends the edge connector to accommodate the additional required signals. Figure 30.4 shows this extended connector and the full implementation of all of its options.
FIGURE 30.4
The 64-bit universal
PCI edge connector.
64-BIT Universal PCI Connector
3.3 volt
Base 32-bit connector
5 volt
64-bit expansion

Boards of the accelerated graphics port (AGP) design follow a physical standard derived
from that of PCI, much as the AGP electrical interface is an offshoot of PCI. The AGP
design assumes that the 3D accelerator chip and frame buffer will have a home on an
expansion board rather than the motherboard, although nothing about the interface
design precludes putting all the circuitry on the motherboard. In a typical computer, an
expansion slot supporting the AGP connection fits on the motherboard like a PCI slot,
with the same spacing and dimensional requirements. Although a given computer or
motherboard may omit a PCI slot in favor of an AGP connection, AGP is electrically separate from the PCI slots and does not reduce the number of slots that may be connected
to a single bridge.
Dimensions
The AGP circuit board design is built around a conventional PCI card, at least for use in
full-size computersthose matching (or approximating) the ATX design. For these systems, the original AGP standard provides for the two sizes of card. A full-size AGP board
measures 4.2 inches (107 millimeters) tall and 12.28 inches (312 millimeters) long. Short
cards measure only 6.875 inches (175 millimeters) long. Figure 30.5 illustrates these principal dimensions.
FIGURE 30.5
Long card = 12.283 in.
Dimensions of an ATXsize AGP video board.
4.2 in.
Short card = 6.875 in.
For low-profile computers that match the NLX standards, the revised AGP 2.0 standard
provides a special, low-slung card form factor that allows the AGP-based graphics adapter
to plug into the motherboard rather than the riser that accommodates other expansion
boards. Figure 30.6 shows the dimensions of the NLX-style AGP board.
FIGURE 30.6
3.25 in.
Dimensions of an
NLX-style AGP video
board.
13.415 in.
The slot spacing for AGP expansion boards in ATX computers is the same as for PCI and
ISA boards, 0.8 inches. AGP boards in such systems use the same retaining bracket design
as ordinary PCI boards. The NLX design, however, requires a specialb bracket, as shown
in Figure 30.7.
FIGURE 30.7
The NLX-styleb AGP
retaining bracket.
1001
1002
Connector
Although the AGP slot bin a computer uses an edge connector, it is an entirely different
connector from those used in PCI slots. This physical difference ensures against inserting
an expansion board in an inappropriate slot. Other aspects of AGP further constrained the
design of the connector. The chosen design interleaves contacts at two depths inside the
connector with the mating fingers on the edge connector arranged in a flip-flop pattern,
as shown in Figure 30.8.
FIGURE 30.8
Layout of the highdensity contacts on an
AGP edge connector.
The interleaved contact fingers on the AGP card make hot-plugging AGP boards impossible. That is, you should not slide an AGP board into its slot when the power is on to
your computer. When you slide a board into a slot, the contact fingers closest to the edge
of the board are apt to wipe across the connector contacts for the deeper contacts. Even a
quick wipe can send weird, even damaging signals through the card and your computer.
Intel recommends that you slide AGP boards into their slots as straight as you can without rocking them back and forth to seat them in the connector. Even when the power is
off, the side-to-side wiping action can damage the miniaturized contacts in the AGP
connector.
Riser Boards
A riser board looks like an expansion board and works like an expansion board. Using the
quack-standardif it looks like a duck and quacks like a ducka riser board is an expansion board. But really its not, at least according to the people who have created the standards that make it possible to consider riser boards as being expansion boards. In its
various riser board standards, Intel claims that its riser boards are not an expansion board
standard.
Rest assured youre not the only one confused by riser boards. The concept underlying
them is the same as an expansion boardput extra circuitry on a special card so that it can
be easily removed and changed. The boards look like expansion boards, pegged at the
same height and using similar expansion connectors. Risers slide into a motherboard slot
placed adjacent to ordinary expansion boards so that, at least from the outside, they look
like simply another expansion board.
Riser Boards
Intel doesnt want you to think of riser boards as expansion boards because it doesnt want
to confuse you with more than one expansion board standard. It doesnt want third-party
manufacturers putting riser boards in boxes and selling them in retail stores. Likely the
company fears someone will buy a riser and discover they cant slide it into a PCI slot, and
the lonely worker handling support will get a barrage of telephone calls asking for help.
Functionally, a riser board is simply a bend in a motherboard. It rises perpendicularly
from the motherboard (much like an expansion board) to hold circuitry that otherwise
would not fit on the motherboard (much like an expansion board). Riser boards can even
use a slot position, much like an expansion board, but they follow their own standards,
which have absolutely nothing to do with PCI. This independence, in fact, is one of the
graces of the riser board. It allows you to change or customize your system using a lowcost product that avoids the complications (and cost) of a full PCI interface. Moreover,
the Audio Modem Riser (AMR) slot doesnt count against the tiny maximum number of
PCI boards that can be connected to your computers PCI bridge circuitry.
Put another way, riser boards offer a system designer three big benefits: space, timeliness,
and isolation. The physical size of the riser board amounts to additional real estate on
which to locate electronic circuitry, as well as bracket space at the back of the computer
for extra connectors. Using a riser board, a motherboard can be more compact. In fact,
Intel created the original standard riser, the Audio Modem Riser, as an adjunct to its
small-footprint microATX motherboard.
Today, communication and networking standards change faster than most computer standards. Keeping up with these changes may require motherboard-makers to revise their
designs more often than they would otherwise like (they would otherwise like never to change
their designs, because each change racks up design expenses). By restricting the changes to a
small riser board, the manufacturer trims the cost of keeping its motherboards timely.
Theres another reason for putting audio circuits on a separate board. Audio circuits are
the most prone to picking up interference from the rest of your computer. Analog audio
uses signals that vary in a wide range, over more than three orders of magnitude.
Interference typically produces tiny signals that are ignored by logic circuits but are
amplified into bothersome noise by analog circuits. A signal one-thousandth the strength
of a normal logic signal would have zero effect on the operation of your computer but
would be audible in high-quality speakers attached to your computer. By putting the
interference-prone circuits on a separate board, designers can physically isolate them from
the other signals on the motherboard, which adds a degree of electrical isolation as well.
With these ideas in mind, Intel and other companies have created three standards for riser
boards meant to work in modern computers. These include the original Audio Modem
Riser standard, the Communications and Networking Riser standard, and the Advanced
Communications Riser standard.
1003
1004
Audio Modem Riser

The first of the riser-board standards independent of a specific motherboard design was
Intels original Audio Modem Riser (AMR). The AMR specification was originally published on June 30, 1998, and updated to its final form (version 1.01) on September 10,
1998. It envisions a board capable of holding up to four audio codecs on a single board
thats at most 6.875 inches long and 4.2 inches high. It uses a single 42-contact edge connector located between the AGP port and other port connectors on the motherboard. It
takes the space of an ordinary expansion board but not the connector or the circuitry, so it
doesnt count against the maximum number of PCI boards that can be inside a single
computer.
The interface was designed to allow a great deal of flexibility. Not only does it allow for
four codecs where two would typically be usedone to implement a modem, one for the
equivalent of a soundboarda special set of split I/O signals allows, for example, a sound
system implementation on the motherboard to communicate with a modem on the riser
board. These signals take the form of four separate serial data channels from the chipset
(typically) to the audio circuitry on the riser board. The interface also provides the power
and clock signals needed to make the audio circuits work, as well as signals for future
applications (including the USB bus).
Although youll still find some references to the AMR design, Intel has replaced it with
the Communications Network Riser standard (discussed next) and has removed the AMR
specification from its Web site.
Communications Network Riser

On February 7, 2000, Intel released its replacement for the AMR design, which it called
the Communications and Networking Riser (CNR). The networking part of the name
marks the major difference. CNR incorporates provisions for networking and USB
interfaces as well as the audio and modem technologies of the earlier design. The full
specification for the CNR board and its interfaces is available at the Intel Web site at
www.intel.com.
The CNR design uses a new 60-pin connector with a preferred location at the edge of the
motherboard: outboard of its normal PCI expansion slots, a position formerly used by a
legacy ISA slot on many motherboards. Significantly, the new connector is incompatible
with AMR boards.
Despite the physical incompatibility, the CNR standard starts with some of the basic functional compatibilities as with boards made for the AMR specification. At heart, CNR
incorporates the Audio Codec (AC) 97 interface. Unlike AMR, which permits four AC
97 channels, the CNR specification originally allowed only two, although it was revised
(to version 1.2) on November 8, 2001, to permit three.
Riser Boards
The networking side of the CNR design defines two different board implementations,
each using the same connector but with different pinouts. The A variation uses eight
contacts for a platform LAN connection interface, meaning that the hard work of generating the network signals is on the motherboard. The riser only packages them, for example, putting them on the same RJ-11 jack as the modem signals. The B variation uses 17
contacts for a Media Independent Interface (MII, as defined under the IEEE 802.3u specification) bus, allowing the riser board circuitry to set up any kind of physical connection
(which makes the riser board circuitry more complex). In other words, the A variation
allows for less complex riser boards, whereas the B variation allows for greater versatility
(for example, permitting the manufacturer to install different network interfaces by changing the riser board).
The CNR design also allows for the integration of a USB interface on the riser board.
With the CNR version 1.1 revision of October 18, 2000, the specification added support
for USB version 2.0 at 480Mbps.
So that the CNR board will be properly identified by the host computer, each board has
five signal lines for a System Management Bus (SMBus) interface. The host computer can
use the bus to interrogate ROM memory on the board to identify its function and its
resource needs each time the host boots up. The CNR design also includes several lines to
provide power to the circuitry on the riser board at 12, 5, and 3.3 volts (DC) as well as
sleep-mode voltage so the circuitry of the CNR board may remain active while the host is
in sleep mode. (This voltage allows the CNR circuitry to wake the host computer upon
modem ring or network request.)
The physical specifications for the CNR board are similar to those of the AMR board.
Both boards are limited to the same maximum height, 4.2 inches (106.68 millimeters),
dictated by host height restrictions within the PCI standard. CNR boards have a slightly
shorter maximum length than AMR boards, at 6.579 inches (167.11 millimeters). The
connector is, of course, different, unique to the CNR board, although the pin spacing on
its edge connector matches that of the standard PCI connector.
Advanced Communications Riser

This need for yet another new connector and other shortcomings of the CNR design led
an industry group to design its own alternative to CNR on February 11, 2000. That organization, which called itself the Advanced Communications Riser Special Interest Group (ACR
SIG), has grown to 55 members at the time of this writing, including, significantly,
Advanced Micro Devices, but not Intel.
Working together, they created the Advanced Communications Riser (ACR) board to add
network and Digital Subscriber Line (DSL) capabilities to simple add-on boards without
losing compatibility with AMR boards or adding the need for a new connector.
1005
1006
Physically, the ACR slot uses a 120-pin PCI connector (for its ready availability), but the
connector is reversed and offset in its position in its slot. This change in orientation and
position makes it compatible with AMR boards. The first 42 contact positions on the
ACR board have the same functions and signals as the AMR board. The ACR design
envisions its slot position on the outside edge of the motherboard, replacing the legacy
ISA slot in many motherboard designs.
An AMR board can slide directly into an ACR slot. An ACR board will, too, although it
wont slide into an AMR slot because of the longer connector on the ACR board.
The electrical functions of the ACR board start with the four Audio Codec 97 channels
of the AMR board and add two more AC 97 channels, for a total of six.
In addition, the ACR design provides two new pins as a serial data channel for board identification. Each ACR board includes an onboard ROM chip to identify the board and its
capabilities to the host computer as part of the Plug-and-Play setup process.
One of the new sets of 14 signals is the Integrated Packet Bus, developed in conjunction
with the ACR design to directly link the host microprocessor with the communications
subsystem and control high-speed Internet connections. Two sets of 18 signals each provide a pair of channels that separately link your chipset to phone-line networking and
ordinary twisted-pair networks following the 10Base-T or 100Base-T standards. The
ACR specification reserves an additional eight contacts for future applications, such as
wireless networking adapters.
The ACR design specifically recognizes home phone-line networking (HomePNA). It envisions modem-makers offering a single multifunction board that includes a V.92 modem, a
HomePNA network adapter, and a DSL adapter. A single ACR card will link to your home
phone-line network, your telephone line as a regular modem (for data as well as dialing and
Internet phone service), and your DSL line, without the need for external adapters. All
phone-line communications functions can then use a single RJ-11 jack on the ACR card.
Although the ACR SIG claims that its standard is open, the specification is available only
to members of the group. Its Web site is located at www.acrsig.org.
PCMCIA Cards
Notebook computers use PCMCIA cards of two types for expansion: 18-bit PC Cards
based on the ISA bus and 32-bit CardBus cards based on the PCI bus. No matter which
standard a given card follows, however, it must follow exactly the same physical specifications as all PCMCIA cards. Every card will slide into every slotwell, almost. The standard allows for three thickness variations, but one of which is almost never found, and
nearly all slots will accommodate the other two sizes. In other words, physically, PC Cards
and CardBus cards are identical, and both use exactly the same connector.
PCMCIA Cards
Under the PCMCIA standard, PC Cards and CardBus cards come in three sizes, differing
only in thickness. The basic unit of measurementthe size of the typical card slotis
based on the medium-thickness card, designated Type II. Measuring 54 by 85 millimeters
(2.126 by 3.37 inches) and 5 mm (about three-sixteenths of a inch) thick.
PC Cards and CardBus cards physically follow the form factor of earlier memory cards
(including the IC Card) standardized by JEIDA (the Japan Electronic Industry
Development Association). The first release of the PCMCIA specification paired this single-size card with a Fujitsu-style 68-pin connector. Under the current PCMCIA specification, version 8.0 (and all versions since the 2.1 specification), this form factor is designated
as the Type I PC Card.
The thinness of the Type I card proved an unacceptable limitation. Even without allowing
for the PC Card packaging, some solid-state devices are themselves thicker than 3.3 mm.
Most important among these fat devices are the EPROMs used for nonvolatile storage.
(Most computers use EPROMs to store their system BIOS, for example.) Unlike ordinary,
thin ROMs, EPROMs can be reprogrammed, but this requires a transparent window to
admit the ultraviolet radiation used to erase the programming of the chip. The windowed
packaging makes most EPROMs themselves 3.3 mm or thicker.
Fujitsu faced this problem when developing the firmware to be encoded on memory cards
and therefore developed a somewhat thicker card that could be plugged into the same
sockets as could standard memory cards. Modem and other peripheral makers found the
Fujitsu fat card more suited to their purposes. To accommodate them, PCMCIA 2.0 standardized an alternative: the Type II PC Card. Essentially based on the old Fujitsu developmental EPROM form factor, Type II PC Cards are 5 mm thick but otherwise conform
to the same dimensions as Type I cards.
The PCMCIA standard puts the extra thickness in a planar bulge, called the substrate area,
in the middle of the card. This thicker area measures 48 mm wide and 75 mm long. Three
millimeters along each side of the Type II card are kept to the thinness of the Type I standard so that the same card guides can be used for either card type. Similarly, the front 10
mm of a Type II card maintain the 3.3 mm thickness of the Type I standard so that the
same connector can be used for either card type. Naturally, the actual card slot for a Type
II card must be wide enough to accommodate the maximum thickness of the card.
In September 1992, PCMCIA approved a third (Type III) form factor for PC Cards.
These still-thicker cards expand the bulge of Type II from 5 mm to 10.5 mm and are
designed to accommodate miniaturized hard disks and similar mechanical components. As
with Type II cards, Type III PC Cards remain thin at the edges to fit standard card guides
and standard connectors. Although a number of hard disk drives appeared in this format,
few are used today in notebook computers. However, high-capacity MP3 players have
embraced the Type III card design.
1007
1008
In practical terms, a Type I card comes closest to being a truly flat, credit-card style card.
Type II cards have small bulges at the top and bottom to accommodate circuitry. Type III
cards have thick lumps to hold a disk drive. Figure 30.9 illustrates the apparent differences
between the three card types.
FIGURE 30.9
The three PCMCIA
card types vary in
thickness from 3.3 mm
to 10.5 mm.
Type l
Type ll
Type lll
Note that many notebook computers lay out their endowment of two PCMCIA slots with
one over the other and without a metal separator. This design allows the two slots to hold
two individual Type II or a single Type III card. Nearly all current PCMCIA slots are
wide enough to accommodate Type II cards. The only place youre likely to encounter a
slot that will accept only Type I cards is equipment manufactured before 1992.
Under the current PCMCIA standard, both Type I and Type II cards can be implemented
in extended form. That is, their depth can be increased by an additional 50 mm (to 135
mm) to hold additional componentryfor example, the antennae of WiFi network
adapters. Such extended cards project about two inches more from standard PCMCIA slots.
To ensure that all cards easily and securely mate with their connectors, the PCMCIA
standard requires that card guides be at least 40 mm long and that the PCMCIA card
connector must engage and guide the connector pins for 10 mm before the connector
bottoms out.
The layout of a PC Card or CardBus card is essentially symmetrical, meaning that it
could inadvertently be inserted upside down. The PCMCIA design allows for such cases
of brain fade by eliminating the risk of damage. Although the cards do not work while
inverted, neither they nor the computers into which they are plugged will suffer damage.
Because the size and placement of labels on the cards is part of the standard, when you are
familiar with the layout of one PC Card, you will know the proper orientation of them all,
and CardBus cards as well. Moreover, other physical aspects of the cardsthe position of
the write-protect switch (if any) and battery (if needed)are standardized as well. The
PCMCIA standard also recommends that the batteries in all cards be oriented in the same
direction (positive terminal up).
In addition to the physical measures that facilitate getting the cards into their sockets,
two pinsone on each side of the connectorallow the computer host to determine
whether the card is properly seated. If the signal (ground) from one is present and the
other is not, the system knows that the card is skewed or otherwise improperly inserted
in the connector.
PCMCIA Cards
Connector
All types of PCMCIA cards use the same 68-pin connector, whose contacts are arranged
in two parallel rows of 34 pins. The lines are spaced at 1.27 mm (0.050 inch) intervals
between rows and between adjacent pins in the same row. Male pins on the card engage a
single molded socket on the host.
To ensure proper powering up of the card, the pins are arranged so that the power and
ground connections are longer (3.6 mm) than the signal leads (3.2 mm). Because of their
greater length, therefore, power leads engage first so that potentially damaging signals are
not applied to unpowered circuits. The two pins (36 and 67) that signal that the card has
been inserted all the way are shorter (2.6 mm) than the signal leads.
1009
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
31
Power
All practical computers made today operate electronically. Moving electronselectricityare the media of their thoughts. Electrical pulses
course from one circuit to another, switched off or on in an instant by
logic chips. Circuits combine the electrical pulses together to make logical decisions and send out other pulses to control peripherals. The computers signals stay electrical until electrons colliding with phosphors in
the monitor tube push out photons toward your eyes or generate the
fields that snap your printer into action.
Of course, your computer needs a source for the electricity that runs it.
The power does not arise spontaneously in its circuits but rather must be
derived from an outside source. Conveniently, nearly every home in
America is equipped with its own electrical supply that the computer can
tap into. Such is the wonder of civilization.
But the delicate solid-state semiconductor circuits of todays computers
cannot directly use the electricity supplied by your favorite utility company. Commercial power is an electrical brute, designed to have the
strength and stamina to withstand the miles of travel between generator
and your home. Your computers circuits want a steady, carefully controlled trickle of power. Raw utility power would fry and melt computer
circuits in a quick flash of miniature lightning.
For economic reasons, commercial electrical power is transmitted
between you and the utility company as alternating current, the familiar
AC found everywhere. AC is preferred by power companies because it is
easy to generate and adapts readily between voltages (including to very
high voltages that make long distance transmission efficient). Its called
1012
alternating because it reverses polarityswapping positive for negativedozens of times a

second (arbitrarily 60Hz in America; 50Hz in Europe).
The changing or oscillating nature of AC enables transformers to increase or decrease
voltage (the measure of driving force of electricity), because transformers only react to
electrical changes. Electrical power travels better at higher voltages because waste (as heat
generated by the electrical current flowing through the resistance of the long-distance
transmission wires) is inversely proportional to voltage. Transformers permit the high
voltages used in transmitting commercial powersometimes hundreds of thousands of
voltsto be reduced to a safe level (nominally 117 volts) before it is led into your home.
As wonderful as AC is to power companies, its an anathema to computer circuits. These
circuits form their pulses by switching the flow of electricity tapped from a constant supply. Although computers can be designed that use AC, the constant voltage reversal would
complicate the design so that juggling knives while blindfolded and riding a roller coaster
would seem tame in comparison. Computers (and most electronic gear) use direct current
(DC) instead. Direct current is the kind of power that comes directly from a primary
sourcea batterya single voltage that stays at a constant level (at least constant as long
as the battery has the reserves to produce it). Moreover, even the relatively low voltage
that powers your lights and vacuum cleaner would be fatal to semiconductor circuits. Tiny
distances separate the elements inside solid-state circuits, and high voltages can flash
across those distances like lightning, burning and destroying the silicon along the way.
Power Supplies
The intermediary that translates AC from your electrical outlets into the DC that your computers circuits need is called the power supply. As it operates, the power supply of your
computer attempts to make the direct current supplied to your computer as pure as possible,
as close to the ideal DC power produced by batteries. The chief goal is regulation, maintaining the voltage as close as possible to the ideal desired by the circuits inside your computer.
The power needs of the circuitry of desktop and notebook computers are the same. Both
are built with the same circuitry with the same voltage and similar current requirements.
Ultimately, both kinds of computers draw power from the same sourceelectric utilities. But
they differ in how they handle the power between your outlet and their circuits. Desktop systems provide a direct route. Utility power gets converted to circuit power once and for all
inside a little tin box called a power supply. Notebook computers add an auxiliary power
sourcetheir batteriesto the mix. In so doing, the power circuitry of the computer system
is rearranged to suit the needs of both the logic circuitry of your computer and its batteries.
In effect, the notebook computer has two power sources: utility power when youre near
an outlet (and plugged in) and battery power when youre in the field (or in an airplane).
But they are not entirely independent. The same circuits regulate the voltage going to the
CHAPTER 31 POWER
Power Supplies
logic of your computer whether your notebook machine is running from AC or battery
power. In fact, should you remove its batteries and the charging and control circuits, the
power functions of a notebook are the same as a desktop machine. Only the location of
the power circuits has been rearranged to suit the system designers. The same technologies that bring a desktop power supply to life also are at work in notebook machines.
Desktop Power Supplies

Most computers package their power supplies as a subassembly thats complete in itself
and simply screws into the chassis and plugs into the system board and other devices that
require its electricity. The power supply itself is ensconced in a metal box perforated with
holes that let heat leak out but prevent your fingers from poking in.
In fact, the safety provided by the self-contained and fully armored computer power supply
is one of the prime advantages of the original design. All the life-threatening voltagesin
particular, line voltageare contained inside the box of the power supply. Only low, nonthreatening voltages are accessiblethat is, touchableon your computers system board
and expansion boards. You can grab a board inside your computer even when the system is
turned on and not worry about electrocution (although you might burn yourself on a particularly intemperate semiconductor or jab an ill-cut circuit lead through a finger).
Grabbing a board out of a slot of an operating computer is not safe for the computers circuits, however. Pulling a board out is apt to bridge together some pins on its slot connector, if but for an instant. As a result, the board (and your computers motherboard) may
find unexpected voltages attacking, possibly destroying, its circuits. In other words, never
plug in or remove an expansion board from a computer that has its power switched on.
Although you may often be successful, the penalty for even one failure should be enough
to deter your impatience.
In many desktop computers, the power supply serves a secondary function. The fan that
cools the power supply circuits also provides the airflow that cools the rest of the system.
This fan also supplies most of the noise that computers generate while they are running.
In general, the power supply fan operates as an exhaust fanit blows outward. Air is
sucked through the other openings in the power supply from the space inside your system.
This gives dust in the air taken into your computer a chance to settle anywhere on your
system board before getting blown out through the power supply.
The prevailing standard for power supplies of todays computers is ATX. This standard
defines not only the power supply itself but also the connectors and voltages available for
the motherboard and devices inside the computer.
Technologies
In electronic gear, two kinds of power supplies are commonly used: linear and switching.
The former is old technology, dating from the days when the first radios were freed from
1013
1014
their need for storage batteries in the 1920s. The latter rates as high technology, requiring
the speed and efficiency of solid-state electronic circuitry to achieve the dominant position
it holds today in the computer power market. These two power supply technologies are
distinguished by the means used to achieve their voltage regulation.
Linear Power Supplies
The design first used for making regulated DC from utility-supplied AC was the linear
power supply. At one time, this was the only kind of power supply used for any electronic
equipment. When another technology became available, it was given the linear label
because it then used standard linear (analog) semiconductor circuits, although a linear
power supply need not have any semiconductors in it at all.
In a linear power supply, the raw electricity from the power line is first sent through a transformer that reduces its voltage to a value slightly higher than required by the computers circuits. Next, one or several rectifiers, usually semiconductor diodes, convert the now
low-voltage AC to DC by permitting the flow of electricity in only one direction, blocking
the reversals. Finally, this DC is sent through the linear voltage regulator, which adjusts the
voltage created by the power supply to the level required by your computers circuits.
Most linear voltage regulators work simply by absorbing the excess voltage made by the transformer, turning it into heat. A shunt regulator simply shorts out excess power to drive the
voltage down. A series regulator puts an impedimenta resistancein the flow of electricity,
blocking excess voltage. In either case, the regulator requires an input voltage higher than the
voltage it supplies to your computers circuits. This excess power is converted to heat (that is,
its wasted). The linear power supply achieves its regulation simply by varying the waste.
Switching Power Supplies
The design alternative is the switching power supply. Although more complex, switching
power supplies are more efficient and often less expensive than their linear kin. Although
designs vary, the typical switching power supply first converts the incoming 60 hertz utility power to a much higher frequency of pulses (in the range of 20,000Hz, above the
range of normal human hearing) by switching it on and off using a transistors.
At the same time, the switching regulator increases the frequency of the commercial
power; it regulates the commercial power using a digital technique called pulse width modulation (PWM). That is, the duration of each power pulse is varied in response to the
needs of the computer circuitry being supplied. The width of the pulses is controlled by
the electronic switch; shorter pulses result in a lower output voltage. Finally, the switched
pulses are reduced in voltage down to the level required by the computer circuits by a
transformer and then turned into pure direct current via rectification and filtering.
Switching power supplies earn their efficiency and lower cost in two ways. Switching
regulation is more efficient because less power is turned into heat. Instead of dissipating
energy with a shunt or series regulator, the switching regulator switches all current flow
CHAPTER 31 POWER
Power Supplies
off, albeit briefly. In addition, high frequencies require smaller, less expensive transformers
and filtering circuits. For these two very practical reasons, nearly all of todays personal
computers use switching power supplies.
Power Needs
Modern computer logic circuits operate by switching voltages with the two different logic
states (true or false, one or zero, for example) coded as two voltage levelshigh and low.
Every family of logic circuits has its own voltage standards.
The primary consumers of power inside a computer are its logic circuits. At one time,
nearly all logic circuits used five volts of direct current. This power level was set by the
design of the electronic circuit components they used, based around the requirements of
transistor-transistor logic (TTL). In a TTL design, high refers to voltages above about 3.2
volts, and low means voltages lower than about 1.8. The middle ground is undefined logically, an electrical guard band that prevents ambiguity between the two meaningful states.
To reduce the power needs of todays high-speed circuits, computer designs are shifting to
3.3-volt logic and require power supplies that deliver that voltage level (often in addition
to 5-volt power). The ATX design and those derived from it (SFX and TFX) provide for
both 5-volt and 3.3-volt supplies.
Some computer circuits, such as microprocessors, run at even lower voltages. The levels
required by their circuits are not available from most computer power supplies. Instead,
motherboard-makers use voltage regulators to reduce a 12-, 5-, or 3.3-volt power source to
the level required by the chips. This design allows standard power supplies to work with
chips rated for almost any voltage.
In addition to the basic logic voltage, computers often require other voltages as well. The
motors of most disk drives (hard and floppy) typically require 12 volts to make them spin.
Other specialized circuits in computers sometimes require bipolar electrical supplies. A
serial port, for example, signals logic states by varying voltages between positive and negative in relation to ground. Consequently, the mirror image voltages (5 and 12 volts) are
usually available inside every computer.
In notebook computers, most of which have no room for generic expansion boards, all
these voltages are often unnecessary. For example, many new hard disks designed for
notebook computers use 5-volt motors, eliminating the need for the 12-volt supply. The
custom-tailored power systems of notebook computers supply only the voltages required
by circuits actually built in to the computer.
Voltages and Ratings
The power supplies you are most likely to tangle with are those inside desktop computers,
and these must produce all four common voltages to satisfy the needs of all potential combinations of circuits. In older desktop computers, the power supply typically produces four
1015
1016
voltages (+5, 5, +12, and 12) that are delivered in different quantities (amperages) because of
the demands associated with each. A separate voltage regulator on the motherboard produces
the lower voltage in the 3.3-volt range required by Pentium-level microprocessors and their
associated circuitry. In some systems, the output voltage of this regulator may be variable to
accommodate energy-saving systems (which reduce the speed and voltage of the microprocessor to conserve power and reduce heat dissipation). Newer power supplies, such as those that
follow the ATX design standard, sometimes also provide a direct 3.3-volt supply.
The typical computer has a lot of logic circuitry, so it needs copious quantities of 5-volt
power, often as much as 20 to 25 amperes). Many disk drives use 12-volt power; the typical
modern drive uses an ampere or so. Only a few components require the negative voltages,
so most power supplies only deliver a few watts of each.
Most power supplies are rated and advertised by the sum of all the power they can make
available, as measured in watts. The power rating of any power supply can be calculated by
individually multiplying the current rating of each of the four voltages it supplies and summing the results. (Power in watts is equal to the product of voltage times current in
amperes.) Most modern full-size computers have power supplies of 150220 watts.
Notebook computers may use from 10 to 50 watts when processing at full speed.
Note that this power rating does not correspond to the wattage that the power supply draws
from a wall outlet. All electronic circuitsand power supplies in particularsuffer from inefficiencies, linear designs more so than switching. Consequently, a power supply requires a wattage in excess of what it provides to your computers circuitsat least when it is producing its
full output. Computer power supplies, however, rarely operate at their rated output. As a result,
efficient switching power supplies typically draw less power than their nominal rating in normal use. For example, a computer with a 220-watt power supply with a typical dosage of memory (say, 4MB) and one hard disk drive likely draws less than 100 watts while it is operating.
When youre selecting a power supply for your computer, the rating you require depends
on the boards and peripherals with which you want to fill your computer. Modern computers are not nearly so power hungry as their forebears. Nearly all computer components
require less power than the equivalents of only half a dozen years ago. The one exception is
the microprocessor. Greater performance requires more power. Although Intels engineers
have done a good job at reducing the power needs of the companys products by shifting to
lower-voltage technologies, the reductions have been matched by increasing demands.
Chips need as much power as they ever have, sometimes more.
In a desktop computer, a 200-watt power supply essentially loafs along; most individual
computers (as opposed to servers or workstations) could get along with 120 watts without
straining. A system board may require 1525 watts; a floppy disk drive, 320 (depending on
its vintage); a hard disk, 550 (also depending on its vintage); a memory or multifunction
expansion board, 510. Table 31.1 summarizes the needs of both vintage and modern computer components.
CHAPTER 31 POWER
Power Supplies
TA B L E 3 1 . 1
Typical Device Power Demands
Device Class
Device Type
Power
Example
Floppy disk drive
Full height, 5.25 inch
12.6 watts
IBM PC diskette drive
Half height, 5.25 inch
12.6 watts
QumeTrak 142
One-inch high, 3.5 inch
1.4 watts
Teac FD-235J
Two-board old technology
16.2 watts
IBM 8514/A
High performance, full length
13.75 watts
Matrox MGA
Accelerated half-card
6.5 watts
ATI VGA Wonder,

Graphics Ultra+
Full height, 5.25-inch
59 watts
IBM 10MB XT hard disk
Half height, 5.25-inch
25 watts
[Estimated]
One-inch high, 3.5 inch
12.25 watts
(peak)
maxtor DiamondMax
Plus 9
2.5 inch
2.2 watts
Quantum Go-Drive
120AT
PCMCIA card
3.5 watts
Maxtor MXL-131-III
Full height, 3.5 inch
12 watts
Quantum ProDrive 210S
1MB SIMM
4.8 watts
Motorola MCM81000
4MB SIMM
6.3 watts
Motorola MCM94000
8MB SIMM
16.8 watts
Motorola MCM36800
128MB SoDIMM
1.8 watts
Micron MT8LSD3264
PCMCIA card
3.5 watts
MultiTech MT1432LT
Internal, half-card
1.2 watts
Boca V.32bis
Network adapter
Ethernet, half-card
7..9 watts
Artisoft AE-2/T
System board
286, AT size
25 watts
[Estimated]
386, XT size
12 watts
Monolithic Systems
MSC386 XT/AT
486 or Pentium, AT
size
25 watts
[Estimated]
Graphics board
Hard disk
Memory
Modem
Reserve power is always a good thing, and with switching power supplies, it comes without a penalty (except in the cost of making the power supply). In other words, although
you could get along with a smaller power supply in your desktop computer, the quasistandard 200 watts remains a good choice.
1017
1018
Supply Voltage
Most power supplies are designed to operate from a certain line voltage and frequency. In
the United States, utility power is supplied at a nominal 115 volts and 60 hertz. In other
nations, the supply voltage and frequency may be different. In Europe, for instance, a 230volt, 50Hz standard prevails.
Most switching power supplies can operate at either frequency, so that shouldnt be a
worry when traveling. (Before you travel, however, check the ratings on your power supply to be sure.) Linear power supplies are more sensitive. Because their transformers have
less reactance at lower frequencies, 60Hz transformers draw more current than their
designers intend when operating on 50Hz power. Consequently, they are liable to overheat and fail, perhaps catastrophically.
Most computer power supplies are either universal or voltage selectable. A universal power
supply is designed to have a wide tolerance for supply current. If you have a computer with
such a power supply, all you need to do is plug the computer in, and it should work properly. Note that some of these universal power supplies accommodate any supply voltage in
a wide range and will accept any standard voltage and a line frequency available in the
worlda voltage range from about 100 to 250 volts and a line frequency of 50 to 60Hz.
Other so-called universal supplies are actually limited to two narrow ranges, bracketing
the two major voltage standards. Because you are unlikely to encounter a province with a
169.35-volt standard, these dual-range supplies are universal enough for worldwide use.
Voltage-selectable power supplies have a small switch on the rear panel that selects their
operating voltage, usually in two ranges115 and 230 volts. If your computer has a
voltage-selectable power supply, make sure that the switch is in the proper position for the
available power before you turn on your computer.
CAUTION
When traveling in a foreign land, always use this power supply switch to
adjust for different voltages. Do not use inexpensive voltage converters.
Often these devices are nothing more than rectifiers that clip half the
incoming waveform. Although that strategy may work for light bulbs, it
can be disastrous to electronic circuitry. Using such a device can destroy
your computer. Its not a recommended procedure.
Standards
The metal modules that serve as power supplies for desktop computers were one of the
first components to be standardized. The technology for building a power supply is common and straightforward, and reverse-engineering a power supply for given computer is
almost trivial. There are no strange codes or complex logic signals coming from a power
supply, just a few, easily measured voltages. Each popular case used a characteristic power
supply that became a de facto industry standard, both in size (the power supply had to
CHAPTER 31 POWER
Power Supplies
match the case mechanically) and connectors (to plug into motherboards and drives). The
earliest of these matched models of IBM computersPC, XT, and ATthe last being the
reigning standard for a decade.
The standards used by modern computers arose when Intel standardized its motherboard
offerings and inspired the industry to follow suit with its ATX design. The new motherboard broke with the past with a different physical layout that required a new case design,
which in turn dictated a different size for the power supply. In addition, Intel chose to
break with the past standards for power connectors to create a new one of its own for the
ATX motherboard. The choice was more than Intel wishing to re-create the industry to
suit itself. A new connector was overduethe old design could not accommodate the 3.3volt supplies that new logic designs required.
The result was the ATX power supply. Roughly patterned after previous designs, the ATX
power supply is a steel-cased modular design that incorporates the computers primary
(and usually, only) external power connection. It supplies all the voltages required by both
past and current logic circuitry: positive direct-current voltages of 5 and 12 volts in addition to the new 3.3-volt supply as well as negative supplies of 5 and 12 volts. But on its
new 20-pin motherboard connector, the ATX design has added several signals aimed at
the needs of modern computers:
Power Good. This signal is a carryover from older designs. The power supply
sends out the Power Good signal to indicate that its output voltages are at the correct level and safe for computer circuits. A computer wont turn on unless the Power
Good signal is present. The technical designation of this signal under the ATX standard is PWR_OK.
On. This is a signal from the motherboard telling the power supply to switch its
main outputs on. This signal allows an external switch to control the computer
power. More importantly, it allows the motherboard to control the power supply.
The motherboard can shift the system to standby and cut its power consumption by
switching off the On signal, and it can shift back to normal operation by restoring it.
This mode of operation is essential for features such as wake-on-alarm, wake-onmodem, and wake-on-network that shift the system to standby awaiting a specific
event (an alarm, incoming call, or network request). The engineering name for this
signal is PS_ON#.
Standby Supply Voltage. This is an auxiliary source of logic voltage that is not controlled by the On signal. It provides the power the motherboard needs when it is on
standby and the main voltages from the power supply have been switched off. The
engineering name for this signal is +5 VSB.
The ATX standard also defines an auxiliary power connector with six pins. This connector is
aimed at high-current power supplies. It gives another channel to the main 5 and 3.3 volts
1019
1020
when the current from the supply exceeds the safe carrying capacity of the 16-gauge wires
recommended for the connections. This connector is used when the 3.3-volt output
exceeds 18 amperes or the 5-volt output exceeds 24 amperes.
ATX power supplies also include several peripheral connectors for disk drives (and, sometimes, auxiliary fans and other accessories). Two styles are usually present: large four-pin
Molex connectors to match hard disk drives and old floppy disks, and miniaturized fourpin connectors for modern floppy disk drives.
With the introduction of ATX version 2.0, Intel added a variation on the ATX power supply called ATX12V. The new design differs from ordinary ATX power supplies only in the
presence of one or more new +12V power connectors. These four-pin connectors provide a
high-current 12-volt supply designed to feed the voltage regulators of high-current
processors or other demanding peripherals.
The ATX specification allows for two types of power suppliesthose meant to cool not
only themselves but also the microprocessor with their fans, and those with fans that cool
only the power supply. The only difference between the two is the location of ventilation
holes. Otherwise, all ATX power supplies look much the same and measure 5.8 inches
(150 millimeters) wide, 5.5 inches (140 mm) long, and 3.4 inches (86 mm) thick.
The ATX design served as a pattern for two additional power supplies designed for cases
more compact than those used by standard ATX motherboards. These include the SFX
and TFX designs. No matter the physical package, these power supplies offer the same
connectors and signals as the basic ATX power supply.
The SFX and SFX12V power supplies are small form-factor designs meant for smaller
computers (hence the SF in the designation). The SFX power supply standard envisions
computers needing from 90 to 180 watts. The standard defines two basic packages. One is
called the 40-millimeter profile design thats about 5 inches (125 mm) long, 4 inches (100
mm) wide, and 2 inches (50 mm rather than 40 mm) thick. In addition, the standard also
allows for a 60-millimeter profile version thats actually 63.5 millimeters (2.5 inches) thick.
The standard also allows manufacturers to add external fans to the 60-millimeter profile
power supplies either as external or internal fans. In either case, the fan adds about 17
millimeters to the thickness of the power supply. The SPX12V design provides one or
more four-pin +12V power connectors, which the ordinary SPX does not.
The TFX12V is a thin form-factor design (hence the designation) made to match microATX
and FlexATX motherboards and fit into low-profile chassis. Under the standard, a TFX12V
power supply measures 6.9 inches (175 millimeters) long, 3 inches (75 mm) high, and 2.6
inches (65 mm) thick. The standard does not define the total output of the supply (or the
current to be supplied at any of its output voltages) but rather envisions supplies with outputs in the range of 180 to 220 watts. The 12V in the name indicates that each TFX12V
power supply makes available one or more four-pin +12V power connectors.
CHAPTER 31 POWER
Power Supplies
The official specifications for ATX, SFX, and TFX power supplies are available on the
Web at www.formfactors.org.
In modern computer designs, only the WTX power supply design does not follow the
connector and signal standards popularized by ATX. WTX power supplies are meant to
match WTX motherboards in workstations and servers, applications demanding more
power than the ATX family can deliver.
WTX recognizes two sizes of power supply: single-fan power supplies capable of delivering from 400 to 460 watts, and a two-fan design meant to supply from 550 to 850 watts.
In addition to harboring an additional fan, two-fan designs are physically larger. A singlefan WTX power supply measures 9.1 inches (230 mm) long, 5.8 inches (150 mm) wide,
and 3.4 inches (86 mm) thick. A two-fan supply measures 9.1 inches (230 mm) long, 8.8
inches (224 mm) wide, and 3.4 inches (86 mm) thick.
Either size of WTX power supply offers five power connectors that do not match the
ATX standard. These include a primary 24-pin connector that supplies most of the power,
a 22-pin connector with control signals (as well as some power), an eight-pin connector
meant to supply voltage converters for low-voltage memory and processors, and two sixpin connectors that supply high-current 12VDC to power pods for high-current processors and similar applications.
The complete WTX specification is available on the Web at www.wtx.org. This standard
is now being phased out. The Server System Infrastructure group at www.ssi.org has
developed a new set of specifications for servers and their power supplies.
Portable Computer Power

In comparison to desktop computers, notebook and subnotebook computers would seem
to have it easy when it comes to power supplies. Notebooks work with battery power,
which is generated inside the battery cells in exactly the right form for computer circuitslow voltage DC. But notebook computers actually have much more complex power
systems, more complex than desktop computers because of their use of two power systems, utility and battery.
Even though battery power is smooth direct current, battery-powered computers require
built-in voltage regulation because the voltage from batteries varies as cells discharge. In
addition, most computer batteries are rechargeable, so they need to get electricity from
somewhere. Moreover, most of the time laptop and notebook computers are close to electrical supplies when they are used, so it makes sense to use utility power rather than battery poweror charge batteries at the same time. Consequently, notebook computers also
use power supplies, but such supplies have a significantly different design, one that splits
the desktop power supply design into pieces.
1021
1022
The power supplies of most notebook computers have an external half that reduces utility
voltage to a safe, near-battery level and rectifies it into DC. These external power supplies
generally create only one voltage, one that will substitute for battery power. But thats not
the end of it. Inside the notebook computers case are several voltage regulators that keep
these low voltages at the constant levels logic circuits require. These voltage regulators
wallpaper over the wide variations that occur to the voltage output of a battery as it discharges. The regulators also ensure that the low-voltage power from the external power
supply exactly matches battery power before it gets to critical circuits. In addition, notebook computers have another, specialized regulator that desktop computers lack, one that
charges the notebook computers battery reserves and keeps the batteries topped up when
external utility power is available.
Unlike the power supplies and voltage regulators for desktop computers, those for notebooks do not follow any industry standards. Each make and model of machine is different.
In general, the power circuitry isnt even in one place or one module (as with desktop
computers) but rather spread throughout the computer system.
The notebook computer is more concerned with minimizing power waste rather than regulating. Todays power-management functions, which save energy in both desktop and
notebook computers, had their origins in the power-saving features of notebook machines.
External Power Supplies

In essence and operation, the external power supply of a notebook computer is little more
than a repackaged version of those inside desktop computers. Line voltage AC goes in,
and low voltage DC (usually) comes out. The output voltage is close to that of the systems battery output, always a bit higher. (A slightly higher voltage is required so that the
batteries are charged to their full capacity.)
In most rudimentary form, the external power supply is nothing more than a transformer
that supplies low-voltage AC to a computer. This approach has two benefits. It moves the
heaviest element of the power supply out of the computer, reducing its weight and making
it more portable. In addition, it keeps dangerous high voltages out of the computer itself,
not only making the computer inherently safer but also reducing the number of hurdles
the manufacturer must leap to get regulatory approval for the entire computer package.
Current external power supply designs are more elaborate. Instead of AC, they supply DC
to the computer. More importantly, they benefit from the latest power technologies and
eliminate the large, heavy (and expensive) power transformers required by older designs.
As a result, the external power supply weighs less and is more energy efficient.
Despite this design change, many people still call the external power supply a transformer.
Another common name is power brick because of the shape of the supply (although the
brick-like weight left the design along with the power transformer).
CHAPTER 31 POWER
Power Supplies
No standard exists for the external battery chargers/power supplies of notebook computers.
Every manufacturerand often every model of computer from a given manufactureruses
its own design. They differ as to output voltage, current, and polarity. You can substitute a
generic replacement only if the replacement matches the voltage used by your computer
and generates at least as much current. Polarity matching gives you two choicesright and
wrongand the wrong choice is apt to destroy many of the semiconductors inside the system. In other words, make extra certain of power polarity when plugging in a generic
replacement power supply. (With most computers, the issue of polarity reduces to a practical matter of whether the center or outer conductor of the almost-universal coaxial power
plug is the positive terminal.)
Most external power supplies are designed to operate from a single voltage (a few are universal, but dont count on it). That means you are restricted to plugging in and charging
your portable computer to one hemisphere (or thereabouts) or the other. Moving from a
117-volt to a 230-volt electrical system requires a second, expensive external charger.
Experienced travelers often buy a second external supply/charger at travelling voltage.
Car and Airplane Adapters

Utility power is not the only external source available for powering computers. Both automobiles and airplanes have their own electrical systems that can be used as an external
power source for notebook computers. However, you must still match the available power
to the requirement of your computer.
Today, the standard power system in automobiles is direct current of approximately 12
volts. The 12-volt rating is only nominal, however. When your car is running and charging its battery, the voltage in the electrical system is often as high as 16 volts. When the
car is starting, the available voltage may dip below 8 volts. In fact, even a so-called 12-volt
car battery, when fully charged, actually delivers 13.2 volts.
A few computers managed to deal with the wide range of voltages in automotive electrical
systems directly. They required only a simple cable with a plug that fit a cigarette lighter
to operate from automotive power. But most current computers are more demanding in
their power requirements, often needing voltages greater than available from standard 12volt automotive systems. Although you can power these computers from an automobile
cigarette lighter, you need a more costly kind of adapter called an inverter.
An inverter is a device that converts DC to AC and increases the voltage to that of utilitysupplied powerinverting what you expect a power supply to do. You simply plug your
notebook computers external power supply into the inverter exactly as you would plug
into utility power; then you plug your computer in normally.
In some commercial airplanesprimarily in first- and business-class sections but occasionally in economyyoull find power jacks meant to supply electricity to portable computers. These jacks supply the equivalent of automobile power, regulated to a constant 15
1023
1024
volts of direct current (usually limited to about 75 watts per seat). This power is not
meant to directly operate notebook computers but to supply an adapter, which in turn
powers the computer. Each computer model is supposed to use its own adapter.
The airline jacks use a special connector manufactured by the Hypertronics Corporation
but commonly referred to as an Empower connector or ARINC 628 power connector. One
exception is American Airlines, which uses an automotive-style cigarette lighter connector
for in-seat power.
You can avoid the need for a special computer power adapter by using an inverter. Several
manufacturers now offer small inverters with both airline-style and automotive connectors. The inverters produce the equivalent of normal house current into which you can
plug your computers standard AC adapter.
Note that these airplane power jacks are meant as a convenience for passengers. The main
electrical power system of most airplanes does not operate at 15 volts nor is it compatible
with house current.
Batteries
Outside of the physical difference in packaging, notebook computers differ from desktop
machines most significantly in that they are designed with self-contained power systems.
They hold their own electricity and can run on batteries. The power versatility is what
gives the notebook machine the ability to compute anywhere.
But batteries play a larger role in modern computing. Wireless keyboards and mice, digital cameras, and MP3 players all rely on batteries as their primary source of power (as
likely does your cell phone and a host of other electronic gadgets you take for granted).
Its enough to make you want to buy stock in a battery companyand shudder to think
what all those throw-away batteries are doing to landfills.
Batteries represent the primitive side of electricity, the chemical side. Its a territory
strewn with things that get your hands dirty, including smudgy black carbon and a host of
toxic materials you hope you never do get on your fingers. Batteries have stubbornly
refused to give in to micro-miniaturization, yielding only small increases in capacity with
every investment in research and development.
We often forget the chemical nature of batteries because the chemistry is all sealed away,
usually permanently. To use the battery usually is a small cylinder that produces electricity
(or, in the case of batteries for notebook computers, an expensive plastic shell that doesnt
hold nearly enough electricity). In any case, the outward manifestations of the battery are
CHAPTER 31 POWER
Batteries
physical and electrical. Its physical size and shape determine where it will fit. Its electrical
ratings determine what it can run.
The most popular batteries for small electronic devices come in standard sizes. The battery packs for notebook computers are usually tailored to a specific model of machine,
although many such packs comprise a set of standard batteries permanently connected
together. All batteries produce the same kind of electricitydirect currentbut they vary
in the amount of energy they can store and several other electrical characteristics.
Cell Types
Batteries can be divided into two types: primary and secondary or storage. In primary batteries, the creation of electricity is irreversible; one or both of the electrodes is altered and
cannot be brought back to its original state except by some complex process (such as resmelting the metal). Secondary or storage batteries are rechargeable; the chemical reaction is reversible by the application of electricity. The electrons can be coaxed back from
whence they came. After the battery is discharged, the chemical changes inside can be
reversed by pumping electricity into the battery again. The chemicals revert back to their
original, charged state and can be discharged to provide electricity once again.
In theory, any chemical reaction is reversible. Clocks can run backwards, too. And pigs
can fly, given a tall enough cliff. The problem is that when a battery discharges, the chemical reaction affects the electrodes more in some places than others; recharging does not
necessarily reconstitute the places that were depleted. Rechargeable batteries work
because the chemical changes inside them alter their electrodes without removing material. For example, an electrode may become plated with an oxide, which can be removed
during recharging.
Primary and secondary (storage) batteries see widely different applications, even in computers. Nearly every modern computer has a primary battery hidden somewhere inside,
letting out a tiny electrical trickle that keeps the time-of-day clock running while the PC
is not. This same battery also maintains a few bytes or kilobytes of CMOS memory to
store system configuration information. Storage batteries are used to power just about
every notebook computer in existence. (A few systems use storage batteries for their clocks
and configuration memory.)
Voltage
The most important of these is voltage, which describes the electrical potential of a battery,
the force with which the battery can move electrons through circuitry. The technical term
is electromotive force (EMF), but most people usually talk about its direct measure, volts.
All batteries have a voltage rating that is both unchangeable and varying. That is, the voltage of a battery cell is characteristic of the cell design and the chemical reaction taking
place inside, and this reaction does not change. But the voltage produced by the reaction
1025
1026
varies with temperature (most batteries produce lower voltage as the temperature declines),
the age of the cell (most batteries produce lower voltage as they age), and load (most batteries produce lower voltages when they are called upon to deliver more current).
These factors result in battery voltage varying widely from the nominal or rated voltage.
Cells may start life producing 1.8 volts and remain useful until their output falls to half
that. Because of the wide variance of cell voltage, most equipment that uses battery power
is either insensitive to the exact voltage supplied or regulates the supplied voltages so that
the internal circuitry of the equipment sees a constant voltage no matter the exact voltage
produced by battery cells. Consequently, typical commercial cells that use carbon-zinc
(nominally rated at 1.5 volts), nickel-cadmium (nominally rated at 1.2 volts), and lithium
disulfide (nominally rated at 1.6 volts) are essentially interchangeable.
Depending on the chemistry used, a single cell can produce anywhere from a small fraction of a volt to somewhat more than three volts. Batteries rated with voltages higher than
about three are composites of several cells linked together. (Technically, the term battery
describes a collection of several individual electrochemical cells, although in common
usage a commercial battery may only be a single cell.)
Current
Current describes the number of electrons the potential can push, the quantity of electricity. Current is measured in amperes (named after the French mathematician and physicist
Andr Marie Ampre, 17751836), usually clipped to the term amps.
Battery cells are limited in the current they can produce by their designs and chemistries.
In theory, if the entire chemical reaction in a battery cell occurred instantly, the cell would
produce unlimited currentfor an instant. Practical factors limit the chemical reaction
rate and the current a cell can produce. Chief among these are the basic reaction rate of
the chemicals, the design of the cell, and the area over which the reaction takes place.
Consequently, some cells are inherently able to produce high currents. Others can only
product weak currents. For example, the currents produced by lead-acid batteries and
nickel-cadmium cells are so high, such batteries can melt metals and start fires when
shorted out. Put an unpackaged ni-cad battery in your pocket, and it may short out
against your keys and loose change. The high current and heat from the short circuit
could start a fire. Consequently, these high-current cells often wear warning labels.
Cell size is also an important factor in determining the reaction area of the cell chemistry
and consequently the current-creating capabilities of the cell. Making a cell larger
increases the current it can produce, so heavy-duty applications often require large cells.
D cells can produce more current than AA cells.
The various factors in cell design and chemistry essentially reduce to a single mathematical factorthe equivalent internal resistance of the cell, which determines current
capabilities. A low internal resistance allows high currents.
CHAPTER 31 POWER
Batteries
Energy and Capacity

Voltage and current are instantaneous values that describe battery characteristics that are
relevant when determining what kind of device the battery can power. Electrical circuits
and motors vary in their voltage and current needs, and they must be tailored to match
the battery used to power them.
The actual power that a battery can produce is the product of the voltage and current and
is described in watts (named in honor of Scottish inventor and engineer James Watt,
17361819). A batterys power is independent of its sizea battery designed to produce a
high current can generate tremendous power, although briefly. For example, even a small
AA ni-cad battery can create enough current to melt metal, but it wouldnt last very long
when challenged with the task of running a notebook computer.
A more relevant measure is the energy a given battery can produce. Energy is the amount
of power a battery can produce over an extended period. A common measure is the watthour, the steady production of one watt of power for one hour.
The rated capacity of a cell or battery is the amount of electricity or electric charge it
can produce when fully charged under specified conditions. As with voltage, the actual
amount of charge the battery can produce varies with its temperature and the discharging
current.
In science, the standard unit for measuring battery capacity is the coulomb (named after
French physicist C. A. de Coulomb, 17361806), which describes the time the battery can
produce a given current. One coulomb is one ampere produced for one second. In practice, however, cell or battery capacity is more commonly expressed in ampere-hours (AH)
or milliampere-hours (mAH), equal to 3600 times the coulomb rating. The total energy in
a battery is its capacity multiplied by its voltage (which results in a measurement of watthours).
Storage Density
The ratio of capacity to the weight (or size) of the battery is called the storage density of
the battery. For you, as a battery user and the person charged with carrying around a
notebook computer weighed down with batteries, this is a most important measure.
The storage density of a battery determines how heavy a load of batteries your computer
needs for a given runtime. The higher the storage density of a battery, the more energy
that can be stored in a given size or weight of a cell and, hence, the more desirable the
battery.
The chemistry used by a battery to store or produce electricity is the primary factor in
determining the storage density of a battery. Table 31.2 lists the storage density of the
major chemical systems used in storage batteries for personal computer and cell phone
applications, expressed in watt-hours per kilogram of weight (Wh/kg).
1027
1028
TA B L E 3 1 . 2
Storage Densities of Common Battery Technologies
Cell Type
Nominal Voltage
Storage Density
Lead-acid
2.1 volts
30 Wh/kg
Nickel-cadmium
1.2 volts
40 to 60 Wh/kg
Nickel-metal hydride
1.2 volts
60 to 80 Wh/kg
Circular lithium ion
3.6 volts
90 to 100 Wh/kg
Prismatic lithium ion
3.6 volts
100 to 110 Wh/kg
Polymer lithium ion
3.6 volts
130 to 150 Wh/kg
The storage density of each chemistry falls into a range of values because the actual construction of a battery cellthe materials used and the layoutalso affects its storage density.
Shelf Life
No battery stores energy forever. The chemicals inside the cells inevitably react and
slowly degrade. As a result, the charge stored by the battery degrades as well. The degradation takes two forms.
Some of the chemical reactions permanently affect the ability of the cell to store chemical
energy. After a while, the battery loses its usefulness and becomes nothing more than a
colorful piece of clutter. The period during which the cell remains useful is termed its shelf
life. The chemistry and construction of the cell determine its shelf life, as do the conditions of storage. Some cells, such as modern lithium designs, have shelf lives in excess of
ten years, whereas other cells may deteriorate in a matter of weeks (for example, zinc-air
batteries once activated). Poor storage conditionsespecially high temperaturesusually
accelerate the degradation of cells, whereas refrigeration (and, with some chemistries,
freezing) often prolongs shelf life.
In secondary cells, the reversible chemical reactions that produce electricity slowly take
place even when the cell is not used. These reactions discharge the cell as if it were being
used and are consequently called self-discharge. As with normal discharge of the cells, the
reactions of self-discharge are reversible by simply recharging the cells. The self-discharge
rates of batteries vary with the same factors affecting shelf life, although in modern cells
the chemistry and cell design are the major determinants. Some chemistries lose as much
as 10 percent of their charge in a day; others less than 1 percent.
Chemical Systems
The chemical reactions in the cell are the most important factor constraining energy density and the usefulness of batteries. In fact, the entire history of battery technology has
been mostly a matter of finding and refining battery chemistries to pack more energy in
CHAPTER 31 POWER
Batteries
ever-smaller packages. Todays batteries use a variety of chemical systems, some dating
from the late 19th Century, as mentioned previously, and some hardly a decade old. The
diversity results from each having distinct benefits for particular applications. The following battery chemistries are the most popular for portable computer, cell phone, power system, and peripheral applications.
Carbon-Zinc
The starting point for battery technology is the carbon-zinc cell, the heir of Georges
Leclanchs 1866 invention of the wet cell for producing electricity. Carbon-zinc cells are
probably the most common batteries in the world, known under a variety of names,
including dry cell and flashlight battery. When you think of batteries, its likely that carbon-zinc cells first come to mind. One company alone, Energizer, sells over six billion
carbon-zinc cells each year. They are the lowest priced primary cells. They also have the
lowest storage density of any common battery.
One reason carbon-zinc cells are so popular is that the name actually describes two or
three different chemistry systems. These include Leclanch cells, zinc chloride cells, and
alkaline batteries.
The name describes the basic chemistry of the cells. In the basic carbon-zinc cell, the
carbon in the name is a cathode current collectora carbon rod in the center of the
cell. The actual material of the cathode is a mixture of manganese dioxide, carbon conductor, and electrolyte. The zinc serves as the anode and often serves as the metal shell of
the battery. The electrolyte is a complex mixture of chemicals that typically includes
ammonium chloride, manganese dioxide, and zinc chloride.
The electrolyte is the chief difference between Leclanch and zinc-chloride cells. The former use a slightly acidic mix of ammonium chloride and zinc chloride in water. The electrolyte in zinc-chloride cells is mostly zinc chloride. Zinc chloride cells produce a slightly
higher open-circuit voltage than Leclanch cells (1.6 versus 1.55 volts).
Although zinc-chloride cells typically have a greater capacity than Leclanch cells, this
difference shrinks under lighter loads, so zinc-chloride cells are often termed heavyduty. In any case, the efficiency of any carbon-zinc cell decreases as the load increases
doubling the current drain more than cuts in half the capacity of the cell. The most
efficient strategy is to use as large a cell as is practical for a given application. Thats
why power-hungry toys demand D batteries and low-drain transistor radios make do
with AA cells.
Alkaline batteries, no matter the advertised claims, are little more than an enhancement of
19th Century carbon-zinc technology. The biggest change in chemistry is an alteration to
the chemical mix in the electrolyte that makes it more alkaline (what did you expect?).
This change helps to increase storage density and shelf life of the cells.
1029
1030
The construction (as opposed to chemistry) of alkaline cells differs significantly from ordinary carbon-zinc cells, however. Alkaline cells are effectively turned inside-out. The shell
of the alkaline battery is nothing more than thata protective shelland it does not play
a part in the overall chemical reaction. The anode of the cell is a gelled mixture of powered zinc combined with the electrolyte (itself a mixture of potassium hydroxidea strong
alkaliand water), and the combination is linked to the negative terminal of the cell by a
brass spike running up the middle of the cell. The cathode, a mixture of carbon and manganese dioxide, surrounds the anode and electrolyte, separated by a layer of nonwoven
fabric such as polyester.
Depending on the application, alkaline cells can last for four to nine times the life of more
traditional carbon-zinc cells. The advantage is greatest under heavy loads that are infrequently usedthat is, something that draws heavy current for an hour once a day rather
than a few minutes of each hour.
Carbon-zinc cells nominally produce 1.5 volts, but this full voltage is available during the
initial discharge of the cell. The voltage of the cell diminishes as the load to the cell
increases and as the charge of the cell decreases.
Standard nine-volt batteries also use carbon-zinc chemistry. To produce the higher voltage, six separate carbon-zinc cells are stacked and connected in series inside each battery.
Higher-voltage carbon-zinc cells can be made similarly, such as the B batteries of the
1950s, which stacked from 45 to 90 volts of cells to power vacuum-tube portable radios.
Ordinarily, alkaline batteries cannot be recharged because the chemical reactions in the
cell cannot be readily reversed. If you attempt to recharge an ordinary carbon-zinc cell, it
acts more like a resistor than a storage cell, turning the electricity you apply to it into
heat. Apply too much power to a cell and it will heat up enough to explodea good reason never to attempt to recharge carbon-zinc or alkaline batteries.
The exceptions to this rule are the Renewal batteries produced under license by Rayovac
Corporation. The Renewal design relies on a two-prong attack on carbon-zinc technology. The Renewal cell is fabricated differently from a standard cell. More importantly,
Renewal batteries are part of a system that requires a special battery charger. Instead of
applying a nearly constant current to recharge the cells, the Renewal charger adds power
in a series of pulses. A microprocessor in the charger monitors how each pulse affects the
cell to prevent overheating. Even with the novel charger, however, Renewal cells have a
limited life, typically between 25 and 100 charge/discharge cycles. In that Renewal cells
cost only about twice as much as standard alkaline cells, they can be very cost-effective in
some applications.
Lead-Acid
The most common storage batteries in the world are the lead-acid batteries used to start
automobiles. Gaston Plant developed the first lead-acid cell in France in 1859, and the
CHAPTER 31 POWER
Batteries
design remains virtually unchanged today. Lead-acid cells use anodes made from porous
lead and cathodes made from lead oxide, both soaked in a sulfuric acid electrolyte.
The lead in the cells makes these batteries inherently heavy. Filled with highly corrosive
acid (which is also heavy), they are cumbersome and dangerous. Not only can the acid and
its fumes damage nearby objects (particularly metals), but overcharging cells also results in
electrolysis of the water component of the internal acid. The hydrogen released by the
electrolysis is highly combustible and, mixed with air and a spark, prone to a Hindenburglike explosion.
The breakdown of the water in the cells also has another effect: It reduces the overall
amount of water in the cell. Too little water reduces the reaction area inside the cell,
reducing its capacity. It also allows the cells to deteriorate by atmospheric action. The
electrodes can flake and possibly short out a cell entirely, reducing its capacity to zero.
Early lead-acid cells consequently required regular maintenance to keep the water/acid
inside the cell at the proper level. Only the water electrolyzes in the battery, so only it
needs to be replaced. To avoid contaminating the battery chemistry, manufacturers recommend you use only distilled water to replenish the battery. Judging the proper amount to
add requires only refilling the battery to its normal level. If the battery carries no mark as
to the proper level, you should fill it so that the liquid in the battery just covers the electrode plates inside the cell.
In stationary applications, lead-acid batteries were once cased in glass. Not only would it
resist the internal acid, but it also allowed maintenance workers to quickly assess the condition of the cells. Automotive applications required a more shatterproof case, for which
engineers developed hard rubber or plastic enclosures.
The convenience of using lead-acid batteries is immensely increased by sealing the cells.
The result is the so-called maintenance-free battery. Because the vapors within the cell cannot escape, electrolysis losses are minimized. The maintenance-free battery never needs
water (or at least it shouldnt).
However, maintenance-free batteries are not entirely trouble free. They still have acid
sloshing around inside them, after all, and it can leak out the battery vent, damaging the
battery compartment or even the equipment in which the battery is located. Engineers
developed two ways of eliminating the slosh. One keeps the liquid acid inside a plastic
separator between cell electrodes (typically a micro-porous polyolefin or polyethylene).
The other alternative chemically combines the liquid electrolyte with other compounds
that turn it into a gela colloidal form like gelatinwhich is less apt to leak out.
Lead-acid cells have several other drawbacks besides the cantankerous nature of their
acidic contents. As noted before, they are heavy. The energy a lead-acid battery stores
per pound of battery is lower than just about any technology short of a potato wired with
zinc plates. This is the chief frustration of automotive engineers who would like to use
1031
1032
low-cost lead-acid batteries to power electric cars. Once a sufficient number of lead-acid
batteries to move the car a worthwhile distance get piled into the car, it weighs more like
an electric truck.
On the other hand, besides being cheap, lead-acid batteries have over 150 years of technical development behind them. They can be custom-tailored to specific applications, such
as those requiring deep discharge cycles (for example, where the batteries are used as the
sole power source for electrical equipment) or for battery backup uses, such as in large
uninterruptible power supply systems in data centers. Lead-acid cells also have a low
internal resistance and therefore can produce enormous currents. They suffer no memory
effect as do some more exotic cell designs, such as nickel-cadmium cells. (This effect, discussed in relation to nickel-cadmium batteries, reduces the capacity of a cell if it gets
recharged before being fully discharged). The cells also have a moderately long, predictable life. And, of course, they are cheap.
Most uninterruptible power systems rely on gelled lead-acid cells for their power reserves.
In this application, they require little maintenance. That is, you dont have to do anything
to keep them going. The power system, however, must be tailored to the needs of the
cells. Gel cells are degraded by the application of continuous low-current charging after
they have been completely charged. (Most lead-acid batteries are kept at full capacity by
such trickle charging methods.) Consequently, gel cells require special chargers that
automatically turn off after the cells have been fully charged. The chargers switch back on
when the battery dischargeseither under load or by self-dischargeto a predetermined
level. Uninterruptible power systems typically check their batteries periodically (usually
weekly) to ensure they maintain a full charge.
Nickel-Cadmium
In consumer electronic equipment, the most popular rechargeable/storage batteries are
nickel-cadmium cells, often called ni-cads. These batteries use cathodes made from nickel
and anodes from cadmium, as the name implies. Their most endearing characteristic is the
capability to withstand a huge number of full charge/discharge cycles, in the range of 500
to 1000, without deteriorating past the point of usefulness. Ni-cads are also relatively
lightweight, have a good energy storage density (although about half that of alkaline cells),
and tolerate trickle charging (when properly designed). On the downside, cadmium is
toxic, thus the warning labels that implore you to be cautious with them and properly dispose of them.
The output voltage of most chemical cells declines as the cell discharges, because the
reactions within the cell increase its internal resistance. Ni-cads have a very low internal
resistancemeaning they can create high currentswhich changes little as the cell discharges. Consequently, the ni-cad cell produces a nearly constant voltage until it becomes
almost completely discharged, at which point its output voltage falls precipitously.
CHAPTER 31 POWER
Batteries
This constant voltage is an advantage to the circuit designer because fewer allowances
need to be made for voltage variations. However, the constant voltage also makes determining the state of a ni-cads charge nearly impossible. As a result, most battery-powered
computers deduce the battery power they have remaining from the time they have been
operating and known battery capacity rather than by actually checking the battery state.
Ni-cads are known for another drawback: memory. When some ni-cads are partly discharged and then later recharged, they may lose capacity. Chemically, recharging ni-cads
before they are fully discharged often results in the formation of cadmium crystals on the
anodes of the cell. The crystals act like a chemical memory system, marking a second discharge state for the cell. When the cell gets discharged to this secondary discharge state,
its output abruptly falls despite further capacity being available within the cell. In subsequent cycles, the cell remembers this second discharge level, which further aggravates the
situation by reinforcing the memory of the second discharge state. The full capacity of the
cell can only be recovered by nudging the cell past this second discharge state. This will
erase the memory and restores full cell capacity.
As a practical matter, the cure for the memory problem is deep dischargedischarging
the battery to its minimum working level and then charging the battery again. Deep discharge does not mean totally discharging the battery, however. Draining nearly any storage battery absolutely dry will damage it and shorten its life. If you discharge a ni-cad
battery so that it produces less than about one volt (its nominal output is 1.2 volts), it may
suffer such damage. Notebook computers are designed to switch off before their batteries
are drained too far, and deep discharge utilities do not push any farther, so you need not
worry in using them. But dont try to deeply discharge your systems batteries by shorting
them outyou risk damaging the battery and even starting a fire.
According to battery-makers, newer ni-cads and nickel-metal hydride cells are free from
memory effects, although this has not been proven in practice. Some lithium batterymakers
claim that the memory effect results from the use of nickel rather than cadmium (a view not
supported by the chemistry), and some users also report contrary experiences with both
nickel-based battery types. In any case, to get the longest life from ni-cads, the best strategy
is to operate them between extremesoperate the battery through its complete cycle.
Charge the battery fully, run it until it is normally discharged, and then fully charge it again.
As with lead-acid batteries, nickel-cadmium cells are also prone to electrolysis breaking
down water in the electrolyte into potentially explosive hydrogen and oxygen. Batterymakers take great steps to reduce this effect. Commercially available ni-cads are sealed to
prevent leakage. They are also designed so that they produce oxygen before hydrogen,
which reacts internally to shut down the electrolysis reaction.
To prevent sealed cells from exploding should gas somehow build up inside them, their
designs usually include resealable vents. You risk the chance of explosion if you encase a
1033
1034
ni-cad cell in such a way it cannot vent. The vents are tiny and usually go unnoticed.
They operate automatically. The warning against blocking the vents applies mostly to
equipment-makers. Standard battery holders wont block the vents, but encapsulating the
battery epoxy to make a solid power module certainly will.
Nickel-Metal Hydride
Chemically, one of the best cathode materials for battery cells would be hydrogen. But
hydrogen is problematic as a material for batteries. At normal temperatures and pressures,
hydrogen is a lighter-than-air gas, as hard to hold to as grabbing your breath in your
hands.
In the late 1960s, however, scientists discovered that some metal alloys have the ability to
store atomic hydrogen 1000 times their own volume. These metallic alloys are termed
hydrides and typically are based on compounds such as LiNi5 and ZrNi2. In properly
designed systems, hydrides can provide a storage sink of hydrogen that can reversibly
react in battery cell chemistry.
The most common cells that use hydride cathodes carry over the nickel anodes from nicad cell designs. These cells typically have an electrolyte of a dilute solution of potassium
hydroxide, which is alkaline in nature.
Substituting hydrides for cadmium in battery cells has several advantages. The most obvious is that such cells eliminate one major toxic material, cadmium. No cadmium also
means that the cells should be free from the memory effect that plagues ni-cad cells. In
addition, hydrogen is so much better as a cathode material that cells based on nickel and
metal hydrides have a storage density about 50 percent higher than nickel-cadmium cells.
In practical terms, that means cells of the same size and about the same weight can power
a notebook computer for about 50 percent longer.
Cells based on nickel and metal hydridesoften abbreviated as NiMH cellsare not perfect. Their chief drawback is that most such cells have a substantially higher self-discharge
rate than do ni-cad cells. Some NiMH cells lose as much as five percent of their capacity
per day, although this figure is coming down with more refined cell designs.
As with ni-cads, NiMH cells have a nominal output voltage of 1.2 volts that remains relatively flat throughout the discharge cycle, falling precipitously only at the end of the useful charge of the cell. (Fully charged, a NiMH cell produces about 1.4 volts, but this
quickly falls to 1.2 volts, where it remains throughout the majority of the discharge cycle.)
In many ways NiMH cells are interchangeable with ni-cads. They have a similar ability
to supply high currents, although not quite as much as ni-cads. NiMH cells also endure
many charge/discharge cycles, typically up to 500 full cycles, but they are not a match for
ni-cads.
CHAPTER 31 POWER
Batteries
Although the discharge characteristics of NiMH and ni-cads are similar, the two cell types
react differently during charging. Specifically, ni-cads are essentially endothermic while
being charged, and NiMH cells are exothermicthey produce heat. As the NiMH cell
approaches full charge, its temperature can rise dramatically. Consequently, chargers are
best designed for one or the other type of cell. NiMH cells work best in chargers designed
for them. NiMH cells do, however, readily accept trickle charging (discussed later).
Lithium Ion
Lithium is the most chemically reactive metal and provides the basis for todays most
compact energy storage for notebook computer power systems. Nearly all high-density
storage systems use lithium because it has an inherent chemical advantage. Lithium has a
specific capacity to store 3860 ampere-hours per kilogram of mass, compared to 820
AH/kg for zinc and 260 AH/kg for lead.
Lithium is also very reactive. Depending on the anode, cells with lithium cathodes can
produce anywhere from 1.5 volts to 3.6 volts per cell, higher voltage than any other
chemistry.
The problem with lithium is that it is too reactive. It reacts violently with water and can
ignite into flame. Batteries based on lithium metal were developed and manufactured in
the 1970s, and in the 1980s some companies introduced commercial rechargeable cells
based on metallic lithium. Such batteries quickly earned a reputation for doubtful safety.
To prevent problems caused by reactive metallic lithium, battery-makers refined their
designs to keep the lithium in its ionic state. In this way, they were able to reap the electrochemical benefits of lithium-based cells without the safety issues associated with the
pure metal. In lithium ion cells, the lithium ions are absorbed into the active material of
the electrodes rather than being plated out as metal.
The typical lithium ion cells use carbon for its anode and lithium cobalt dioxide as the
cathode. The electrolyte is usually based on a lithium salt in solution.
Lithium batteries offer higher storage densities than nickel-metal hydride cells, which
equates to using them in notebook computer systems for about fifty percent longer without
a recharge. Lithium ion cells also lack the memory effect that plagued early nickel-cad cells.
On the other hand, current lithium cells have a higher internal resistance than nickel-cadmium cells and consequently cannot deliver high currents. A ni-cad could melt a screwdriver, but a lithium cell cannotthats why lithium cells dont wear the same warnings as
ni-cads. The available power is sufficient for a properly designed notebook computer that
minimizes surge requirements (meaning that certain devices, such as disk drives, may
require a fair surge of power during certain phases of operation, notably spin-up).
Moreover, the life of lithium cells is more limited than that of nickel-based designs,
although lithium ion cells withstand hundreds of charge/recharge cycles.
1035
1036
Because lithium ion cells use a liquid electrolyte (although one that may be constrained in
a fabric separator), cell designs are limited to the familiar cylindrical battery form.
Although such designs are no more handicapped than they are with other battery
chemistries, the lithium ion chemistry lends itself to other, space-saving designs based on
polymerized electrolytes.
Lithium Polymer
A refinement of familiar lithium chemistry, called the lithium solid polymer cell, promises
more power for portable applications, but through packaging rather than high-energy
density. Whereas conventional lithium ion cells require liquid electrolytes, solid polymer
cells integrate the electrolyte into a polymer plastic separator between the anode and cathode of the cell. As an electrolyte, lithium polymer cells use a polymer composite such as
polyacrylonitrile containing a lithium salt. Because theres no liquid, the solid polymer cell
does not require the chunky cylindrical cases of conventional batteries. Instead, the solid
polymer cells can be formed into flat sheets or prismatic (rectangular) packages better able
to fit the nooks and crannies of notebook computers.
Although the energy density of solid polymer cells is similar to ordinary lithium ion cells,
computer manufacturers can shape them to better fit the space available in a notebook
machine, squeezing more capacity into otherwise unused nooks and crannies. For example, simply by filling the empty space that would appear in the corners around a cylindrical cell, a solid polymer battery can fit in about 22 percent more chemistry and energy
capacity. In addition, solid polymer batteries are environmentally friendly, lighter because
they have no metal shell and safer because they contain no flammable solvent. Most battery-makers and computer-makers are switching to the lithium solid polymer cell design.
Lithium-Iron Disulfide
Unlike other lithium cells that have chemistries tuned to obtaining the greatest capacity in
a given package, lithium-iron disulfide cells are a compromise. To match to existing
equipment and circuits, their chemistry has been tailored to the standard nominal 1.5-volt
output (whereas other lithium technologies produce double that). These cells are consequently sometimes termed voltage-compatible lithium batteries. Unlike other lithium technologies, lithium-iron disulfide cells are not rechargeable.
Internally, the lithium-iron disulfide cell is a sandwich of a lithium anode, a separator, and
iron disulfide cathode with an aluminum cathode collector. The cells are sealed but
vented.
Compared to the alkaline cells with which they are meant to compete, lithium-iron disulfide cells are lighter (weighing about 66 percent of same-size alkaline cells), higher in
capacity, and longer in life. Even after ten years of shelf storage, lithium-iron disulfide
cells still retain most of their capacity.
CHAPTER 31 POWER
Batteries
Lithium-iron disulfide cells operate best under heavier loads. In high-current applications,
they can supply power for about 260 percent the time of a same-size alkaline cell. This
advantage is less at lower loads, however, and at very light loads may disappear entirely.
For example, under a 20 mA load, one manufacturer rates its AA-size lithium-iron disulfide cells as providing power for about 122 hours, whereas its alkaline cells will last for
135 hours. With a one-ampere load, however, the lithium-iron disulfide cells last for 2.1
hours versus only 0.8 for alkaline.
You can use lithium-iron disulfide cells wherever you might use zinc-carbon batteries,
although they are cost-effective only under high-current loadsflashlights, motor-driven
devices, and powerful electronics. They are not a wise choice for clocks and portable
radios.
Zinc-Air
Of the current battery technologies, the one offering the densest storage is zinc-air. One
reason is that one of the components of its chemical reaction is external to the battery.
Zinc-air batteries use atmospheric oxygen as their cathode reactant, hence the air in the
name. Small holes in the battery casing allow air in to react with a powered zinc anode
through a highly conductive potassium hydroxide electrolyte.
Originally created for use in primary batteries, zinc-air batteries were characterized by
their long stable storage life, at least when kept sealed from the air and thus inactive. A
sealed zinc-air cell loses only about 2 percent of its capacity after a year of storage. Once
air infiltrates the cell, zinc-air primary cells last only for months, whether under discharge
or not.
Some battery-makers have adapted zinc-air technology for secondary storage. Zinc-air
cells work best when frequently or continuously used in low-drain situations. The chief
drawback of zinc-air batteries is, however, a high internal resistance, which means zinc-air
batteries must be huge to satisfy high-current needsfor notebook computers that means
an auxiliary battery pack about the size of the computer itself.
Standards
Contrary to appearances, most rechargeable batteries used by notebook computers are
standard sizes. Computer manufacturers, however, package these standard-size batteries in
custom battery packs that may fit only one model of computer. Usually you cannot change
the batteries in the pack but must buy a new replacement if something goes awry.
Attempts at standardizing batteries for notebook computers have fallen flat, as witnessed
by the near extinction of Duracells standard designs. Computer-makers find proprietary
cells more profitable. They can put a few dollars of individual cells in a cheap but odd
plastic package and sell the assembly for a hundred dollars or more.
1037
1038
Anyone who has ever received a holiday gift and discovered the batteries werent included
knows the basic rule about the different standard sizes for batteries: The one you need is the
one you dont have or cannot find. Frustrating as the multiplicity in standard sizes may be,
the situation is better than having no existing standards. For example, nearly all battery
packs for notebook computers are nonstandard and consequently very costly. Youre captive
to the manufacturers pricing policy in many cases, although there are lower-cost second
sources for batteries for various notebook computers, cellular phones, and other common
portable electronic devices that use rechargeable batteries. Many people remain suspicious
of the build quality, longevity, and safety of some of these second-source batteries, although
most are likely fine for the majority of targeted applications. Standard batteries, on the other
hand, are readily available from multiple sources and far less expensive than rechargeables.
The chief battery standards now in use originally applied to carbon-zinc cells, but other
technologies (some varieties of lithium, nickel-cadmium, and nickel-metal hydride cells)
now follow the same size standards. These standards specify the basic dimensions of the
batteries, allowing many manufacturers to produce interchangeable cells. Table 31.3 lists
the dimensions of many standard battery types.
TA B L E 3 1 . 3
Dimension of Common Battery Types
Cell Type
Shape
Height
Diameter
AAAA
Cylindrical
42.5 mm
8.3 mm
AAA
Cylindrical
44.5 mm
10.5 mm
AA
Cylindrical
50.5 mm
14.5 mm
Cylindrical
50.0 mm
26.2 mm
Cylindrical
61.5 mm
34.2 mm
Rectangular
48.5 mm
35.55 by 9.2 mm
Cylindrical
29.35 mm
11.95 mm
9 volt
Rectangular
48.5 mm
26.5 by 17.5 mm
Note that these sizes are a physical characteristic only. You might find any battery chemistry
in any cell size, some of which might not be suited to a give application. In other words, just
because a battery fits into a holder is no indication that it will work properly there.
Battery Charging
Just as your electronic gear is sensitive to the kind of electricity you supply it, battery
chemistry is extremely sensitive to the electricity used for charging cells. If the voltage
applied is too low, the cell will output current instead of accepting itthat is, discharging
rather than charging. If the voltage is too high, undesirable reactions can take place that
CHAPTER 31 POWER
Batteries
can destroy the cell. For example, raising the voltage inevitably raises the current, and too
much current can cause the cell to overheat. In addition, trying to charge a cell beyond its
capacity can result in the production of explosive gasesand an explosion itself.
Modern battery chargers are consequently sophisticated electronic devices with many different types of safety circuitry to protect both you and the batteries you want to charge.
Most batteries require a specific charger tailored to their needs. Plug the wrong batteries
into the wrong charger (or the wrong charger into your battery pack), and you may
destroy the batteries, the charger, and even the equipment connected to the batteries.
The chief difference between battery chargers is the mode in which they operate. The
choice is between constant voltage and constant current.
Constant voltage chargers are the simplest. They always produce a specific voltage level but
deliver a current that depends on the charge level of the battery (and environmental factors).
As the battery becomes charged, its voltage increases while that of the charger remains the
same, so there is less difference in potential between the charger and battery. As a result, less
current can flow through the system. When the two are equal, no current flows.
A constant-voltage charger requires little more than a transformer (to reduce line voltage
to the level required for battery charging) and a rectifier (to change the alternating current of utility power into the direct current used to charge batteries). Such simple designs
are often found in the battery chargers used for charging car and boat batteries.
The lead-acid cells used for cars and backup power systems typically use constant-voltage
chargers. In addition, lithium ion cells often use constant-voltage systems, although these
usually are more complex with added circuitry to protect both the batteries and your
safety.
The alternate design for battery chargers maintains a constant current and alters the voltage applied to the batteries to maintain this current. These constant-current chargers vary
the voltage they apply to the batteries to maintain a constant current flow, switching off
when the voltage reaches the level of a full charge. (Remember, the voltage produced by
any cell falls as it becomes discharged.) This design is usually used for nickel-cadmium
and nickel-metal hydride cells or batteries.
In addition to charging cells at the proper rate, battery chargers also face another vital
issue: the proper time to switch off. A charger can destroy a battery by overcharging it.
Depending on the requirements of the battery to be charged and the sophistication of the
charger, the charger may use any of several technologies to determine the proper time to
turn off.
The most straightforward way of determining charge is by the voltage produced by the
battery. The charger monitors the batterys voltage and switches off when it reaches the
cutoff voltage. The voltage-sensing technique is not adequate for many kinds of batteries
1039
1040
in particular, ni-cads have a very linear discharge curve that makes the cutoff voltage difficult to determine.
More advanced charging systems use temperature cutoff. That is, the charger monitors the
temperature of the battery cells and switches off or reduces the charging rate when the
battery begins to heat up (which indicates an overcharge condition). Typically battery
packs using temperature cutoff have built-in thermometers that relay a control signal to
the charger circuitry.
More sophisticated chargers combine the voltage and temperature cutoff (VTCO).
Chargers using this technology may switch from high-current charging to a lower or
maintenance charge rate using circuitry that senses both temperature and voltage.
Standard battery chargers supply less current than the batterys discharge rate. High-current chargers supply a current greater than the batterys nominal discharge rate. Trickle
chargers supply a charging current at a rate so low it only compensates for the self-discharge of the battery (by definition, a trickle charger is one that compensates for self-discharge). The trickle charging rate typically is about one-twentieth to one-thirtieth the
batterys nominal discharge rate. Modern battery chargers often operate at several charging rates, starting at high current and switching to low current as the battery nears full
charge. If the battery is of a type that tolerates trickle charging (ni-cads, for example, do
not), the charge will switch to a trickle rate at the end of the charging cycle.
Engineers design most chargers for computer and cell phone batteries. These chargers
can be plugged into the batteries continuously without any detrimental effects on the batteries.
Smart Battery Specifications

Battery charging is actually an interaction between the battery and the charger. The
charger must attend to the state of the battery if it is to avoid damaging the battery by
overcharging the battery and damaging your state of mind by undercharging the battery
so that you run out of power long before you run out of the need for your computer.
Charging and monitoring the charge of batteries has always been problematic. Both
capacity and charge characteristics vary with the battery type and over the life of a given
battery. The smartest conventional battery chargers monitor not the voltage but the temperature of their subjects, because a sharp rise in temperature is the best indication available to the charger of the completion of its work. Even this rise varies with battery
chemistry, so ni-cad and NiMH batteries present differentand confusingtemperature
characteristics that would lead to a charger mistaking one for the other, possibly damaging
the battery.
CHAPTER 31 POWER
Batteries
The Smart Battery system, developed jointly by battery-maker Duracell and Intel and first
published as the Smart Battery Data Specification, Version 1.0, on February 15, 1995,
eliminates these problems by endowing batteries with enough brains to tell of their condition. When matched to a charger with an equivalent I.Q. that follows the Smart Charger
specification, the Smart Battery gets charged perfectly every time with never a worry
about overcharging.
The Smart Battery system defines a standard with several layers that distribute intelligence between battery, charger, and your computer. It provides for an inexpensive communication link between them and outlines the information that a battery can convey to
its charger and the message format for doing so.
Among other data that the battery can relay are its chemistry, its capacity, its voltage, and
even its physical packaging. Messages warn not only about the current status of the batterys charge but even how many charge/recharge cycles the battery has endured so that
the charger can monitor its long-term prognosis. The specification is independent of the
chemistry used by the battery and even the circuitry used to implement its functions.
The battery packs of nearly all new notebook computers follow the Smart Battery specifications or a similar replacement standard called the Control Method Battery Interface,
because battery management is required for compatibility with the latest versions of
Windows. Use of either of the two battery-management standards is required by the
Advanced Configuration and Power Interface (ACPI). The Control Method Battery
Interface is part of the ACPI standard and effectively supercedes Smart Battery.
Battery Safety
The maximum current any battery can produce is limited by its internal resistance. Zinccarbon batteries have a relatively high resistance and produce small currents, on the order
of a few hundred milliamperes. Lead-acid, nickel-cadmium, and nickel-hydride batteries
have very low internal resistances and can produce prodigious currents. If you short the
terminals of one of these batteries, whatever produces the short circuitwires, a strip of
metal, a coin in your pocketbecomes hot because of resistive heating. For example, you
can melt a wrench by placing it across the terminals of a fully charged automotive battery.
You can also start a fire with something inadvertently shorting the terminals of the spare
nickel-cadmium battery for your notebook or subnotebook computer. Be careful and
never allow anything to touch these battery terminals except the contacts of your notebook computer.
When a battery is charged, a process called electrolysis takes place inside. If you remember
your high-school science experiments, electrolysis is what you did to break ordinary water
into hydrogen and oxygen using electricity. Hydrogen is an explosive gas; oxygen is an oxidizer. Both are produced when charging batteries. Normally these gases are absorbed by
1041
1042
the battery before they can do anything (such as explode), but too great a charging current
(as results from applying too high a voltage) can cause them to build up. Trying to charge a
primary battery produces the same gas build-up. As a result, the battery can explode from
too great an internal pressure or from combustion of the gases. Even if the battery does not
catastrophically fail, its life will be greatly reduced. In other words, use only the charger
provided with a portable computer battery and never try to hurry things along.
Nearly all batteries contain harmful chemicals of some kind. Even zinc-carbon batteries
contain manganese, which is regarded as hazardous. All batteries present some kind of
environmental hazard, so be sure to dispose of them properly. Some manufacturers are
beginning to provide a means of recycling batteries. Encourage them by taking advantage
of their offers.
Power Management
With few advances in battery storage density expected in the near-term future, computermakers have relied on reducing the power consumption of their notebook computers to
extend the time a machine can operate between battery charges.
Engineers can use two basic strategies to reduce the power consumption of computers.
They can design circuits and components to use less power, and they can manage the
power used by the devices. Managing power needs usually means switching off whatever
system components arent being actively used. Although the two design methods can be
used separately, they are generally used in tandem to shrink computer power needs as
much as possible.
Microprocessors, the most power hungry of computer circuits, were among the first
devices to gain built-in power management. System Management Mode endowed processors
with the ability to slow down and shut off unnecessary circuits when they were idle.
Similarly, makers of hard disk drives have added sleep modes to spin down their platters
and reduce power needs. Most computers also incorporate timers to darken their screens
to further conserve power.
Although these techniques can be successful in trimming power demands, they lack a unified control system. In response, the industry developed the Advanced Power
Management (APM) interface to give overall control to the power-savings systems in
computers. More recently, APM has been updated and augmented by the Advanced
Configuration and Power Interface specification.
Advanced Power Management

The Advanced Power Management interface specification was jointly developed by Intel
and Microsoft to integrate the control of hardware power-saving features with software
CHAPTER 31 POWER
Power Management
control. First published in January 1992, as the APM BIOS Interface Specification, the
current version, 1.2, was published in February 1996.
Although nominally a BIOS interface, the APM specification describes a layered control
system that manages computer devices to reduce power consumption using both BIOS
and API interfaces. To be fully functional, APM requires a compatible BIOS and hardware
devices that recognize APM control. In addition, hardware devices may have their own
built-in automatic power-management functions that are not controlled by your computers software. For example, a hard disk drive may automatically power down after a
given period without a specific command from your computer. The APM specification
tolerates but does not affect these built-in functions.
States
APM is an overall system feature. Although it has the ability to individually control the
features of each device it manages, the basic-design APM controls all devices together to
conserve power. It manages system power consumption by shifting the overall operating
mode of the computer, called APM states. APM shifts the operating state of the system
based on the needs of the system as determined from a combination of software commands and events. The various APM states provide for power savings that occur in five
levels (six, if you count normal power-hungry operation). The APM specification gives
each of these levels a specific state name.
The first state, Full On, means that the system is operating at full power without any management at all. The APM software is not in control, and no power savings can be achieved.
A system without APM or with its APM features disabled operates in Full On state.
When the APM system is active, all devices run in their normal, full-power consumption
modes. The system is up and ready to do business, operating in what the specification
calls the APM Enabled state.
In the APM Standby state, the microprocessor may stop, and many of the system devices
are turned off or operate at reduced power. The system usually cannot process data, but
its memory is kept alive and the status of all devices is preserved. When your activity or
some other event requires system attention, the computer can rapidly shift from the APM
Standby to the APM Enabled state.
In the APM Suspend state, the system shifts to its maximum power-savings modemost
devices that follow the APM standard are switched off, and the microprocessor switches to
its lowest power state with its clock turned off. Your computer becomes a vegetable.
Hibernation is a special implementation of the APM Suspend state that allows the system
to be switched entirely off and still be restored to the point at which it entered the APM
Suspend state. When entering the APM Suspend state, the system saves all its operating
parameters. In entering the Hibernation state, the system copies memory and other status
1043
1044
data to nonvolatile storage, such as the hard disk, allowing you to switch off memory
power. A system event can shift back to the APM Enabled state from APM Suspend or
Hibernation, but changing modes from APM Suspend to APM Enabled takes substantially longer than from APM Standby to APM Enabled.
The Off state is exactly what the name implies. Power to the system is entirely off. The
computer is more a mineral than vegetable. The only event that restores the system is
turning it back on. If you enter the Off state directlysay, by switching your computer
offno status information or memory gets saved. The system must run through the
entire boot-up process and starts with a clean slate.
Structure
APM adds a layered control system to give you, your software, and your hardware a
mechanism to shift states manually or automatically.
The bottom layer of the system is the APM BIOS, which provides a common software interface for controlling hardware devices under the specification. APM specifies that the BIOS
have at least a real-mode interface that uses interrupt 15(hex) to implement its functions. In
addition, the APM BIOS may also use 16- or 32-bit protected mode using entry points that
are returned from the protected-mode connection call using the real-mode interrupt.
The APM BIOS is meant to manage the power of the motherboard. Its code is specific to
a given motherboard. Under the APM specification, the APM BIOS can operate independently of other APM layers to effect some degree of power savings in the system by itself.
Your computers operating system can switch off this internal BIOS APM control to manage system power itself, still using the APM BIOS interface functions to control hardware
features.
Linking the APM BIOS to your operating system is the APM driver. The driver provides
a set of function calls to the operating system, which it translates to BIOS interrupts. The
driver is more than a mere translator, however. It is fully interactive with both the BIOS
and operating system. For example, the BIOS may generate its own request to power
down the system, and the driver then checks with the operating system to determine
whether it should permit the power-down operation.
The APM system has a built-in failsafe. The APM driver must interact with the BIOS at
least once per second. If it does not, after a second, the BIOS assumes the operating system has malfunctioned and takes self-contained control. The driver can regain control by
sending the appropriate commands (interrupts) to the BIOS.
Certain system events termed wake-up calls tell the APM system to shift modes. Interrupts
generated by events such as a press of the resume button, the modem detecting an incoming telephone ring, or an alarm set on the real-time clock can command the APM BIOS
to shift the system from the APM Suspend to the APM Enabled state.
CHAPTER 31 POWER
Power Management
Advanced Configuration and Power Interface

Customizing the power use of your computer is a natural part of setting up your system.
Understanding that, Intel, Microsoft, and Toshiba decided that the best way to integrate
power management with your system was to combine it with the computers setup facilities and to give both a common interface. The result is called the Advanced Configuration
and Power Interface, a formal specification that was first published by the threesome as
ACPI version 1.0 in December 1996.
ACPI is an integral part of the Microsoft-inspired OnNow initiative, which was created to
minimize the delays inherent in starting up and shutting down a computer burdened with
megabytes of operating system overhead, to let the computer run tasks while it appears to
be off, and to lower the overall power requirement of the computer. At the time OnNow
was proposed, operating systems required time to test the host computer, to check out
Plug-and-Play devices, and to set up their structures. These functions required a lengthy
period to carry out, so booting your system took so long it seemed to be warming up from
absolute zero. OnNow was designed to eliminate that wait. The first inkling of OnNow
appeared in Windows Me, and Windows XP continues to develop the concept.
OnNow also sought to integrate the power and configuration interfaces of modern (meaning Windows) operating systems so that programmers can write to a common standard.
To bring these features to life, the OnNow design moves the operating system to the center of power management using ACPI and builds a new table structure for storing and
organizing configuration information.
As a power-management system, the ACPI specification can accommodate the needs of
any operating system, integrating all the necessary power-management features required
in a computer, from the application software down to the hardware level. It enables the
operating system to automatically turn on and off and adjust the power consumption of
nearly any peripheral, from hard disk drives to displays to printers. It can reach beyond
the computer to other devices that may be connected into a single system some time in
the futuretelevisions, stereos, VCRs, telephones, and even other appliances. Using the
Smart Battery specification, under ACPI the operating system takes command of battery
charging and monitoring. It also monitors the thermal operation of the system, reducing
speed or shutting down a computer that might overheat.
The ACPI standard itself defines the interface for controlling device power and a means
of identifying hardware features. The interface uses a set of five hardware registers that
are controlled through a higher-level application programming interface through the
operating system. The descriptive elements identify not only power management but also
device features through a nested set of tables. The ACPI standard supplements Plug-andPlay technology, extending its existing structure with an architecture-independent implementation, and replaces the Plug-and-Play BIOS with a new ACPI BIOS.
1045
1046
Soft Off
The fundamental and most noticeable change made by ACPI is the power button on the
front of new computers. In systems equipped to handle ACPI, this is a soft switch or set of
two switches. Although one of these switches may be labeled Power and imply that it is
an on/off switch, in the ACPI scheme of things the power switch does not actually switch
the power to the system on and off. Rather, it sends a command to the system to shut
itself offand not exactly what you think is off.
Using the front panel Off button actually puts the computer in a new mode called Soft
Off. In this mode, the computer acts like youve shut it off and requires rebooting to
restart it. But it doesnt remove all power from the system. A slight bit of power continues
to be supplied to the motherboard and expansion boards, enabling them to monitor external events. For example, a network board will still listen to network traffic for packets targeted at it. A modem or fax board may lie in wait for a telephone call. Or you may set a
time (such as midnight) at which the tape backup system starts. When any of these designated external events occurs, the computer automatically switches itself back on to deal
with it.
ACPI envisions that some manufacturers will also put a sleep switch (or standby button) on
the front panel or on the keyboard. Pressing it puts the computer in a sleep mode that
uses somewhat more power than Soft Off but allows the system to resume operation more
quickly.
States
As with APM, the ACPI design works by shifting modes called ACPI states. The states differ substantially from those in APM. Under ACPI, four basic types of states are defined
Global, Special Sleep, Microprocessor, and Deviceand these can be further subdivided.
Most importantly, ACPI lets the operating system control all aspects of the power consumption of a computer by shifting the single devices or the entire system between these
states.
The ACPI Global states most closely correspond to the APM modes, affecting the entire
operation of the computer and how you deal with it. Formally, the states are termed G0
through G3, but they are effectively the same as the APM states, with the addition of
Soft Off. The chief operational difference between them is how long it takes your computer to move from each state to normal operation, which can be instantaneous in
Sleeping state (corresponding to Suspend), a bit longer in Soft Off (similar but not identical to Hibernation), and a full boot-up from Power Off. Table 31.4 lists the Global
ACPI states.
CHAPTER 31 POWER
Power Management
TA B L E 3 1 . 4
ACPI States
State
Name
Appearance
Recovery Time
G0
Working state
Normal execution
Zero
G1
Sleeping state
No activity
Fast
G2
Soft Off
No activity
Slow
G3
Power Off
No activity
Boot-up
Under ACPI, each device, such as your hard disk, modem, or display screen, operates
under several similar states (but termed with the letter D) corresponding to various
power-saving modes. Similarly, the microprocessor (designed with state names starting
with C) can switch to various power-saving states. As a device is needed, your operating
system can switch it on or off using the ACPI protocols. Most of these changes happen
invisiblyyou may only detect a slight (or longer) lag as a device powers up from a
lower state.
Configuration
To handle its configuration function, ACPI must manage a tremendous amount of data,
not only about the power needs and management capabilities of the system but also
describing the features available for all the devices connected to the system. ACPI stores
this information in a hierarchy of tables.
The overall master table is called the Root System Description Table. It has no fixed place in
memory. Rather, upon booting up, the BIOS locates a pointer to the table during the
memory scan thats part of the boot-up process. The Root System Description Table itself
is identified in memory because it starts with the signature RSDT. Following the signature is an array of pointers that tells the operating system the location of other description
tables that provide it with the information it needs about the standards defined on the
current system and individual devices.
One of these tables is called the Fixed ACPI Description Table. In it the operating system
finds the base address of the registers used for controlling the power-management system.
In addition, the Fixed ACPI Description Table also points to the Differentiated System
Description Table, which provides variable information about the design of the base system. Some of the entries in this table are differentiated definition blocks, which can contain
data about a device or even a program that sets up other structures and defines new attributes. ACPI defines its own languages for programming these functions.
1047
1048
Energy Star
Rather than a technology, the Energy Star standard is a goal. The Energy Star standard
itself was created by the United States Environmental Protection Agency in 1992 with the
goal of encouraging manufacturers to create business equipment that minimizes power
consumption. Since then, the standard has been adopted by Japan, New Zealand, and
Sweden.
Energy Star is a certification program that allows compliant computers and peripherals (as
well as other devices unrelated to computers) to wear a certification badge. The program
specifically covers computers, monitors, printers, fax machines, copiers, scanners, and
multifunction devicesas well as telephones, home appliances (including washing
machines), and even entire buildings. Equipment conforming to the Energy Star standard
must meet strict guidelines on power consumption published by the Environmental
Protection Agency.
Manufacturers have a strong incentive to embrace the Energy Star standard and put the
label on their productssome businesses and many federal contracts require that new PC
equipment meet the Energy Star standards.
All that said, the actual Energy Star standards for PCs are quite simple. Energy Star version 2.0, which applies to products shipped after October 1, 1995, asks only that a PC or
monitor be able to switch to a low-power mode that consumes less than 30 watts after 15
to 30 minutes of inactivity (a default you are allowed to adjust). Combination monitorand-PC units are allowed the full 60 watts. Printers able to generate seven or fewer pages
per minute must reduce their drain to 15 watts after 15 minutes; 14 or fewer ppm, 30
watts after 30 minutes; faster or high-end color printers, 45 watts after an hour.
You can find out more about the Energy Star program at its Web site,
www.energystar.gov.
Power Protection
Normal line voltage is often far from the 115-volt alternating current you pay for. It can
be a rather inhospitable mixture of aberrations such as spikes and surges mixed with noise,
dips, and interruptions. None of these oddities is desirable, and some can be powerful
enough to cause errors in your data or damage to your computer. Although you cannot
avoid them, you can protect your computer against their ill effects.
Power Line Irregularities

Power line problems can be broadly classed into three basic categories: over-voltage,
under-voltage, and noise. Each problem has its own distinct causes and requires a particular kind of protection.
CHAPTER 31 POWER
Power Protection
Over-Voltage
The deadliest power-line pollution is over-voltagelightning-like high potential spikes
that sneak into your computer and actually melt down its silicon circuitry. Often the damage is invisibleexcept for the very visible lack of image on your monitor. Other times,
you can actually see charred remains inside your computer as a result of the over-voltage.
As its name implies, an over-voltage gushes more voltage into your computer than the
equipment can handle. In generaland in the long runyour utility supplies power thats
very close to the ideal, usually within about ten percent of its rated value. If it always
stayed within that range, the internal voltage regulation circuitry of your computer could
take its fluctuations in stride.
Short duration over-voltages larger than that may occur too quickly for your utilitys
equipment to compensate, however. Moreover, many over-voltages are generated nearby,
possibly within your home or office, and your utility has no control over them. Brief peaks
as high as 25,000 volts have been measured on normal lines, usually due to nearby lightning strikes. Lightning doesnt have to hit a power line to induce a voltage spike that can
damage your computer. When it does hit a wire, however, everything connected to that
circuit is likely to take on the characteristics of a flash bulb.
Over-voltages are usually divided into two classes by duration. Short-lived over-voltages
are called spikes or transients and last from a nanosecond (billionth of a second) to a
microsecond (one millionth of a second). Longer-duration over-voltages are usually
termed surges and can stretch into milliseconds.
Sometimes power companies do make errors and send too much voltage down the line,
causing your lights to glow brighter and your computer to teeter closer to disaster. The
occurrences are simply termed over-voltages.
Most AC-power computers are designed to withstand moderate over-voltages without
damage. Most machines tolerate brief surges in the range of 800 to 2,000 volts. On the
other hand, power cords and normal home and office electrical wiring break (by arcing
between the wiring conductors) at potentials between about 4,000 and 6,000 volts. In
other words, electrical wiring limits the maximum surge potential your computer is likely
to face to no more than about 6,000 volts. Higher voltage surges simply cant reach your
computer.
Besides intensity and energy, surges also differ in their mode. Modern electrical wiring
involves three conductors: a hot, neutral, and ground. Hot is the wire that carries the
power; neutral provides a return path; and ground provides protection. The ground lead is
ostensibly connected directly to the earth.
A surge can occur between any pairing of conductors: hot and neutral, hot and ground,
or neutral and ground. The first pairing is termed normal mode. It reflects a voltage
1049
1050
difference between the power conductors used by your computer. When a surge arises
from a voltage difference between hot or neutral and ground, it is called common mode.
Surges caused by utility switching and natural phenomenafor the most part, lightning
occur in the normal mode. They have to. The National Electrical Code requires that the
neutral lead and the ground lead be bonded together at the service entrance (where utility
power enters a building) as well as at the utility line transformer, typically hanging from a
telephone pole near your home or office. At that point, neutral and ground must have the
same potential. Any external common mode surge becomes normal mode.
Common mode surges can, however, originate within a building because long runs of wire
stretch between most outlets and the service entrance, and the resistance of the wire
allows the potential on the neutral wire to drift from that of ground. Although opinions
differ, recent European studies suggest that common mode surges are the most dangerous
to your equipment. (European wiring practice is more likely to result in common mode
surges because the bonding of neutral and ground is made only at the transformer.)
Under-Voltage
An under-voltage occurs when your equipment gets less voltage than it expects. Undervoltages can range from sags, which are dips of but a few volts, to complete outages or
blackouts. Durations vary from nearly instantaneous to hoursor even days, if you
havent paid your light bill recently.
Very short dips, sags, and even blackouts are not a problem. As long as they are less than a
few dozen millisecondsabout the blink of an eyeyour computer should purr along as
if nothing happened. The only exceptions are a few old computers that have power supplies with very sensitive Power Good signals. A short blackout may switch off the Power
Good signal, shutting down your computer even though enough electricity is available.
Most computers are designed to withstand prolonged voltage dips of about 20 percent
without shutting down. Deeper dips or blackouts lasting for more than those few milliseconds result in a shutdown. Your computer is forced to cold start, booting up afresh. Any
work you have not saved before the under-voltage is lost.
Noise
Noise is a nagging problem in the power supplies of most electronic devices. It comprises
all the spurious signals that wires pick up as they run through electromagnetic fields. In
many cases, these signals can sneak through the filtering circuitry of the power supply and
interfere with the signals inside the electrical device.
For example, the power cord of a tape recorder might act like an antenna and pick up a
strong radio signal. The broadcast could then sneak through the circuitry of the recorder
and mix with the music it is supposed to be playing. As a result, you might hear a CB
radio maven croaking over your Mozart.
CHAPTER 31 POWER
Power Protection
In computers, these spurious signals could confuse the digital thought coursing through
the circuitry of the machine. As a practical matter, they dont. All better computers are
designed to minimize the leakage of their signals from inside their cases into the outside
world to minimize your computers interfering with your radio and television. The same
protection that prevents signals leaking out works well in warding off other signals from
getting in. Personal computers are almost automatically well-protected against line noise.
You probably wont need a noise filter to protect your computer.
Then again, noise filtering doesnt hurt. Most power-protection devices have noise filtering built in to them because its cheap, and it can be an extra selling point (particularly to
people who believe they need it). Think of it as a bonus. You can take advantage of its
added protectionbut dont go out of your way to get it.
Over-Voltage Protection
Surges are dangerous to your computer because the energy they contain can rush through
semiconductor circuits faster than the circuits can dissipate themthe silicon junctions of
your computers integrated circuits fry in microseconds. Spike and surge protectors are
designed to prevent most short-duration, high-intensity over-voltages from reaching your
computer. They absorb excess voltages before they can travel down the power line and
into your computers power supply. Surge suppressors are typically connected between the
various conductors of the wiring leading to your computer. Most short out surges that rise
above a preset level.
The most important characteristics of over-voltage protection devices are how fast they
work and how much energy they can dissipate. Generally, a faster response time or clamping speed is better. Response times can be as short as picosecondstrillionths of a second.
You get better protection from devices that have higher energy-handling capacities, which
are measured in watt-seconds or joules. Devices claiming the capability to handle millions
of watts are not unusual.
Four kinds of devices are most often used to protect against surges: metal-oxide varistors
(MOVs), gas tubes, avalanche diodes, and reactive circuits. Of these, the MOVs dominate
because the parts are inexpensive and they work effectively against most surges.
The MOV is a disc-shaped electronic component typically made from a layer of zincoxide particles held between two electrodes. The granular zinc oxide offers a high resistance to the flow of electricity until the voltage reaches a break-over point. The electrical
current then forms a low-resistance path between the zinc-oxide particles that shorts out
the electrical flow. The energy-handling capability can be increased simply by enlarging
the device (typical MOVs are about an inch in diameter; high-power MOVs may be twice
that). Figure 31.1 shows a typical MOV.
1051
1052
FIGURE 31.1
A metal-oxide varistor.
GE-MDV
VR160N
The downside to MOVs is that they degrade. Surges tend to form preferred paths
between the zinc-oxide particles, reducing the resistance to electrical flow. Eventually, the
MOV shorts out, blowing a fuse or (more likely) overheating until it destroys itself. The
MOV can end its life in flames or with no external change at allexcept that it no longer
offers surge protection.
Thanks to the laws of thermodynamics, the excess energy in a surge cannot just disappear;
it can only change form. With most surge suppression technologies (all except reactive
devices), the over-voltage is converted into heat thats dissipated by the wiring between
the device and the origin of the surge as well as inside the surge suppressor itself. The
power in a large surge can destroy a surge suppressor so that it yields up its life to protect
your computer.
Because they degrade cumulatively with every surge they absorb, MOVs are particularly
prone to failure as they age. Eventually, an MOV will fail, sometimes in its own lightninglike burst. Although its unlikely this failure will electrically damage the circuits of your
computer, it can cause a firewhich can damage not just your computer, but your home,
office, or self.
An MOV-based surge suppressor also can fail more subtlyit just stops sucking up surges.
Unbeknownst to you, your computer can be left unprotected. Many commercial surge
suppressors have indicators designed to reveal the failure of an internal MOV.
In any case, a good strategy is to replace MOV-based surge suppressors periodically to
ensure that they do their job and to lessen the likelihood of their failure. How often to
replace them depends on how dirty an electrical diet you feed them, but most MOV-based
surge protectors should work effectively for a few years before needing replacement.
CHAPTER 31 POWER
Power Protection
Blackout Protection
Protecting against total blackouts requires a local source of electricity, either a generator
or storage batteries. A local generator is the choice when you want to continue to work as
normal, running from local power for hours or days. Battery backup lasts only as long as
the batteries, typically a few minutes to allow the orderly shutdown of your computer so
you can start back up when normal utility power returns.
Battery backup systems are often called uninterruptible power systems (UPSs) because they
supply power continuously, without interruption. UPSs are often used in conjunction with
generators to bridge over the few seconds that power would otherwise not be available
while the generator is starting. A battery backup system is built around powerful batteries
that store substantial current. An inverter converts the direct current from the batteries
into alternating current that can be used by your computer. A battery charger built in to
the system keeps the reserve power supply fully charged at all times. Long runtime UPSs,
which have extra batteries, sometimes substitute for generators, keeping computer servers
(particularly those in remote locations where they are not readily accessible) running for
hours.
Although the term UPS has become the industry standard for any kind of battery backup
system, there are actually two kinds of UPSs, only one of which provides truly uninterruptible power. An offline or standby power system switches the input of your computer from
utility power to backup power when the utility fails. An online power system keeps your
computer constantly connected to the backup power source, so it never has to switch.
Offline Backup Systems

As the name implies, the standby power system constantly stands by, waiting for the
power to fail so that it can leap into action. Under normal conditionsthat is, when utility power is availablethe battery is offline and its charger draws only a slight current to
keep its source of emergency energy topped off. The AC power line from which the
offline supply feeds is directly connected to its output, and thence to the computer. The
batteries are out of the loop.
When the power fails, the offline supply switches into actionswitch being the key word.
The current-carrying wires inside the power supply that lead to your computer are physically switched (usually by a mechanical relay) from the utility line to the current coming
from the battery-powered inverter.
Most offline power systems available today switch within one-half of one cycle of the AC
current they are suppliedthats less than ten milliseconds, quick enough to keep nearly all
computers running as if no interruption occurred. Although the standby power system
design does not protect again spikes and surges, most offline power systems have other protection devices installed in their circuitry to ensure that your computer gets clean power.
1053
1054
Line-interactive power systems are offline designs with an added feature. They react to
changes in line voltage and compensate when the voltage gets too high or too low. To
change the voltage, they use multitap autotransformers. When the voltage falls low, the
line-interactive UPS switches to a tap that boosts the voltage back to the appropriate level.
When the voltage gets too high, it switches to a different tap that bucks the voltage down,
reducing it. If the line voltage is too far from normal for the transformer taps to compensate, the UPS reacts as if it suffered a power failure, switching to battery power. The lineinteractive design offers the big benefit of compensating for most brownouts without
draining its battery reserves.
Online Backup Systems

Online backup systems use several designs to guarantee that there is never an interruption
in their output power. The traditional design is the double-conversion UPS. These devices
earn their name from converting the power twice. First, incoming utility power is transformed down to battery level and rectified to direct current. Then this direct current is
inverted and transformed back up to utility voltage to be supplied to your computer. The
process would seem to be wasteful and redundant except that a set of batteries connects in
the middle. When utility power is available, the constant supply at low voltage keeps the
batteries charged. When utility power fails, the batteries maintain the low voltage that
feeds the inverter, so the power to your computer is never interrupted.
This traditional double-conversion design is wasteful and expensive. It requires two large
transformers capable of carrying the entire load (your computers and its peripherals),
which are very costly and resistant to miniaturization technology. UPS-makers consequently have shifted to a slightly different design that eliminates the transformers. These
newer double-conversion UPSs simply rectify the incoming utility voltage to DC and
then send it to an inverter, which changes it back to AC. The batteries connect in the
middle, but instead of a direct link, they couple through a DC-to-DC converter that
matches their voltage to the power line. DC-to-DC converters are high-tech electronic
devices that cost substantially less than conventional transformers. As a result, this kind of
double-conversion UPS not only has replaced other online designs, but it can also be
cost-competitive with offline designs.
In effect, a double-conversion UPS acts like your computers own generating station, one
that is only inches away from the machine it serves. It keeps your system safe from the polluting effects of lightning and load transients found on long-distance power lines.
Moreover, dips and surges can never reach the computer. Instead, the computer gets a genuinely smooth, constant electrical supply, exactly like the one for which it was designed.
An alternate online design is more like an offline standby system but uses clever engineering to bridge over even the briefest switching lulls. These UPSs connect both the input
power and the output of their inverters together through a special transformer, which is
then connected to your computer or other equipment to be protected. Although utility
CHAPTER 31 POWER
Power Protection
power is available, this kind of UPS supplies it through the transformer to your computer.
When the utility power fails, the inverter kicks in, typically within half a cycle. The inductance of the transformer, however, acts as a storage system and supplies the missing halfcycle of electricity during the switchover period.
The double-conversion UPS provides an extreme measure of surge and spike protection
(as well as eliminating sags) because no direct connection bridges the power line and the
protected equipmentspikes and their kin have no pathway to sneak in. Although the
transformer in the new style of UPS absorbs many power-line irregularities, overall it
does not afford the same degree of protection. Consequently, these newer devices usually
have other protection devices (such as MOVs) built in.
Specifications
The most important specification to investigate before purchasing any backup power device
is its capacity as measured in volt-amperes (VA) or watts. This number should always be
greater than the rating of the equipment to which the backup device is to be connected.
In alternating current (AC) systems, watts do not necessarily equal the product of volts
and amperes (as they should by the definition that applies in DC systems) because the
voltage and current can be out of phase with one another. That is, when the voltage is at a
maximum, the current in the circuit can be at an intermediary value. So the peak values of
voltage and amperage may occur at different times.
Power requires both voltage and current simultaneously. Consequently, the product of
voltage and current (amperage) in an AC circuit is often higher than the actual power in
the circuit. The ratio between these two values is called the power factor of the system.
What all this means to you is that volt-amperes and watts are not the same thing. Most
backup power systems are rated in VA because it is a higher figure thanks to the power
factor. You must make sure the total VA used by your computer equipment is less than the
VA available from the backup power system. Alternatively, you must make sure that the
wattage used by your equipment is less than the wattage available from the backup power
system. Dont indiscriminately mix the VA and watts in making comparisons.
To convert a VA rating to a watt rating, multiply the VA by the power factor of the
backup power supply. To go the other waywatts to VAdivide the wattage rating of the
backup power system by its power factor. (You can do the same thing with the equipment
you want to plug into the power supply, but you may have a difficult time discovering the
power factor of each piece of equipment. For computers, a safe value to assume is 2/3.)
Both online and offline backup systems also are rated as to how long they can supply battery power. This equates to the total energy (the product of power and time) that they
store. Such time ratings vary with the VA the backup device must supplybecause of
finite battery reserves, it can supply greater currents only for shorter periods. Most
1055
1056
manufacturers rate their backup systems for a given minutes of operation with a load of a
particular size instead of in more scientific fashion using units of energy. For example, a
backup system may be rated to run a 250 volt-ampere load for 20 minutes.
If you want an idea of the maximum possible time a given backup supply can carry your
system, check the ratings of the batteries it uses. Most batteries are rated in ampere-hours,
which describes how much current they can deliver for how long. To convert that rating
to a genuine energy rating, multiply it by the nominal battery voltage. For example, a 12volt, 6 amp-hour battery could, in theory, produce 72 watt-hours of electricity. That figure is theoretical rather than realistic because the circuitry that converts the battery DC to
AC wastes some of the power and because ratings are only nominal for new batteries.
However, the numbers you derive give you a limit. If you have only 72 watt-hours of battery, you cant expect the system to run your 250 VA computer for an hour. At most, you
could expect 17 minutes; realistically, you might expect 12 to 15.
You probably will not need much time from a backup power system, however. In most
cases, five minutes or less of backup time is sufficient because the point of a backup supply
is not to keep a system running forever. Instead, the backup power system is designed to
give you a chance to shut down your computer without losing your work. Shutting down
shouldnt take more than a minute or two.
UPS-makers warn that no matter the rating of your UPS, you should never plug a laser
printer into it. The fusers in laser printers are about as power hungry as toastersboth
are resistive heaters. The peak power demand when the fuser switches on can overload
even larger UPSs, and the continuing need for current can quickly drain batteries.
Moreover, theres no need to keep a print job running during a power failure. Even if you
lose a page, you can reprint it when the power comes back at far less expense than the cost
of additional UPS power capable of handling the lasers needs. Some printers, such as
inkjets, are friendlier to UPSs and can safely be connected, but youll still be wasting
capacity. The best strategy is to connect only your computer, your monitor, and any external disk drives to the UPS. Plug the rest of your equipment into a surge suppresser.
To handle such situations, many UPSs have both battery-protected outlets and outlets
with only surge protection. Be sure to check which outlets you use with your equipment,
making sure your computer has battery-backed protection.
Interfaces
An ordinary UPS works effectively if youre sitting at your computer and the power fails.
You can quickly assess whether it looks like the blackout will be short or long (if the world
is blowing away outside your window, you can be pretty sure any outage will be prolonged). You can save your work, haul down your operating system, and shut off your
computer at your leisure. When a computer is connected to a network or is running unattended, however, problems can arise.
CHAPTER 31 POWER
Power Protection
During a prolonged outage, a simple UPS only prolongs a disaster with an unattended
computerit runs another dozen minutes or so while the power is off, then the UPS runs
out of juice and the computer plummets with it. Of course, if a server crashes without
warning, no one is happy, particularly if a number of files were in the queue to be saved.
To avoid these problems, better UPSs include interfaces that let them link to your computer, usually through a serial port. Install an appropriate driver, supplied by the UPSmaker, and your computer can monitor the condition of your power line. When the power
goes off, the software can send messages down the network warning individual users to save
their work. Then, the UPS software can initiate an orderly shut down of the network.
Some UPSs will continue to run even after your network or computer has shut itself
down. Better units have an additional feature termed inverter shutdown that automatically
switches off the UPS after your network shuts down. This preserves some charge in the
batteries of the UPS so that it can still offer protection if you put your computer or network back online and another power failure follows shortly thereafter. A fully discharged
UPS, on the other hand, might not be ready to take the load for several hours.
Battery Life
The gelled electrolyte batteries most commonly used in uninterruptible power systems
have a finite life. The materials from which they are made gradually deteriorate, and the
overall system loses its ability to store electricity. After several years, a gelled electrolyte
battery will no longer be able to operate a UPS, even for a short period. The UPS then
becomes nonfunctional. The only way to revive the UPS is to replace the batteries.
Battery failure in a UPS usually comes as a big surprise. The power goes off and your computer goes with it, notwithstanding your investment in the UPS. The characteristics of the
batteries themselves almost guarantee this surprise. Gelled electrolyte batteries gradually
lose their storage capacity over a period of years, typically between three and five. Then,
suddenly, their capacity plummets. They can lose nearly all their total storage capability in a
few weeks. Figure 31.2 illustrates this characteristic of typical gelled electrolyte batteries.
UPS battery capacity

slowly deteriorates,
then suddenly
plummets.
Battery Capacity
FIGURE 31.2
Battery age
1057
1058
Note that the deterioration of gelled electrolyte batteries occurs regardless of whether
they are repeatedly discharged. They deteriorate even when not used, although repeated
heavy discharges will further shorten their lives.
To guard against the surprise of total battery failure, better UPSs incorporate an automatic testing mechanism that periodically checks battery capacity. A battery failure indication from such a UPS should not be taken lightly.
Phone and Network Line Protection

Spikes and surges affect more than just power lines. Any wire that ventures outside holds
the potential for attracting lightning. Any long wiring run is susceptible to induced voltages, including noise and surges. These over-voltages can be transmitted directly into your
computer or its peripherals and cause the same damage as a power line surge.
The good news is that several important wiring systems incorporate their own power protection. For example, Ethernet systems (both coaxial and twisted pair) have sufficient
surge protection for their intended applications. Apple LocalTalk adapters are designed to
withstand surges of 2000 volts with no damage. Because they are not electrical at all, fiberoptical connections are completely immune to power surges.
The bad new is that two common kinds of computer wiring are not innately protected by
surges. Telephone wiring runs long distances through the same environments as the
power-distribution system and is consequently susceptible to the same problems. In particular, powerful surges generated by direct lightning hits or induction can travel through
telephone wiring, through your modem, and into the circuitry of your computer. In addition, ordinary serial port circuitry includes no innate surge suppression. A long unshielded
serial cable can pick up surges from other cables by induction.
The best protection is avoidance. Keep unshielded serial cable runs short whenever possible. If you must use a long serial connection, use shielded cable. Better still, break up
the run with a short-haul modem, which will also increase the potential speed of the
connection.
Modem connections with the outside world are unavoidable in these days of online connectivity and the Internet. You can, however, protect against phone-line surges using special suppressors designed exactly for that purpose. Better power-protection devices also
have modem connections that provide the necessary safeguards. Standalone telephone
surge suppressors are also available. They use the same technologies as power-line surge
suppressors. Indeed, the voltage that rings your telephone is nearly the same as the
110120 volt utility power used in the United States. Most phone-line suppressors are
based on MOV devices. Better units combine MOVs with capacitors, inductors, and fuses.
WINN L. ROSCH
HARDWARE BIBLE,
SIXTH EDITION
CHAPTER
32
Cases
The case is your computer. Its what you kick when it wont behave. Its
the big, bulky box that makes too much noise and takes up too much
spaceexcept if you have a notebook computer. Its the thing that gets in
the way, the thing you pay too much for. Its all the things that you hate
about your computer.
The case is the most visible part of your computer. Its the most physical.
But dont be misled. The physical side of your computers case is almost
trivial. Its almost all style. Although the substance of it is vital for holding your computer together, thats all blacksmith technology. The case is
no mere shell. It is a sophisticated piece of engineering. In addition to its
mechanical function, it also serves to keep your computer cool, thereby
extending the life of its circuits. In addition, the case is a protective
shield that looks after you as well as the delicate circuits inside. It even
protects the world from the high-frequency (and low) signals inside.
Mechanical Construction
The obvious function of the case is mechanicalyou can see it and
touch it as a distinct object. And it steals part of your desktop, floor, or
lap when you put it to work. It has a definite sizealways too small
when you want to add one more thing but too large when you need to
find a place to put it (and particularly when that place happens to be
inside your carry-on luggage). The case also has a shape, which may be
functional to allow you the best access to all those computer accoutrements, such as the slot into which you shove your backup tapes. But
shape and color also are part of your computers style, which can set one
system apart from the boring sameness of its computer kin.
1062
Modern computers have wings, waves, and flaresall only aesthetics calculated to make a
given manufacturers machines stand out among the masses, style to make you think they
are more modern and capable. The features, such as some of the more interesting paint
shades with which some manufacturers have started experimenting, are design gimmicks,
the tail-fins of the new century. Theres nothing wrong with a computer that looks like
youve stolen it from the deck of an aircraft carrier, but theres nothing inherently better
about it, either. Beneath the plastic youll find that same basic mechanical design and construction of systems built a decade ago.
When a computer is being built, two issues are paramount: how the case is put together
and how a computer is put together inside it. In the first case, how the case gets constructedwhether with screws, rivets, or weldsis not so important as the size of the finished product. The size of the case determines what it can hold, which in turn limits how
much you can expand the system to add all the features you want. How components
install inside the case affect how much you want to expand your system. If sliding in an
expansion board or just getting at a drive bay to install a new disk makes you think the
system was designed by someone in league with the devil (or another employee of
Microsoft), youre not going to look on expansion with much favor.
Some computer companies actually sympathize with your plight, knowing the happy customer brings goodwill, and a satisfied customer is less likely to make calls to the support
staff thats so expensive to pay. Over the years, they have developed several schemes to
make working on and expanding your computer easiereverything from cases you can
open without tools to cases you cannot open at all, relying instead on external expansion.
In computers, form dictates function as much (if not more) than it does for any other type
of office equipment. Computers have the shape they have so that they can hold what you
want to put inside themprimarily all those expansion options that give your machine
power and capabilities. It has to be large enough to accommodate the expansion boards
you want to plug in as well as provide adequate space for all the disk drives your computer
and life would not be complete withouthard, floppy, DVD, and the like.
For most manufacturers, the sizes of both boards and drives are preordained. Both are
covered by industry standards that were set years ago and remain invariant. The case must
be designed around these standards if it is to accommodate the generally available motherboards, expansion boards, and disk drives.
Standards
All current desktop computer cases follow the motherboard mounting dimensions laid
down by the ATX motherboard standard, the microATX standard, or the NLX standard.
These standards describe not only the minimum dimensions for a computer case but also
the locations of mounting standards for the screws that hold the motherboard in place.
CHAPTER 32 CASES
The guidance given by these standards is, in fact, minimal. The standards define only the
footprint of the motherboard and a space on the rear panel of the case for port connectors. Even so, you can be sure that an ATX motherboard will fit into an ATX case.
With notebook computers, all bets are off. Despite attempts at defining a notebook
motherboard standard, every manufacturer goes its own wayboth in motherboard and
case design.
Size
A computer can be any size to carry out its job. Thanks to miniaturization, even the
smallest computers can rival the power of the largestwith one exception. You can stuff
more stuff into a big box than you can a small one.
Thats the essence of size differences among computers. Bigger means more expansion,
more places to put power-enhancing accessories. As long as you dont need an extraordinary amount of expansion, even the physically smallest computer will suffice.
A number of terms are used to describe the size of computers, although none approaches
being an official designation.
Compact
The basic computerthat is, the smallest with the least future potentialuses a compact
device thats sometimes called a small footprint computer. In the jargon of the computer
industry, the footprint is the space a computer occupies on your desk, so the small footprint steals less desk space, making it an ideal office companion. Compact desktop computers are sometimes called low-profile computers because they are short, usually as short as
possible, and still hold a whole computer.
To achieve compact size, computer-makers must do some serious trimming from the
designs of larger computers. The principal loss is in expansion. Compact systems have
fewer slots and bays than other computers. They have a minimum of extra space for
expansiononce filled with standard factory equipment, perhaps as little as a single open
slot and a single unfilled bay. The underlying philosophy is that the compact computer
comes with everything you need already installed, so theres no need to add anything
more.
To shave down the height of low-profile cases while conserving the capability of handling
tall expansion boards, compact machines move their options on edge. Instead of boards
sliding into slots vertically, these tiny computers align them horizontally. The motherboard boards remain horizontal in the case and have a single master expansion slot. A
special expansion board rises vertically with additional slot connectors on one side to
allow you to slide in several ordinary expansion boards (typically three) parallel to the
system board.
1063
1064
Although the exact assortment of drive possibilities varies with the design of compact systems, most small-footprint machines offer similar drive options. They are constrained by
what you expect and need in any computer. A floppy disk drive and a CD or DVD drive
leer out from the front panel, and a hard disk hides safely inside. If any drive expansion is
available inside these systems, it comprises a single bay suitable for an inch-tall hard disk.
Desktop
At one time, desktop meant a particular case designa low, flat cabinet meant to sit on a
desk and hold a monitor on top. The basic configuration and even the standard dimensions carried over from the original IBM Personal Computer. The result was a case measuring 21 inches wide by 17 inches deep, designed to accommodate two large disk drives
measuring 5.25 inches across, side by side. Its height, to allow for expansion boards and
three-eighths inch feet underneath for airflow (just enough so you could lose a pencil
maybe the designers did have some kind of inspiration!), measured 5.5 inches.
Because few people need more than one large (5.25-inch) drive bay, most new desktop
computers opt to shrink the other bay area to the 3.5-inch size, shaving about two inches
of the computer in the process. This intermediate package, measuring about 5.5 by 19 by
17 inches is todays most popular in desktop systems.
Towers
Stand a conventional computer on edge and you get a tower. At least thats how the tower
format evolved. The design has proved so popular that these days, one form of tower or
another probably is the standard computer.
The upright design actually makes the ultimate small-footprint computer. It steals no
space from your desk if you set it next to the furniture. A smaller footprint is hard to
imagine.
Freed from restraints required when sharing a desktop, tower computers could expand to
suit even the most fanciful dreams. The largest have nearly as many bays as the Eastern
seaboardas many as 11 in a single system. On the other hand, they usually do not
extend the number of expansion slots over eight. The reason is not a spatial limitation but
an economic one. Most towers use standard, off-the-shelf motherboards. Raising the slot
number would involve expensive redesign (as well as more motherboard, circuitry, and
assembly costs).
With modern components, vertically mounting computer components causes no problems. Electronic circuits dont know which way is up. Although mechanical devices such as
disk drives may care, the only practical effects with current hardware are really your problem. Putting a CD drive on edge can turn loading a disc into a dash across your office
floor and down the hallway as you race after the errant disc that dropped and rolled away.
Most towers keep their bays horizontal to avoid the office chase scene. Compaq even
CHAPTER 32 CASES
developed a computer transformerby remounting the drive bay and changing the front
panel bezel, you can switch the Deskpro EP series from desktop to tower and back.
The tower design has been so successful, designers have fine-tuned the vertical package to
meet a variety of expansion needs. Mini-tower cases are the most compact and usually accommodate only Mini-AT and smaller motherboards and a few drive options. Full-size tower cases
hold full-size motherboards and more drive bays than most people know what to do with.
Recently, midi-tower cases, with accommodation falling in-between, have become a popular
option. There is no standardization of these terms. One manufacturers mini is anothers midi.
Choose a computer with a tower-style case for its greater physical capacity for internal
peripherals and its flexibility of installation wherever there are a few vacant feet of floor
space. You also need to be critical about the provisions for physically mounting mass-storage devices. Some towers provide only flimsy mounting means or require you to work
through a Chinese puzzle of interlocking parts to install a drive. You need a system that
provides sufficient drive-mounting options.
Notebook
Back in the days before micro-miniaturization, anything instantly became portable the
moment you attached a handle. The first generation of portable televisions, for example,
was eminently portableat least for anyone accustomed to carrying a carboy of mineral
water under each arm. The first generation of computers had similar pretenses of portability, challenging your wherewithal with a weighty bottle of gas and photons and a small
but hardly lightweight picture tube. The typical weight of a first-generation portable computer was about 40 poundsabout the limit of what the market (or any reasonable human
being) would bear. These portables were essentially nothing more than a repackaging that
combined a conventional computer with an integral monitor.
Replacing the bottle with a flat-panel display gave designers a quick way to cut half the
weight and repackage systems into lunchbox computers. The name referred to the slabsided design with a handle on top, reminiscent of what every kid not party to the school
lunch program toted to class. But with some weighing in at 20 to 25 pounds, these packages were enough to provide Paul Bunyan with his midday meal. The largest of these did
allow the use of conventional motherboards with space for several conventional expansion
slots. Overall, however, the design was one that only a mother could love, at least if she
advocated an aggressive weight-training program.
The ultimate in computer compression is the notebook computer, machines shrunk as
small as possible while allowing your hands a grip on their keyboards (and eyes a good
look at the screen) and as thin as componentry allows. Manufacturers have managed to
squeeze the typical notebook computer down to a little more than six pounds. Size is
limited by the choice of display screen. With bleary-eyed users demanding the biggest
possible screensone manufacturer now offers a notebook computer with a 16-inch
1065
1066
screencases have been tailored to match. Although easy on your eyes when youre working or playing games, the big screens make todays notebook a bit more troublesome to
carry and almost impossible to use in an economy-class airline seat.
The notebook design spawned the subnotebook for people on the move who want to
carry the minimal computer. Typically, subnotebooks forgo several features, such as port
connectors and optical drives, to achieve their smaller size. In general, a subnotebook is a
portable computer weighing under four pounds, has a 12-inch or smaller LCD screen,
and is a little larger than an inch-thick stack of writing tablets.
Fitting into the small notebook and subnotebook packages forces manufacturers to sacrifice some degree of convenience. For example, a number of subnotebook machines have
been developed with keyboards reduced to 80 percent of the standard size. Most people
adapt to slightly cramped keyboards and continue to touch-type without difficulty. Smaller
than that, however, and touch-typing becomes challenging. In other words, handheld
computers are not for extensive data entry.
Besides length and width, notebook computermakers also have trimmed the depth of
their keyboards, reducing the height of keytopsnot a noticeable changeas well as
reducing key travel. The latter can have a dramatic effect on typing feel and usability.
Although the feel and travel of a keyboard are mostly user preference, odds favor greater
dissatisfaction with the shrunken, truncated keyboards in miniaturized computers compared to full-size machines.
Notebook computers have forced manufacturers to miniaturize a number of components that
dont naturally lend themselves to miniaturization. Both CD and DVD-ROM drives now fit
into a space one-third their desktop size, but they suffer the penalty of lower top speeds.
Hard disk drives suffer the same fate. Drive-makers have shrunk their products to fit a
2.5-inch form factor with little capacity penalty. Instead, they have sacrificed speed.
Notebook drives spin more slowly than desktop models (which helps them gain some of
the capacity they need).
Internal Access
Two philosophies surround the issue of access into your computers case. One school of
thought holds that it should be as easy as possible so that you can make upgrades quickly
and easily. The other school believes that you can only get yourself into trouble inside
your computer. Students of this school believe that the best case is one thats sealed
against access. If they had their way, your computer would be embedded in a block of
Lucite so you could see its pretty lights but not touch a thing inside.
The curriculum divides roughly on application lines. Computers used at home or in
a small businessthat is, one small enough that youre responsible for your own
CHAPTER 32 CASES
computerare best with open access. Large businesses that have a unified Information
Systems department headed by an all-powerful fuehrer want to keep in absolute control
which means keeping you out. The needs of these folks led the move to the Network
Computer and other abortive attempts at electronic mind control.
Tool-free entry allows you to open and upgrade your system without special tools or any
tools at all. You wont have to steal a knife from the silverware drawer to get into your
computer or buy a special screwdriver to pull out an expansion board. In the ideal case,
the lid unlatches and slides off your computer with minimum effort. Disk drives snap into
place, and knurled posts that you can tighten with your fingers lock expansion boards into
place. Some current systems approach this ideal, but few go all the way.
Bays
Bays come in two basic varieties: those with front panel access and those without. Frontpanel-access bays are for devices using removable media. You have to have access to the
front of the drive to slide in a disk or tape cartridge. Floppy disk drives, CD and DVD
drives, and tape drives all must be mounted in bays with front panel access.
Internal bays lack front panel access. They may be tucked away inside the computer or just
below the bays with front panel access. Internal bays suit devices that require no direct
user interaction, those you dont have to see or touch. The chief occupant of an internal
bay is the hard disk drive.
Form Factors
Disk drives come in a variety of heights and widths. The basic unit of measurement of the
size of a drive is the form factor, which is simply the volume of a standard drive that handles a particular medium. Several form factors regularly find their way into discussions of
personal computers, ranging in size from 8 inches to 1.3 inches, most of which allow for
one or more device heights.
The full-size drive, the unit that defines the form factor and occupies all of its volume, is
usually a first-generation machine. Its exact dimensions, chosen for whatever particular
reason, seemed fitting, perhaps allowing for the mood of the mechanical engineer on the
day he was drafting the blueprints. If the drive is a reasonable size and proves particularly
successfulsuccessful enough that other manufacturers eagerly want to cash in, too
others follow suit and copy the dimension of the product, making it a standard.
Device Heights
The second generation of any variety of hardware inevitably results in some sort of size
reduction. Cost cutting, greater precision, experience in manufacturing, and the inevitable
need to put more in less space gang up to shrink things down.
1067
1068
The succession of heights first appeared among 5.25-inch drives. At 5.25 inches, devices
are measured in increments of the original full-height package. Devices that are twothirds height, half-height, one-third, or quarter-height have all been manufactured at one
time or another.
At the 3.5-inch form factor, sizes are more pragmatic, measured as the actual height in
inches. The original 3.5-inch drives may be considered full height and typically measure
about 1.6 inches high. The next most widely used size was an even inch in height (defining five-eighths height, for the fractious folk who prefer fractions). Sub-inch heights have
been used for some devices, some as small as 0.6 inches.
At 2.5 inches, the size of most notebook computer drives, height is described as actually
measured. Typical high-capacity drives measure 12.5 millimeters high (roughly half an
inch) with 9.5 millimeters preferred for the drives in subnotebook computers.
Smaller drives have also been manufactured, although the one standard for tiny drives fits
the CompactFlash package. These have superceded the 1.8-inch size that won particular
favor for fitting into Type 3 PC Cards that follow the most recent standards promulgated
by PCMCIA.
Racks
Most drive-mounting schemes are meant for permanent installations. Your hard disk drive
is supposed to last for the life of your computer. In certain situations, however, you may
want something less permanent.
If you deal in secrets or have valuable information that you store on your computer, you
might not want to leave it out on your desk untended all night long. Although packing up
and putting your computer away for the evening is more trouble than most people want to
bother with, pulling out a disk drive and sliding it into a safe is not.
Although hard disks are quite reliable, every one you add to your system increases the
chances that one will fail. Multiple disk drives and the need for reliable operation go
hand-in-hand in RAID systems. To speed repairs, most RAID systems have moved to
mounting drives for hot-swapping.
In either case, the choice is mounting your drive in a removable rack. The typical hotswap rack allows you to install a 3.5-inch hard disk in a 5.25-inch bay. More than an
adapter, it puts the drive on a submodule that slides into the rack, connects, and locks in
place. You can remove a drive for the evening or swap a new drive for an old one at any
time. Each rack manufacturer uses its own designs, so theres no intercompatibility among
different rack systems.
CHAPTER 32 CASES
Cooling
Slots
All expansion slots are not the same. Put another way, a slot is not a slot is not a slot.
Expansion slots differ in two primary waysthe standard followed and length.
To push up slot totals, many manufacturers also add that their systems (or motherboards)
include an accelerated graphics port (AGP) slot. Technically they are correct. The AGP
slot works exactly like an expansion slot. But it is irrelevant when analyzing the availability
of expansion. Any computer with an AGP slot likely fills it with an AGP graphics adapter
(why else would the slot be there?), so the AGP slot is never available for expansion
(although it does allow you to upgrade your video system). Ordinary expansion boards
both PCI and ISAwill not fit into the AGP slot in a computer.
Slots also vary in length. A full-size slot allows you to install any length of expansion
board. A short slot cannot hold a full-length expansion board. Although PCI describes the
dimensions of a short card, computer-makers make short slots of any length that suits
their designs. In general, these short slots are more than half the length of a full-size slot
and will accommodate most short cards. Most modern expansion boards are short, so you
ordinarily wont have problems matching slot length.
Cooling
A case can be confining. It can keep just about everything from escaping, including the
heat that electronic circuits produce as a by-product of performing their normal functions.
Some of the electricity in any circuit (except one made from superconductors) is turned
into heat by the unavoidable electrical resistance of the circuit. Heat is also generated
whenever an element of a computer circuit changes state. In fact, nearly all the electricity
consumed by a computer eventually turns into heat.
Inside the protective (and confining) case of the computer, that heat builds up, thus driving
up the temperature. Heat is the worst enemy of semiconductor circuits; it can shorten their
lives considerably or even cause their catastrophic failure. Some means of escape must be
provided for the excess heat. In truth, the heat build-up in most computers may not be
immediately fatal to semiconductor circuits. For example, most microprocessors shut down
(or simply generate errors that shut down your computer) before any permanent damage
occurs to them or the rest of the components inside your computer. However, heat can
cause circuits to age prematurely and can trim the lives of circuit components.
The design of the case of a computer affects how well the machine deals with its heat
build-up. A case thats effective in keeping its internal electronics cool can prolong the life
of the system.
1069
1070
Passive Convection
The obvious way to make a computer run cooler is to punch holes in its case to let the
heat outbut to keep the holes small enough so that other things such as mice and
milkshakes cant get in. In due time, passive convectionless dense hot air rising with
denser cool air flowing in to take its placelets the excess thermal energy drift out of
the case.
Any impediment to the free flow of air slows the passive cooling effect. In general, the
more holes in the case, the merrier the computer will be. Remove the lid, and the heat can
waft away along with temperature worries.
Unfortunately, your computers case should be closed. Keeping a lid on it does more than
just restrict coolingit is also the only effective way to deal with interference. It also
keeps your computer quieter, prevents foreign objects and liquids from plummeting in,
and gives your monitor a lift.
Moreover, passive cooling is often not enough. Only low-power designs (such as notebook
and Green computers) generate little enough heat that convection can be entirely successful. Other systems generate more heat than naturally goes away on its own.
Active Cooling
The alternative to passive cooling is, hardly unexpectedly, active cooling, which uses a
force of some kind to move the heat away from the circuits. The force of choice in most
computers is a fan.
Usually tucked inside the power supply, the computers fan forces air to circulate both
inside the power supply and the computer. It sucks cool air in to circulate and blows the
heated air out.
The cooling systems of early computers, however, were particularly ill-conceived for
active cooling. The fans were designed mostly to cool off the heat-generating circuitry
inside the power supply itself and only incidentally cooled the inside of the computer.
Moreover, the chance design of the system resulted in most of the cool air getting sucked
in through the floppy disk drive slots. Along with the air came all the dust and grime
floating around in the environment, polluting whatever media you had sitting in the drive.
At least enough air coursed through the machine to cool off the small amount of circuitry
that the meager power supply of the computer could provide.
Modern computers do much better. Many have carefully designed air channels to route
cooling air over the places most apt to get work (such as memory chips). Some manufacturers opt to put one or more additional fans (besides the one in the power supply) into
their systems to keep the overall chassis cool.
CHAPTER 32 CASES
Cooling
Modern microprocessors generate so much waste heat that a single system fan cant keep
them cool. They require an additional means to keep air circulating. Most microprocessors
in desktop computers now use fans integrated into their heatsinks to keep the chips cool.
Each heatsink design requires a fan custom tailored to it.
Large cooling fans make noise. Some computer manufacturers try to minimize this noise
by making their fans thermostatically controlled. They switch on only when your computer needs cooling. Sometimes this feature is controlled through the BIOS; other times
it is an invariant feature. If you suspect fan failure (discussed later in this chapter in more
detail), be sure your system does not have a thermostatically controlled fan that simply has
not switched on.
Many people believe you cannot overdo coolingmore fans is always better. In fact, the
cooler you keep your chips, the longer they will last. Heat is their worst enemy. The only
downside to blowing hurricanes through your computer is that air is inevitably fouled
with dust and lint. A substantial amount of foreign material may accumulate inside your
computer, and it may alter or stop airflow in some places in the system, causing them to
overheat. To prevent such problems, fans may be equipped with filters. In any case, whenever a computer is operated in an environment likely to cause dust and lint contamination
(say, for instance, you have a cat), you should vacuum the inside of your computer periodically to remove any accumulation.
Fan Failure
The fan inside a computer power supply is a necessity, not a luxury. If it fails to operate, your computer wont falterat least not at first. But temperatures build up inside.
The machinethe power supply in particularmay even fail catastrophically from
overheating.
The symptoms of fan failure are subtle but hard to miss. You hear the difference in the
noise your system makes. You may even be able to smell components warming past their
safe operating temperature.
Should you detect either symptom, hold you hand near where the air usually emerges
from your computer. (On most computers, thats near the big round opening that the
fan peers through.) If you feel no breeze, you can be certain your fan is no longer doing
its job.
A fan failure constitutes an emergency. If it happens to your system, immediately save
your work and shut the machine off. Although you can safely use it for short periods,
the better strategy is to replace the fan or power supply as soon as you possibly can.
1071
1072
Radiation
Besides heat, all electrical circuits radiate something elseelectromagnetic fields. Every flow
of electrical energy sets up an electromagnetic field that radiates away. Radio and television
stations push kilowatts of energy through their antennae so that this energy (accompanied
by programming in the form of modulation) radiates over the countryside, eventually to be
hauled in by a radio or television set for your enjoyment or disgruntlement.
The electrical circuits inside all computers work the same way but on a smaller scale. The
circuit board traces act as antennae and radiate electromagnetic energy whenever the
computer is turned on. When the thinking gets intense, so does the radiation.
You cant see, hear, feel, taste, or smell this radiation, just as you cant detect the emissions
from a radio station (at least not without a radio), so you would think there would be no
reason for concern about the radiation from your computer. But even invisible signals can
be dangerous, and their very invisibility makes them more worrisomeyou may never
know if they are there or not. The case of your computer is your primary (often only) line
of defense against radiation from its electronic circuitry.
The problems of radiation are twofold: the radiation interfering with other, more desirable signals in the air, and the radiation affecting your health.
Radio Frequency Interference

The signals radiated by a computer typically fall in the microwatt range, perhaps a billion
times weaker than those emitted by a broadcasting station. You would think that the
broadcast signals would easily overwhelm the inadvertent emissions from your computer.
But the strength of signals falls off dramatically with distance from the source. They follow the inverse-square law; therefore, a signal from a source a thousand times farther
away would be a million times weaker. Radio and television stations are typically miles
away, so the emissions from a computer can easily overwhelm nearby broadcast signals,
turning transmissions into gibberish.
The radiation from the computer circuitry occurs at a wide variety of frequencies, including not only the range occupied by your favorite radio and television stations but also aviation navigation systems, emergency radio services, and even the eavesdropping
equipment some initialed government agency may have buried in your walls. Unchecked,
these untamed radiations from within your computer can compete with broadcast signals,
not only for the ears of your radio but those of your neighbors. These radio-like signals
emitted by the computer generate what is termed radio frequency interference (RFI), so
called because they interfere with other signals in the radio spectrum.
The government agency charged with the chore of managing interferencethe Federal
Communications Commissionhas set strict standards on the radio waves that personal
CHAPTER 32 CASES
Radiation
computers can emit. These standards are complex, and ensuring your computer is actually
in compliance would require expensive test equipment. The law does not require that you
checkthe burden is on the manufacturer. Moreover, at their hearts, the FCC standards
simply enforce a good-neighbor policy. They require that the RFI from computers be so
weak that it wont bother your neighbors, although it may garble radio signals in your
own home or office.
The FCC sets two standards: Class A and Class B. Computer equipment must be verified
to meet the FCC Class A standard to be legally sold for business use. Computers must be
certified to conform with the more stringent FCC Class B standard to be sold for home
use.
Equipment-makers, rather than users, must pass FCC muster. You are responsible, however, for ensuring that your equipment does not interfere with your neighbors. If your
computer does interfere, legally you have the responsibility for eliminating the problem.
While you can sneak Class A equipment into your home, you have good reasons not to.
The job of interference elimination is easier with Class B certified equipment because it
starts off radiating lower signal levels, so Class B machines give you a head start.
Moreover, meeting the Class B standards requires better overall construction, which helps
ensure that you get a better case and a better computer.
Minimizing Interference
Most television interference takes one of two forms: noise and signal interference.
Noise interference appears on the screen as lines and dots that jump randomly about. The
random appearance of noise reflects its origins. Noise arises from random pulses of electrical energy. The most common source for noise is electric motors. Every spark in the
brushes of an electric motor radiates a broad spectrum of radio frequency signals that your
television may receive along with its normal signals. Some computer peripherals may also
generate such noise.
Signal interference usually appears as a pattern of some sort on your screen. For example, a
series of tilted horizontal bars or noise-like snow on the screen that stays in a fixed pattern
instead of jumping madly about. Signal interference is caused by regular, periodic electrical signals.
Television interference most commonly occurs when you rely on a rabbit ear antenna
for your television reception. Such antennae pull signals from the air in the immediate
vicinity of the television set, so if your computer is nearby, its signals are more likely to be
received. Moving to cable or an external antenna relocates the point your TV picks up its
signals to a distant location and will likely minimize or eliminate interference from a computer near the TV set.
1073
1074
You can minimize the interference your computer radiates and improve your television
reception by taking several preventive measures.
The first step is to make sure the lid is on your computers case and that it and all expansion boards are firmly screwed into place. Fill all empty expansion slots with blank panels.
Firmly affixing the screws is important because they ground the expansion boards or blank
panels, which helps them shield your computer. This strategy also helps minimize the
already small fire hazard your computer presents.
If the interference persists after youve screwed everything down in your computer, next
check to see if you can locate where the interference leaks out of your computer. The
most likely suspects are the various cables that trail out of your computer and link to
peripherals such as your monitor, keyboard, and printer. Disconnect cabled peripherals
one at a time and observe whether the disconnection reduces the interference.
Because it operates at the highest speed (and therefore the highest frequency), an external
SCSI cable is most prone to radiating interference. All external SCSI cables should be
shielded.
Your mouse is the most unlikely part of your computer to cause TV interference. The
mouse operates at serial data rates, which are much too low to interfere even with VHF
television.
If disconnecting a cable reduces onscreen TV interference, the next step is to get the
offending signal out of the cable. The best way is to add a ferrite core around the cable.
Many computer cables already have ferrite cores installed. They are the cylindrical lumps
in the cable near one or the other connector. Install the ferrite core by putting it around
the offending cable near where the cable leaves your computer. You can buy clamp-on ferrite cores from many electronic-parts stores.
Unplugging one cableyour computers power cableshould completely eliminate the
interference radiated by your computer. After all, the computer wont work without power
and cant generate or radiate anything. You can reduce the interference traveling on the
power line by adding a noise filter between your computers plug and its power outlet. You
can usually obtain noise filters from electronic-parts suppliers. Although a noise filter is
not the same thing as a surge suppresser, most of the better surge suppressers also include
noise filtering.
Health Concerns
Some radiation emitted by computers is of such low frequencies that it falls below the
range used by any radio station. These very low frequency and extremely low frequency signals (often called VLF and ELF) are thought by some people to cause a variety of health
problems.
CHAPTER 32 CASES
Radiation
Your computers case is the first line of defense against these signals. A metal case blocks
low-frequency magnetic fields, which some epidemiological studies have hinted might be
dangerous, and shields against the emission of electrical fields. Plastic cases are less effective. By themselves they offer no electrical or magnetic shielding. But plain plastic cases
would also flunk the FCC tests. Most manufacturers coat plastic cases with a conductive
paint to contain interference. However, these coatings are largely ineffective against magnetic fields. Most modern systems now use metal cases or internal metal shielding inside
plastic cases to minimize radiation.
No matter the construction of your computer, you can minimize your exposure to radiation from its case by ensuring that it is properly and securely assembled. Minimizing
interference means screwing in the retaining brackets of all the expansion boards inside
your computer and keeping the lid tightly screwed into the chassis. Keeping a tight computer not only helps keep you safe, it also keeps your system safe and intact.
1075
Index
Numbers
2.4MHz wireless networking, 433
2 transfer mode, 873
3D pipeline, 722
3D sound, 897-898
3M Company, 625
4 transfer mode, 873-874
8B/10B coding, 363
9-pin connectors, 381-382
10Base-T hardware standard (networks), 428-430
+12V power connectors,
1020
16-state QAM, 475
18-bit PC Cards, 1006
24-bit color systems, 709
25-pin connectors, 380-381
30/30 disk drive, 585
32-bit CardBus, 1006
40-millimeter profiles,
1020
42-Line Bible, 923
48-bit addressing, 307
486 microprocessors,
119-120
50-pin external connectors,
SCSI, 338-340
50-pin internal connectors,
SCSI, 338
60-millimeter profiles,
1020
64-state QAM, 475
68-pin connectors, SCSI,
340-341
80-pin SCA connectors,

SCSI, 340-341
100Base-T hardware standard (networks), 428-430
256-state QAM, 475
117-118
386 microprocessors, 118
802.11a wireless network
standard, 433-434
802.11b wireless network
standard, 432-433
802.11b+ wireless network
standard, 435
802.11g wireless network
standard, 435-436
116-117
8253 Timer/Counter Chip,
164-165
A
A connector, 395-397
ABC (Atanasoff Berry
Computer), 6
aborted frames, IrDA, 371
absolute lead-out area, 676
AC (alternating current),
1012
links, 910
power, 810
AC-3, 899-900
AC97 Bit clock, 162
Accelerated Graphics Port.
See AGP
1078
accelerator chips
accelerator chips, 828-830
access
cases
front panel access, 1067
internal, 1066-1067
hard disk performance,
606-607
memory, 517
bank-switched memory,
518-519
direct access, 517-518
optical media drives, 658
access points, 444-445

accessed flags, 542
accumulators, 54
acknowledge lines, 402
Acknowledge signal (DMA),
316
ACPI (Advanced
Configuration and Power
Interface), 185-186, 1041
APM (Advanced Power
Management), 1045
configuration, 1047
expansion boards, 186-188
Machine Language (AML),
203
Soft Off mode, 1046
Source Language (ASL), 203
states, 1046-1047
tables, BIOS legacy data, 203
ACR (Advanced
Communications Riser),
1005-1006
ACR SIG (Advanced
Communications Riser
Special Interest Group),
1005-1006
active cooling, 104,
1070-1071
active matrix LCDs, 855
Active mode, Attachment
power management,
309-310
active speakers, 921
active systems, 921
adapters
cables, RS-232C ports, 386
display, 813, 827-828
memory, 832-833
RAMDACs, 832
video controllers, 828-832
graphics, 819
parallel ports, 398
transparency, 780
Windows network model,
422-423
addresses
buses, 85
extension registers, 289
lines, expansion buses,
242-243
register (enhanced parallel
ports), 412
routines, entry points, 196
spaces, TLB (translation lookaside buffer), 100
addressing, 55-56, 482

AT Attachment, 304
CHS, 305-306
LBA (logical block addressing), 306-307
CD-ROMs, 666
chipsets, 157-158
CIDR, 486
CRTs (cathode ray tubes),
843-844
DHCP (Dynamic Host
Configuration Protocol),
488-489
forbidden addresses, 488
IANA private network reservations, 487-488
IP addresses, 480-483
IPv6, 487
network classes, 485-486
printers, 941
SCSI parallel interface,
326-327
subnet masks, 483-485
Windows limits, 550-551
ADM (Advanced Micro

Devices), 143
Athlon, 144-146
Athlon MP, 146
Duron, 146-147
mobile, 147-148
Opteron, 149
Adobe Systems pagedescription language, 955

ADSL (Asymmetrical
Digital Subscriber Line),
470
Riser (ACR), 1005-1006
Riser Special Interest
Group (ACR SIG),
1005-1006
Advanced Configuration
and Power Interface. See
ACPI
Advanced Intelligent Tape,
639
Advanced Keyboard layout
(IBM), 735-736
Advanced Micro Devices
(AMD), 282
Advanced Power
Management. See APM
Advanced Programmable
Interrupt Controller. See
APIC
Advanced Research Project
Agency. See ARPA
AEG (Allgemeine
Elektricitaets
Gesellschaft), 532
AGP (Accelerated Graphic
Port), 869-870
2 transfer mode, 873
4 transfer mode, 873-874
architecture, 871
boards, 1000-1002
operation, 871-873
INDEX
applications
airplane adapters,
1023-1024
algorithms, cyclical redundancy check (CRC) algorithm, 304
alignment, color, 838
all points addressable
graphics (APA display),
706
Allgemeine Elektricitaets
Gesellschaft. See AEG
alpha channel, 719
alpha-blending, 719
Alternate Mark Inversion
(AMI), 468
alternating current. See AC
Alto workstation, 695
ALU (arithmetic/logic
unit), 84
AMD (Advanced Micro
Devices)
HyperTransport, 282
PowerNow!, 107-109
American digital television

transmission standard
(ATSC), 900
American National
Standards Institute. See
ANSI
American Telephone and
Telegraph Company, 458
AMI (Alternate Mark
Inversion), 468
AML (ACPI Machine
Language), 203
AMP (manufacturer of
electronic component
sockets), 574
Ampre, Andr Marie, 1026
amplification, 63
amplifiers, 906
gamma, 859
video, 858
amplitude
modulation, 233
sound, 885-886
AMR (audio modem riser),

1004
AMS (Array Management
Software), 610
analog audio, 884-885
amplitude, 885-886
decibels, 886-887
distortion, 887-888
frequencies, 885
voltage, 886
analog captures, 902-903

analog connection systems,
875-877
analog mixers, 910
Analog Protection System
(APS), 681
analog services
modems, 450-451
background, 450-451
channels, 452-454
connection-enhancing technologies, 454-458
hardware, 462-466
Shannons Limit, 451-452
standards, 458
ITU (International
Telecommunications
Union), 458
modem speed standards,
458-461
telephone company, 458
analog standards, 961

analog video output, 809
analog voltage level, 876
analog-to-digital converters, 822, 908
Analytical Engine, 6
angular velocity, 656
ANSI (American National
Standards Institute), 320
ATA standards, 320
344-345
ANSI T13 Technical

Committee, 307
anti-glare treatment, CRTs
(cathode ray tubes),
844-846
antialiasing filters, 889
APA display (all points
addressable graphics), 706
aperture
digital camera lenses, 791-793
grilles, 840-841
APIC (Advanced
Programmable Interrupt
Controller), 162
APIs (Application
Programming Interfaces),
51
APM (Advanced Power
Management), 1042-1043
ACPI (Advanced
Interface), 1045-1047
BIOS, 1044
driver, 1044
states, 1043-1044
structure, 1044
Apple
Lisa, 695
QuickTime, 895
applets, 38
application interfacing
(scanners), 782-784
Application layer (OSI
model), 421
application programming
interfaces (APIs), 51
applications, 8, 34-36
applets, 38
application suites, 37
databases, 36
digital electronics, 79-80
disassembling, 35
drawing/painting, 36
email, 36
Microsoft Office, 37
multimedia, 36
How can we make this index more useful? Email us at [email protected]
1079
1080
applications
operating systems. See operating systems
pushbuttons, 736
spreadsheets, 36
utilities, 37-38
Web browsers, 36
word processors, 36
APS (Analog Protection

System), 681
arbitration
FireWire, 364-365
PCI (Peripheral Component
Interconnect) bus mastering,
270
343-344
architecture
Port), 871
cable digital services, 473-474
networks, 424-428
Interconnect), 264-267
ARINC 628 power connectors, 1024

arithmetic/logic unit (ALU),
84
arithmetic/logic unit
(microprocessors), 86
ARPA (Advanced Research
Project Agency), 481
ARPAnet, 481
Array Management
Software (AMS), 610
arrays, hard disks drives,
609-610
data striping, 610
implementations, 611-615
interfacing, 616
parallel access arrays, 615
principles, 610-611
redundancy, 610-611
reliability, 610-611
ASL (ACPI Source

Language), 203
aspect ratio, CRTs, 849
assemblers, 45
assembly languages, 45
Asymmetrical Digital
Subscriber Line. See
ADSL
asymmetrical modems,
453-454
asynchronous buses, 243
asynchronous communications, 247-248
asynchronous SCSI, 325
AT Attachment, 300
addressing, 304-307
capacity, 304-307
compatibility, 317-318
connectors, 313-315
device identification, 311-312
logical interface, 312-313
operation, 313
packet interface, 308-309
performance, 301-304
power management, 307-310
power management modes,
310
security, 310-311
signals, 313-317
standards, 300
wiring, 318-320
AT Attachment Packet
Interface, 308, 902
AT power supplies, 1019
ATA interface, 22
Atanasoff Berry Computer
(ABC), 6
Atanasoff, John Vincent, 6
ATAPI (AT Attachment
Packet Interface), 308,
902
Athlon, 144-146
atmospheric perspective,
718
lighting effects, 718
ray-tracing, 718-719
shading, 719
transparency, 719-720
z-buffering technique, 719
ATSC (American digital

television transmission
standard), 900
attribute phase (data transfers), 273
attributes
bytes, 702
graphic, 708
memory, PC cards, 292-293
monochrome display attributes, 702
nonresident, 605
video, 702-704
ATX
motherboards, 988-989
power supplies, 1019
ATX12V, 1020
audio, 29, 290
Audio Codec (Intel), 890,
910
audio digital, 888
bandwidth, 890-891
capturing sounds, 901-904
compression, 891-900
output systems, 913-915
resolution, 890
sampling rate, 888-890
transmission, 891
audio modem riser (AMR),

1004
audio systems, 883-884
hardware, 904-909
amplifiers, 906
control, 910-911
drivers, 907-908
loudspeakers, 906-907
INDEX
BIOS
quality, 911-913
tone generators, 905-906
transducers, 915-921
sound, 884-887
authoring, 684
Auto Answer indicators,
466
Auto Feed XT, 404
auto-focus system, 803-804
automatic exposure control,
800-801
automatic gain control, 677
Automatic mode
(PowerNow!), 108
automatic park-and-lock,
594
Automatic Sequence
Controlled Calculator, 6
autosensing, dual-speed
NICs, 442
Auxiliary Input functions,
903
auxiliary power connectors,
1019
AV drives, 608
average seek time, 606
B
B connector, 397
B plugs, 357
Babbage, Charles, 6
back buffer, 720
back porches, 817
back-side bus, 122
backlights, 857
digital camera sensing, 799
switches, 801
backplanes, 363, 974

backup power systems, 31
backup rollers, 933
backups, CDs, 904
Baffles, 919-920
Bain, Alexander, 959
bandwidth, 100
communication channels, 227
emissions limitations, 228
standards-based limitations,
228-229
digital audio, 890-891
displays, 861-863
518-519
Bardeen, John, 66
barrel distortion, 850-851
base addresses, RS-232C
ports, 390
base memory. See real
mode memory
baseboards, 974
Basic Input/Output System.
See BIOS
Basic Rate Interface. See
BRI
bass reflex speakers, 920
batch languages, 48-49
batteries, 1024
capacity, 1057-1058
charging, 1038-1040
chemical systems, 1028
carbon-zinc cells, 1029-1030
lead-acid cells, 1030-1032
Lithium, 1035-1037
ni-cads, 1032-1034
nickel-metal hydrides,
1034-1035
zinc-air cell, 1037
digital cameras, 809-810
electrical characteristics,
1024-1025
capacity, 1027
cell types, 1025
currents, 1026
energy, 1027
shelf life, 1028
storage density, 1027-1028
voltages, 1025-1026
maintenance-free, 1031
primary, 1025
safety, 1041-1042
secondary, 1025
Smart Battery system,
1040-1041
standards, 1037-1038
Battery Optimized Mode

(SpeedStep), 107
Battery Saver mode
(PowerNow!), 108
bays (cases), 31, 1067-1068
BEDO (Burst EDO
DRAM), 562
beep codes, POST (PowerOn Self Test), 181-182
Bell, Alexander Graham,
886
Bell 103 modem standard,
458
Bell 212A modem standard,
458
Berry, Clifford, 6
BFT (binary file transfers),
964
bidirectional microphones,
916
bidirectional parallel ports,
392-394
bilinear filtering, 717
bilingual connectors, 366
bin feed mechanisms, 947
binary file transfers (BFT),
964
binaural recording, 896
BIOS (Basic Input/Output
System), 19-21, 53,
175-176
ACPI, 1045
APM, 1044
data area (real mode memory), 553
1081
1082
BIOS
extensions, 182-183
firmware, 176-177
functions, 177
initialization, 179-188
interface, 188-191, 195-198
IPL (Initial Program Load),
178, 183-184
storage, 199-220
interrupt vector table,
191-195
shadowing, 219
teletypes, 698
upgrades, 220-221
bit-block transfers, 713

bit-coding, serial channels,
248-249
bit-depth, 888-890
bit-stuffing, 250-251
BitFAT, 603
bitmaps, 700, 705-707
bit memory cells, 514
black, composite black, 944
blackouts, protection from,
1053
offline backup systems,
1053-1054
online backup systems,
1054-1058
blade servers, 973

blanking display systems,
817
blind transfers, AT
Attachment, 302
blocks
graphics, 705-706
secondary storage sequential
data organization, 530
blue CD-R discs, 651

blue screening, 822
Bluetooth, 353, 371-373
components, 378
hardware, 375-377
history, 372-373
software, 374-375
standards, 378
BNC connectors, 878-879

Boole, George, 74
Boolean algebra, 62
Boolean logic, digital electronics, 74-75
boot block Flash memory,
220, 558
boot sector, 184
booting
cold boots, 179-180
devices, CMOS advanced
setup, 214
hot boots, 179
178, 183-184
warm boots, 179-180
BOSS (Bus
Owner/Supervisor/
Selector), 365
bottles. See CRT (cathode
ray tube technology)
branches
FireWire, 362
prediction logic, 96
Brattain, Walter, 66
BRI (Basic Rate Interface),
472
bridges
expansion buses, 245-246
latency, 246
broadcast model, satellite

systems, 476
browsers. See WWW
(World Wide Web),
browsers
BubbleJet, 930
buffer underrun, 668,
675-676
buffered SDRAM. See registered SDRAM
buffers
back, 720
double-buffering, 720
frame, 51, 706
front, 720
bulk erase Flash memory,

558
burning the PROM, 557
Burst EDO DRAM. See
BEDO
burst mode
expansion mode, 243
Interconnect), 267
Bus Owner/Supervisor/
Selector (BOSS), 365
bus topology. See linear
topologies
bus-master DMA, 304
bus-oriented computers,
970
buses, 21-22, 261-262,
997-998
address, 85
AGP, 1000-1002
asynchronous, 243
back-side bus, 122
burst mode, 243
clock, CMOS advanced setup,
211
communications channels,
225, 228, 239-241
address lines, 242-243
bridges, 245-246
data lines, 241
expansion boards, 240
flow control, 243
masters, 243-244
non-bussed signals, 245
parallel, 237
peripherals, 240
power distribution, 245
system control, 243
timing, 243
controllers
DMA (Direct Memory
Access), 171
USB, 356
INDEX
capacity
data, 85
fairness, 244
front-side, 99-100, 122
InfiniBand Architecture,
276-277
communications, 278-279
history, 277-278
standards, 281
structure, 279-280
memory I/O, 54
multiplexing, 242
Interconnect), 262-264, 998
arbitrated bus-mastering,
270
architecture, 264-267
burst mode, 267
connectors, 999-1000
data integrity signals,
267-268
dimensions, 998-999
hot-plugging, 271
low-voltage evolution, 270
multiplexing, 266
PCI Express, 274-276
PCI-X, 272-274
serialized interrupts,
268-269
setup, 272
slot limits, 270-271
speed, 265-266
standards, 276
PCMCIA cards, 1006-1009
peer, 246
priorities, 244
riser boards, 993, 1002-1006
SCSI, 326, 342
signals, HyperTransport,
284-285
standards, 262-264, 984
synchronous, 243
USB, 354-355
bus controllers, 356
cable, 358-359
connectors, 357-358
function, 355
hosts, 356
hubs, 355
Plug-and-Play, 357
protocol, 356-361
speed, 356
standards, 361
busy lines, parallel ports,

401-402
busy signal, 402
buttons, mouse, 757-758
bytes, 515
attribute bytes, 702
parallel ports, 406-407
C
C connector, parallel ports,
398
Cable TV Media Access
Control and Physical
Protocol Working Group,
474
cables
digital services, 473
standards, 474-476
FireWire, 366-367
floppy disks, 348-350
network hardware, 445
coaxial cables, 445
twisted-pair wires, 445-447
ribbon, 25
RS-232C ports, 385-388
SCSI
connections, 332-334
maximum cable length, 336
parallel interface, 336
terminations, 334-336
USB, 358-359
caching
CMOS advanced setup, 211
data, 102
hard disks, 608
memory, 100
microprocessors, 100-103
calculations, computers
and, 8-9
decision making, 14-15
processing, 19-21
calculators, 6
Atanasoff Berry Computer
(ABC), 6
Automatic Sequence
Controlled Calculator, 6
cameras (digital), 787-788

exposure, 798-801
flash, 801-802
focusing, 802-804
image storage, 805
imaging sensors, 789-791
lenses, 791
aperture, 791-793
conversion, 795
focal length, 793-794
supplemental, 795-796
power, 809-810
shutters, 796-798
still, 788
video, 788-789
view panels, 807-808
viewfinders, 806
Web, 789
white balance, 804
zoom, 794-795
cameras, digital, 28
Canon BubbleJet, 930
capacitive keyboards, 729
capacitors, 521, 916
capacity
AT Attachment, 304
CHS addressing, 305-306
CHS translation, 306
LBA (logical block addressing), 306-307
batteries, 1027, 1057-1058
CD-Rs, 669
CD-ROMs, 666, 669
1083
1084
capacity
printer paper, 948
326-327
UPSs, 1057-1058
capturing sound, 901

analog captures, 902-903
CD-ripping, 901-902
legal issues, 904
car adapters, 1023-1024

carbon-zinc cells (batteries), 1029-1030
card guides, 977
card identification structure
(CIS), PC cards, 291-292
card select numbers, 187
card services, PC cards, 291
CardBus, 263, 293-294
Carpal Tunnel Syndrome,
737
carrier detect (CD),
RS-232C ports, 384
Carrier Detect indicators,
465
Carrier Sense Multiple
Access with Collision
Avoidance (CSMA/CA),
440
Carrier Sense, Multiple
Access with Collision
Detection (CSMA/CD),
429
carrier wave modulation,
229-232, 649
cartridges, 112-113,
952-953
CAS (Column Address
Strobe signal), 217, 560,
563
cases, 31, 979-980, 1061
access, 1067
bays, 1067-1068
cooling, 1069-1071
expansion slots, 1069
internal access, 1066-1067
mechanical construction,
1061-1062
radiation, 1072-1075
sizes, 1063-1066
categories, twisted-pair
wires, 446
cathode ray tube (CRT)
technology, 834-835
addressability, 843-844
anti-glare treatment, 844-846
aperture grilles, 840-841
aspect ratios, 849
color temperature, 836-837
convergence, 838-839
dot-pitch, 841
electron guns, 837-838
image controls, 851-852
image distortion, 850-851
image sizing, 850
line width, 842
monitors, 28
over scans, 848-849
phosphors, 835-836
purity, 839
resolution, 843-844
screen curvature, 842-843
screen size, 846-848
shadow masks, 839-840
under scans, 848-849
CCD (charge coupled

device), 815
CD (carrier detect), 384
CD-DAs (Compact Disc,
Digital Audio), 660-661
CD-I (Compact Disc
Interactive), 644
CD-ROM standards, 663
capacity issues, 669
logical format, 663-667
CD-Rs (Compact DiscRecordable), 667

recording materials, 649-651
standards, 667-669
logical blocks, 669-670
operation, 670-677
CD-RWs, 652, 677-678

CDs (Compact Discs), 25,
539, 643
ATAPI-based (IDE), 902
backups, 904
disc medium, 645-648
drives, 654-655
access time, 658
compatibility, 659
controls, 658-659
speed, 655-658
history, 643-645
internal, 902
physical format, 653-654
reading, 657
ripping, 901-902
standards, 659-660
track pitch, 653
cells (batteries), 1025

bits, 514
carbon-zinc, 1029-1030
lead-acid, 1030-1032
Lithium, 1035-1036
lithium solid polymer cell, 1036
lithium-iron disulfide,
1036-1037
ni-cads, 1032-1034
1034-1035
zinc-air, 1037
central processing units. See

CPUs
Centronics Data Computer
Corporation, 391
CFA (CompactFlash
Association), 576
chaining interrupts, 195
Change Display Type dialog
box, 869
channel adapters, 280
channels (modem technology), 225-227, 452
alpha channel, 719
asymmetrical modems,
453-454
bandwidth, 227-229
INDEX
clients
Bluetooth, 376
echoplex mode, 452-453
expansion buses, 225, 228,
239-241
bridges, 245-246
data lines, 241
flow control, 243
masters, 243-244
parallel, 237
peripherals, 240
system control, 243
timing, 243
full-duplex modems, 452
group-coding, 251-254
control, 257-258
error handling, 256-257
frames, 254-256
half-duplex modems, 452
modulation, 229-236
parallel, 236-237
ports, 237-239
serial, 246
bit-coding, 248-249
clocking, 247-248
No Return-to-Zero (NRZ),
249
No Return-to-Zero Inverted
(NRZI), 249-250
Return-to-Zero coding
(RTZ), 249
Return-to-Zero Inverted
(RTI), 249
switching modems, 453
character boxes, 701-702

character ROM, 701
character technologies,
699-700
character boxes, 701-702
character-mapping, 700
direct writing, 700-701
video attributes, 702-704
video pages, 704
character-mapping, 700,
943
charge coupled device
(CCD ), 815
chargers
constant voltage chargers,
1039
high-current, 1040
trickle, 1040
charging
batteries, 1038-1040
rollers, 933
chemical systems, batteries,

1028
carbon-zinc cells, 1029-1030
lead-acid cells, 1030-1032
Lithium, 1035-1036
lithium solid polymer cell,
1036
lithium-iron disulfide cell,
1036-1037
ni-cads cells, 1032-1034
1034-1035
zinc-air cell, 1037
Chiariglione, Leonardo,
824
chips, 68
errors, 524-525
memory packaging, 567
chipsets, 19-20, 153-155

functions, 155-156
memory control, 156
addressing, 157-158
refreshing, 158
peripheral control, 172-173
system control, 159
clocks, 161-164
DMA (Direct Memory
Access), 169-171
oscillators, 160-161
PCI (programmable interrupt controller), 165-169
power management, 171
timers, 164-165
timing circuits, 159-160
cholesteric technology,
LCDs (liquid crystal
displays), 855
chroma keying, 821
chrominance, 820
CHS addressing, AT
Attachment, 305-306
CHS translation, AT
Attachment, 306
CIDR (Classless InterDomain Routing), 486
circuits
integrated circuits, 68-69
logic, 62-64, 1015
Moores Law, 69-70
printed, 70-73
support, 153-154
support chips, 154
timing, 159-160
CIS (card identification

structure), PC cards,
291-292
CISC (Complex Instruction
Set Computers), 90
Class A (network class),
486
Class B (network class),
486
Class C (network class),
486
Class D (network class),
486
Class E (network class),
486
classes, interface (fax
modems), 964-965
Classless Inter-Domain
Routing. See CIDR
cleaning blades, 933
clear to send (CTS),
RS-232C ports, 383-384
client/server systems
(networks), 427
clients, Windows network
model, 423
1085
1086
CLK66
CLK66, 162
clocked logic, digital electronics, 80-81
clocking
SCSI parallel interface, 324
asynchronous SCSI, 325
speed, 342-343
synchronous SCSI, 325-326
serial channels, 247-248
clocks, 159
Bluetooth, 375
bus, 211
chipsets, 161-164
over clocking, 216
speed. See speed, clock
Close-up converters, 795

closing the disc, 676
clusters, FAT, 601-603
CMOS (Complementary
Metal Oxide
Semiconductor), 67-68
BIOS system configuration,
204-205
advanced setup, 209-214
memory assignments,
205-208
resetting CMOS, 208-209
CNR (Communications and

Networking Riser),
1004-1005
coating, anti-glare treatments for displays, 845
coaxial cables, 445
cobalt, as magnetic material, 534
Cocke, John, 90
code
bit-coding, 248-249
boot sector, 184
entry points, 196
control, 257-258
frames, 254-256
249
(NRZI), 249-250
object, 47
passing parameters, 195-196
(RTZ), 249
(RTI), 249
source, 47
terminating, 964
codecs, 892
Audio Codec 97, 910
image compression, 823
cold boots, 179-180

cold fusers, 931
collation, printer paper, 951
collectors, 67
color coding, 710-711
color spaces, 711-712
color-mapping, 712
direct-mapping, 712
graphic commands, 712-713
bit-block transfers, 713
drawing commands, 713
panning, 714-715
sprites, 713
windowing, 714
twisted-pair wires, 447
color
alignment, CRTs, 838
coding. See color coding
display attributes, 704
LCDs (liquid crystal displays),
857
mapping, 712
planes, 708-710
printers, 943-946
spaces, 711-712
temperature, 804, 836-837
tracking, 858
triads, 835
triplets, 835
colored substrate, CD-Rs,

650-651
Column Address Strobe
signal. See CAS
command driven control
systems, 695
Command Packet (ATAPI),
308
commands
complex instructions, 77
graphic, 712-715
sets, 44, 77, 954
software, keyboards, 742-744
View menu, Resources, 167
Common Command Set,

SCSI, 323
common memory, 288, 293
communications, 21,
225-226
channels, 225-227
bandwidth, 227-229
expansion buses, 225, 228,
237-246
modulation, 229-236
parallel, 236-237
ports, 237-239
serial, 246-250
Communications and
Networking Riser (CNR),
1004-1005
278-279
input/output ports, 22
interfaces, 22
Internet, 23
LANs, 23
packetized communications,
257
telecommunications, 23,
449-450
analog services, 450-466
digital services, 466-477
WANs, 23
wireless, 23
INDEX
connection-enhancing technologies
Compact Disc Interactive
(CD-I), 644
Compact Disc, Digital
Audio. See CD-DAs
Compact Disc-Recordable.
See CD-Rs
Compact Discs. See CDs
CompactFlash Association
(CFA), 576-577
Compaq Computer
Corporation, 625
compatibility
ATA Attachment, 317-318
BIOS interface, 190-191
keyboards, 746-747
optical media drives, 659
acknowledge lines, 402
busy lines, 401-402
data lines, 401
fault signal, 403
initialize printer signal,
403-404
nAutoFd, 404
PError, 403
select input signal, 404
select signal, 402-403
strobe lines, 401
sound systems, 909-910
SuperDisks, 628
compiled programming
languages, 46-47
compilers, 47, 91
Complementary Metal
Oxide Semiconductor. See
CMOS
Complex Instruction Set
Computers. See CISC
complex instructions, 77
complex modulation, 236
components, 17-19
BIOS, 19-21
Bluetooth, 378
chipsets, 19-20
external, 903
infrastructure, 30-31
input devices, 26-29
integrated circuits, 20
interfaces, 22
memory, 24-26
modems, 23
monitoring devices, 28-29
composite black, 944

composite sync, 816
composite video, 816, 819
compression
digital audio, 891-894
3D sound, 897-898
AC-3, 899-900
DTS (Digital Theater
Systems), 900
Internet sound systems, 895
looped sound, 894-895
multichannel sound,
895-896
quadraphonic recording, 897
SDDS (Sony Dynamic
Digital Sound), 900
stereo, 896
surround sound, 898-899
FAT, 603
Group 3 standards, 962-964
ratios, 544
compromise computers,
972-973
computer interfaces as
modem hardware, 462
computers, 5, 10. See also
PCs
blade servers, 973
boards, 71
bus-oriented, 970. See also
buses
calculations, 8-9
decision making, 14-15
processing, 19-21
cases, 979-980
components. See components
compromise, 972-973
digital electronics, 12-13,
61-62
electronic design, 62-73
logic, 13-14
logic design, 74-81
electronics, 11
functions. See functions
history, 5-7
intelligence, 8-10
low-profile, 988
low-profile computers, 1063
memory, 8-10, 24-26
motherboard design, 969-970
multitasking, 79
power, 979
present day, 7-9
programmability, 14
single-board, 971
small footprint, 1063
concurrent refresh, CMOS

advanced setup, 212
condenser microphones,
916
Config/OS layer (PCI
Express), 275
configurations
ACPI, 1047
BIOS system, 203-214
FireWire, 364
monitors, 868-869
PC cards, 290
connection-enhancing
technologies (modems),
454
combining voice and data,
457-458
data compression, 456
echo cancellation, 455
error-checking and errorcorrection systems, 456-457
fallback, 455
line compensation, 454
multiple-carrier modems, 455
1087
1088
connections
connections
cable, SCSI, 332-334
keyboards, 747-749
connectivity (networks), 21,

417-418
architecture, 424
hierarchies, 426-428
topologies, 424-426
hardware, 441
cabling, 445-447
hubs, 443-444
NICs, 442-443
routers, 445
Internet model, 422
OSI model, 418-421
standards, 428
Ethernet, 428-431
HomePlug, 438-441
HomePNA, 436-438
wireless network standards,
431-436
Windows model, 422-423
connectors, 977-979
+12V power connectors, 1020
AGP boards, 1002
ARINC 628, 1024
auxiliary power, 1019
edge, 978
Empower, 1024
FireWire, 365-366
floppy disks, 350
microphones, 918
monitors, 874-879
parallel ports, 395
A, 395-397
adapters, 398
B, 397
C, 398
cable, 398-399
electrical operation, 399-410
PCI boards, 999-1000

PCMCIA cards, 1009
RS-232C ports, 380
9-pin, 381-382
25-pin, 380-381
50-pin external, 338-340
50-pin internal, 338
68-pin, 340-341
80-pin SCA, 340-341
USB, 357-358
constant angular velocity,

653
equipment, 586
medium, 653
recording, 586-587

1039
consumables, printers,
951-954
contact
keyboards, 727
mechanical switches, 727
membrane switches, 728
non-contact, 728-729
rubber dome switches, 728
materials, 574
contaminants, magnetic
storage, 597-598
content, WWW (World
Wide Web), 481
control
digital camera exposure,
799-801
functions, sound systems,
910-911
hierarchy, hardware interfaces,
50-51
keys (keyboards), 735-737
Control Method Battery

Interface, 1041
control unit (microprocessors), 86
controllers
bus, 356
floppy disks, 347
keyboard, 739
RAID, 616
video, 828-832
controls
image, 851-852
optical media drives, 658-659
convergence, CRTs (cathode ray tubes), 838-839

conversion lenses (digital
cameras), 795
converters
analog-to-digital, 908
analog-to-digital converter
(A-to-D converter), 822
DAC (digital-to-analog converters), 910
digital-to-analog, 908
RAMDACs, 832
cooling
active, 104
cases, 1069-1071
passive, 104
coprocessors, graphics, 829

copy generation management system, 682
Corel WordPerfect Office,
37
coulombs, 1027
country codes, top-level
domains, 490-498
CPUs (central processing
units), 31, 84
CRC (cyclical redundancy
check) algorithm, 304
crossover cables, 386-388
CRT (cathode ray tube)
technology, 834-835
anti-glare treatment, 844-846
aspect ratios, 849
INDEX
data transfers
controllers, 831-832
dot-pitch, 841
image sizing, 850
line width, 842
monitors, 28
over scans, 848-849
phosphors, 835-836
purity, 839
resolution, 843-844
CSMA/CA (Carrier Sense

Multiple Access with
Collision Avoidance), 440
CSMA/CD (Carrier Sense,
Multiple Access with
Collision Detection), 429
CTS (clear to send), 384
Curie temperature, 538
currents
AC (alternating currents),
1012
batteries, 1026
DC (direct current), 1012
cut-sheet feeders, 946-947

cutoff voltage, 1039
CW (carrier wave) modulation, 231-232
Cycle Frame (FRAME#),
266
cycle shaving, 217
cyclical redundancy check
(CRC) algorithm, 304
cylinders, hard disks,
598-599
D
D-max scanners, 778
DAC (digital-to-analog
converter), 910
DAE (digital audio extraction), 655
daisy-chain wiring system,
876
FireWire, 362
SCSI
bus width, 342
connectors, 333-334
terminations, 334-336
DASD (Direct Access

Storage Device), 585
DAT (Digital Audio Tape),
639
data access arrangements,
463
data area, BIOS, 199-200
data bits, 254
data buses, 85
data caches, 102
data coding, CD-ROMs,
664-665
Data Communication
Equipment (DCE), 383
data compression, 456,
543-545
compression ratios, 544
disk compression systems, 544
file compression, 545
lossless compression systems,
544
data density, optical memory storage, 539

Data Encryption Standard
algorithm. See DES algorithm
data integrity signals, PCI
(Peripheral Component
data lines
expansion buses, 241
parallel ports, 401, 410
ports, 238-239
Data Link layer

Ethernet, 429-430
OSI model, 420
PCI Express, 275
Data Migration Facility, 392

data organization
Data Organization Layer, PC
cards, 292
secondary storage (memory),
529
random-access media, 531-532
sequential media, 529-531
Data Over Cable Service

Interface Specification. See
DOCSIS
data packets, USB, 360
data preparation, as modem
hardware, 463
data rates
Group 3, 961-962
Group 4, 964
IrDA, 369
Data Recording Format

Layer, PC cards, 292
data registers (enhanced
parallel ports), 412-413
data set ready (DSR), 383
data striping, 610
data strobe coding, 363
Data Terminal Equipment
(DTE), 383
Data Terminal Ready
(DTR), 383
data transfers
attribute phase, 273
hard disk mechanism, 590
hard disk performance,
607-608
non-cache coherent, 273
PC cards, 289
register-to-register transfers,
273
1089
1090
database software
database software, 36
date, BIOS legacy data, 200
daughter boards, 975
DBPSK (differential
quarternary phase shift
keying), 439
DCE (Data Communication
Equipment), 383
DDC (Display Data
Channel), 865-868
DDR DIMMs, 570
DDR memory (double datarate memory), 563-564
de Coulomb, C. A., 1027
De Forest, Lee, 65
Dealey, William, 733
debouncers, 729
decibels, sound, 886-887
decision making, computer
calculations, 14-15
Deep Blue, 8
definitions, RS-232C ports,
382-384
deflection angles, anti-glare
treatments for displays,
846
demand paging, virtual
memory, 541-542
demodulation, 230
densities, 596
double-density, 619
depth cueing, 718

depth effects, 717
DES algorithm (Data
Encryption Standard algorithm), 476
design
blade servers, 973
bus-oriented computers, 970
CardBus, 263
compromise computers,
972-973
digital electronics, 62, 74

Boolean logic, 74-75
clocked logic, 80-81
electronics, 64-65
interrupts, 80
logic circuitry, 62-64
logic gates, 75-76
memory, 76
microprocessor instructions,
76-77
Moores Law, 69-70
pin-in hole technology, 72-73
printed circuits, 70-72
programs, 79-80
registers, 77-79
semiconductors, 66-68
surface mount technology, 73
vacuum tubes, 65
motherboard designs, 969-970
optical mouse, 755-757
read/write heads
electrical design, 591-592
head actuators, 593-594
landing zone, 594-595
physical design, 590-591
PRML (Partial Response
Maximum Likelihood),
592-593
write precompensation, 592
single-board computers, 971
desktop cases, 1064

desktop power supplies,
1013
ATX, 1019
linear, 1014
logic circuits, 1015
ratings, 1015-1017
switching, 1014-1015
voltages, 1015-1018
detection, memory errors,

526
detection/correction, 527-528
fake parity, 527
parity, 526
developing rollers, 933

development environments,
48
Device Manager
DMA (Direct Memory
Access), 170
I/O port usage, 199
interrupt assignments,
167-168
device-dependent language,
955
devices
drivers, 52-53
heights, 1067-1068
identification, 311-312
input, 696
interfaces, 54-56
pass-through, 362
point, 695-696
pointing, 753
drivers, 767-768
game controllers, 764-765
interfaces, 765-767
mouse, 753-759
protocols, 767-768
TouchPads, 761-763
trackballs, 759-761
TrackPoints, 763-764
power demands, 1017
DFP (Digital Flat Panel),

879
DHCP (Dynamic Host
488-489
dialog boxes
Change Display Type, 869
Windows Select Network
Component Type, 422
diaphragms, 916-918
Difference Engine, 6
differential quarternary
phase shift keying
(DQPSK), 439
differentiated definition
blocks, 1047
INDEX
direct-mapping
digital audio, 884, 888
bandwidth, 890-891
capturing sounds, 901-904
3D sound, 897-898
AC-3, 899-900
Systems), 900
Internet sound systems, 895
multichannel sound,
895-896
SDDS (Sony Dynamic
Digital Sound), 900
stereo, 896
resolution, 890
transmission, 891
digital audio extraction

(DAE), 655, 901
Digital Audio Tape (DAT),
639
digital cameras, 28,
787-788
exposure, 798
control, 799-801
sensing, 798-799
flash, 801-802
focusing, 802-804
image storage, 805
imaging sensors, 789-791
lenses, 791
aperture, 791-793
conversion, 795
power, 809-810
shutters, 796-798
still cameras, 788
video cameras, 788-789

view panels, 807-808
viewfinders, 806
Web cameras, 789
white balance, 804
zoom, 794-795
Digital Data Storage,

639-641
digital electronics, 12-13,
61-62
electronic design, 62
electronics, 64-65
logic circuitry, 62-64
Moores Law, 69-70
vacuum tubes, 65
logic, 13-14, 74
interrupts, 80
logic gates, 75-76
memory, 76
76-77
programs, 79-80
registers, 77-79
digital film, 805

Digital Flat Panel (DFP),
879
digital magnetic systems,
535-536
digital mixers, 910
digital photography, 787
cable services, 473-476
satellite services, 476-477
telephone services, 467
DSL, 469-471
ISDN, 472-473
maximum data rates, 467
SDS 56, 471-472
T1, 468-469
VDSL, 471
digital signal processor. See

DSP
Digital Subscriber Line. See
DSL services
Digital Theater Sound
(DTS), 681
Digital Versatile Discs. See
DVDs
digital video interface
(DVI), 879-880
digital zooming, 795
digital-to-analog converters
(DAC), 908-910
digitization, 62
dimensions
AGP boards, 1000-1001
PCI boards, 998-999
DIMMs (Dual Inline

Memory Modules), 568
DDR DIMMs, 570
ECC memory (errorcorrection code memory),
569
old technology DIMMs, 569
presence detect pins, 569
SDR DIMMs, 570
serial presence detect pins,
569
DIN connectors,
PS/2-style, 747-749
DIP (dual inline pin), 110
DirecPC satellite service,
476-477
Direct Access Storage
Device (DASD), 585
direct current (DC), 1012
Direct Memory Access. See
DMA
Direct Stream Digital
(DSD), 662-663
direct thermal printers,
935-936
direct writing, 700-701
direct-mapping, 712
1091
1092
directional antennae
directional antennae (access
points), 444
directionality, microphones,
916-917
dirty flags, 542
disassembling programs, 35
disc mediums, optical
media, 645-648
disc-at-once mode (writing
to CDs), 673-674
disc-writing modes, CD-Rs,
672
disc-at-once, 673-674
Mount Rainier, 674-675
packet writing, 674
track multisession mode,
672-673
track-at-once, 672
discrete switch design, 727

disks. See also floppy disks
caching, hard disk performance, 608
compression systems, 544
drives, 587. See also hard disks
30/30, 585
bays, 31
Winchester, 585
parameter table, 201-202, 598
display adapters, 813,

827-828
memory, 832-833
RAMDACs, 832
Display Data Channel. See

DDC
displays, 28-29, 813,
833-834. See also monitors
compared to monitors, 833
CRT (cathode ray tube),
834-835
anti-glare treatment,
844-846
aspect ratios, 849

dot-pitch, 841
image sizing, 850
line width, 842
over scans, 848-849
phosphors, 835-836
purity, 839
resolution, 843-844
display adapters, 827-828
memory, 832-833
RAMDACs, 832
dot-clocks, 862
electronics, 858-859
bandwidth, 861-863
DDC (display data channel),
865-868
Energy Star, 863-864
hard-wired coding, 865
identification, 864
interlacing, 861
manual configuration,
868-869
synchronizing frequency
range, 859-861
flat-panel, 834, 852-853
frequencies, 815
image compression, 822-823
codecs, 823
filters, 823
JPEGs, 823-824
MPEGs, 823-826
interfaces, 869
AGP (Accelerated Graphics
Port), 869-874
connectors, 874-879
digital video interface (DVI),

879-880
UMA (Unified Memory
Architecture), 874
834, 852-853
cholesteric technology, 855
color issues, 857
double-scanned displays, 856
nematic technology, 853-855
reflective displays, 858
response time, 856
transmissive displays, 857
yields, 856-857
portrait, 848
video integration, 818
standards, 818-821
video capture, 822
video overlay, 821-822
video standards, 826-827
video technology, 814
back porches, 817
blanking, 817
front porches, 817
retracing, 816-817
scanning, 814-815
synchronizing signals, 815-816
vertical intervals, 817-818
DisplayWriter, 617
distortions
CRT images, 850-852
sound, 887-888
Distribution Media Format

(DMF), 623
dithering, 944
DMA (Direct Memory
Access), 169-170, 302
Acknowledge signal, 316
bus controller, 171
bus master, 304
multi-word DMA transfers,
303-304
Plain UltraDMA, 304
Request signal, 316
Single-word DMA transfers,
303
UltraDMA modes, 304
INDEX
DVD-Audio
DMF (Distribution Media
Format), 623
DNS (Domain Name
System), 480, 489,
499-500
root name servers, 480, 501
root zone files, 501
subdomains, 499
top-level domains, 489, 500
country codes, 490-498
organizational versus
national, 489
URLs (Uniform Resource
Locators), 499
DOCSIS (Data Over Cable

Service Interface
Specification), 474-475
Dolby Digital, 899-900
Dolby Pro Logic, 898
Dolby Surround, 898-899
Domain Name System. See
DNS
dome-based keyboards, 730
doping, 66
DOS CD-ROM extensions,
667
DOS tables, 202-203
dot-clocks, 832, 862
dot-pitch, CRTs (cathode
ray tubes), 841
dots per inch (dpi), 708
double data-rate memory.
See DDR memory
double-buffering, 710, 720,
833
double-clocking, 326, 873
double-density, 619
double-scanned displays,
LCDs (liquid crystal displays), 856
double-scanned supertwist
nematic (DSTN), 856
double-shot keycaps, 731
double-supertwist nematic
display, 854
double-words (memory
measurements), 515
downstream data rates
(ADSL), 470
dpi (dots per inch), 708
DQPSK (differential
quarternary phase shift
keying), 439
draft mode, 940
dragging the mouse, 695
DRAM (Dynamic Random
Access Memory), 559
drawing commands, 713
drawing software, 36
Drive Active/Drive 1
Present signal, 317
Drive Reset signal, 317
drivers, 907-908, 918
APM, 1044
pointing devices, 767-768
printer, 945
drives
arrays, hard disks, 609-615
AT Attachment, 319-320
bays, 980
cabling, 349
CDs, 654-655
access time, 658
compatibility, 659
controls, 658-659
speed, 655-658
DVDs, 654-655
access time, 658
compatibility, 659
controls, 658-659
speed, 655-658
embedded servo drive, 599
select jumpers
AT Attachment wiring, 320
floppy disks, 349
driving the line, 388

drum scanners, 772-773
DS1 format, 468
DSD (Direct Stream
Digital), 662-663
DSL services (Digital

Subscriber Line), 469
ADSL (Asymmetrical Digital
SDSL (Symmetrical Digital
speed, 505
VDSL (Very-high-data-rate
471
DSP (digital signal processor), 756

DSR (data set ready), 383
DSTN (double-scanned
supertwist nematic), 856
DTE (Data Terminal
Equipment), 383
DTR (Data Terminal
Ready), 383
Sound), 681, 900
Dual Inline Memory
Modules. See DIMMs
dual inline pin (DIP), 110
dual-cavity PGA, 121
dual-speed NICs, 442
dual-standard systems
(wireless network standard), 436
ducted speakers, 920
dumb terminals, 698
duplex communications,
452
duplex operation, 951
Duron, 146-147
DV. See FireWire
DVD+R, 685
DVD-Audio, 682-683
rewritable formats, 685
Blu-ray, 688-689
DVD+RWs, 687-688
DVD-RAMs, 685-686
DVD-RWs, 686-687
write-once DVD formats,
684-685
1093
1094
DVD-R(A)
DVD-R(A), 684
DVD-RAMs, 685-686
DVD-ROMs, 678-679
DVD-RWs, 686-688
DVD-Videos, 679-682
DVDs (Digital Versatile
Discs), 25, 539, 644
drives, 654-659
history, 643-645
reading, 657
standards, 659-660
DVI (digital video interface), 879-880

Dvorak, August, 733
Dvorak-Dealey (keyboard
layout), 733-734
dye-diffusion printers,
937-938
Dynamic Execution, 121
Dynamic Host
Configuration Protocol.
See DHCP
dynamic memory, 521-522
Dynamic Random Access
Memory. See DRAM
dynamic range scanners,
777-778
E
email programs, 36
E.SPEC VER.1 (Ethernet),
429
E.SPEC VER.2 (Ethernet),
429
E1 high-speed telephone
service, 469
ECC memory (errorcorrection code memory),
527, 569
echo cancellation, 455
echoplex mode, 452-453

ECM (error-correction
mode), 963
ECP (extended capabilities
port), 393-394, 408-414
EDAC (Error Detection
and Correction), 527
edge connectors, 978
edge-triggered interrupts,
168
EDID (Extended Display
Identification), 865-866
Edison Effect, 65
EDO (Extended Data Out),
560-562
EDPT (enhanced fixed disk
parameter table), 202
EEPROM (electrically
erasable programmable
read-only memory),
557-558
eight-inch disks, 595
eight-millimeter digital
recording system, 639
EISA (Enhanced ISA), 264
EL panels, 857
electrical characteristics
microphones, 917
microprocessors, 103
operating voltages, 104-106
packaging, 109-114
power management,
106-109
thermal constraints, 103-104
electrical components,
cases, 1072
health concerns, 1074-1075
minimizing interference,
1073-1074
radio frequency interference,
1072-1073
electrical design, read/write

heads, 591-592
electrical functions, keyboards, 738-739
host interface, 744-745
ports, 739-742
software commands, 742-744
USB port, 750
electrical interfacing (scanners), 782

electrical operations
byte mode, 406-407
compatibility mode, 400-404
Enhanced Parallel Port
(EPP) mode, 407-408
Extended Capabilities Port
(ECP) mode, 408-410
nibble mode, 405-406
RS-232C ports, 379
electrical relays, 64
electrically erasable programmable read-only
memory (EEPROM),
557-558
electromotive force (EMF),
1025
electron guns, CRTs (cathode ray tubes), 837-838
electronic design, digital
electronics, 62
digital logic circuitry, 62-64
electronics, 64-65
Moores Law, 69-70
vacuum tubes, 65
electronic displays, 858-859

bandwidth, 861-863
865-868
INDEX
expansion buses
identification, 864
interlacing, 861
manual configuration, 868-869
range, 859-861
Electronic Numerical
Integrator and Computer
(ENIAC), 7
electronic view finder
(EVF), 806
Electronics, 70
electrons, 11
electrophotography, 932
elevator seeking, 607
ELF (extremely low frequency), 1074
embedded servo drives, 599
embedded servo technology, 593
EMF (electromotive force),
1025
emissions, communication
channel limitations, 228
emitters, 67
Empower connector, 1024
Enabled state (APM), 1043
enclosures, loudspeakers,
919-920
encoding, 963
energy, batteries, 1027
Energy Star, 863-864, 1048
Engelbart, Douglas C., 695,
757
enhanced fixed disk parameter table (EDPT), 202
Enhanced ISA (EISA), 264
(EPP), 393, 407-408
address register, 412
data registers, 412-413
Enhanced SpeedStep, 107

ENIAC (Electronic
Numerical Integrator and
Computer), 7
entry points, 196

EPIC (Explicitly Parallel
Instruction Computing),
98-99
EPP (Enhanced Parallel
Port), 393, 407-408
address port, 412
address register, 412
data registers, 412-413
EPROM (erasable programmable read-only

memory), 557
Epson inkjet printers, 931
equalization, 919
equipment
HomePlug, 441
HomePNA, 438
ergonomic designs (keyboards), 737-738

Error Detection and
Correction. See EDAC
error-checking systems,
456-457
error-correction code
memory (ECC) memory,
527, 569
error-correction mode
(ECM), 963
error-correction systems,
456-457
errors
buffer underrun, 668
codes, POST (Power-On Self
Test), 181
error handling
chipsets, 158-159
memory, 523
detection and prevention,
526-528
hard errors, 525-526
soft errors, 524-525
Ethernet, 428
Data Link layer, 429-430
Gigabit Ethernet, 431
Physical layer, 430-431
EuroDOCSIS, 475
Everybodys Autobiography
(italic), 480
EVF (electronic view
finder), 806
expanded memory, 518
expansion boards, 21, 30,
240, 974, 997-998
ACPI, 186-188
AGP, 1000-1002
expansion slots, 976-979
full-length expansion boards,
977
nomenclature, 974-976
PCI, 998-1000
PCMCIA cards, 1006-1009
Plug-and-Play, 186
riser, 993, 1002-1006
expansion buses, 21-22,

261-262. See also buses;
data buses
asynchronous, 243
back-side, 122
burst mode, 243
225, 228, 239-241
bridges, 245-246
data lines, 241
flow control, 243
masters, 243-244
parallel, 237
peripherals, 240
system control, 243
timing, 243
expansion boards. See expansion boards
fairness, 244
front-side, 99-100, 122
276-280
memory I/O, 54
multiplexing, 242
1095
1096
expansion buses
arbitrated bus-mastering,
270
burst mode, 267
data integrity signals,
267-268
hot-plugging, 271
multiplexing, 266
PCI-X, 272-274
serialized interrupts,
268-269
setup, 272
speed, 265-266
standards, 276
peer, 246
priorities, 244
SCSI, 342
signals, 284-285
standards, 262-264, 984
synchronous, 243
expansion slots, 21, 976-977

cases, 1069
connectors, 977-979
layout, 986-987
numbers, 985
retaining brackets, 977
spacing, 985-986
substrates, 977
Explicitly Parallel
Instruction Computing.
See EPIC
exposure
compensation controls, 801

(ECP) mode, 393-394,
408-414
extended control register,
413
Extended Data Out. See

EDO
Extended Display
Identification (EDID),
865-866
extended memory. See protected mode memory
extensions, BIOS, 182-183
external components, capturing sounds, 903
external computer modems
computer interfaces, 462
packaging hardware, 464
external devices, SCSI cable

external power supplies,
1022-1023
extraction, digital audio,
901
extremely low frequency
(ELF), 1074
F
f-stops, 791-793
fabrication, printed circuits,
71-72
facsimile transmissions. See
faxes
fairness buses, 244
fairness interval, 364
fake parity, detection and
prevention of memory
errors, 527
fallback, 455
fans
failures, cases, 1071
Farnsworth, Philo T., 814

fast page-mode RAM. See
FPM RAM
fast parallel port control
register, 413
FAT (File Allocation Table),

601
clusters, 601-603
compression, 603
hard disks, 601-602
NTFS, 604-606
fault signal, parallel ports,

403
fault tolerance, 611
faxes, 29, 957-958
group standards, 961-964
hardware for fax machines,
959
modems, 959-960
multifunction printers,
959-961
interface classes, 964-965
reasonable, 957
technologies, 958-959
teleprinters, 957
FC-AL (Fibre Channel

Arbitrated Loop), 345
FDPT (fixed disk parameter
table), 202
feature matching (scanners), 781
Federal Communications
Commission, 818
Fibre Channel, 345-346
Fibre Channel Arbitrated
Loop (FC-AL), 345
figure-eight microphones,
916
File Allocation Table. See
FAT
file compression, 456, 545
File NODE. See FNODEs
file systems, FAT, 601-602
clusters, 601-603
compression, 603
NTFS, 604-606
film, digital, 805

filtering, 717
filters, 471
antialiasing filters, 889
image compression, 823
INDEX
functions
finalization process, 676
firewalls, 503-505
FireWire, 353, 361-363
arbitration, 364-365
cable, 366-367
configuration, 364
connectors, 365-366
daisy-chaining, 362
leaf, 362
signaling, 363
firmware, BIOS, 176-177

first-in, first-out buffer, 388
Fisheye converters, 796
fixated for appending, 676
fixation, CD-Rs, 676-677
Fixed ACPI Description
Table, 1047
fixed disk parameter table
(FDPT), 202
fixed disks, 585
fixed focus, 803
flash digital cameras,
801-802
Flash memory, 558-559
flat-panel displays, 834,
852-853. See also LCDs
flatbed scanners, 773
Fleming, John, 65
FlexATX motherboards,
991-992
flip-flop circuits, 522
flip-flop memory, 76
floating-point unit (microprocessors), 87-88
floppy disks, 346-347,
616-617
cable, 348-350
chipsets, 173
controllers, 347
formats, 622-625
history, 617-618
density key, 621-622
hubs, 620
insertion key, 620-621
liners, 620
magnetic properties,
618-619
medium, 618
shell, 619-620
writing protection, 621
mechanisms, 628-629
operation, 347-348
SuperDisks, 625-628
floppy drives, 24-25

flow control
AT Attachment, 302
ports, 239
RS-232C ports, 385
flux transitions, 252

FNODEs (File NODE),
604
focal length, digital camera
lenses, 793-794
focus (cameras), 793,
802-804
fog color, 718
fogging, 718
form factors, 1067
formats
CD-ROMs, 667
hard disks, 600-601
IrDA, 371
SuperDisks, 626-627
four-color printing, 944

FPM RAM (fast page-mode
RAM), 560-561
frame buffer memory, 554
frames
buffers, 51, 706
IrDA, 371
rates, 815, 859
freezing the disc, 677

frequencies
Bluetooth, 376-377
display systems, 815
dividers, 160
LFE (low frequency effects),

899
modulation, 233-234, 253
multipliers, 160
Nyquist, 888
response, 911
sound, 885
synchronizing, 859-861
frequency shift keying

(FSK), 232, 377
fretting, 574
friction feeds, 947
front buffer, 720
front panel access, 1067
front porches, 817
front-side buses, 99-100,
122
Frozen mode, AT
Attachment security, 311
FSK (frequency shift keying), 232, 377
Full On state (APM), 1043
full-bleed printing, 950
full-duplex modems, 452
full-length expansion
boards, 977
full-range speaker systems,
920
full-size slot, 1069
full-size tower cases, 1065
fully formed character
printers, 924
function keys (keyboards),
735-737, 740
functions, 17-19
BIOS, 177
initialization, 179-188
interface, 188-191, 195-198
178, 183-184
storage, 199-220
chipsets, 155-156
communications, 21, 225-226
channels, 225-258
1097
1098
functions
interfaces, 22
Internet, 23
LANs, 23
telecommunications, 23
WANs, 23
wireless, 23
input devices, 26-29
memory. See memory
monitoring devices, 28-29
processing, 19
BIOS, 19-21
chipsets, 19-20
USB, 355
G
G.Lite, 470
galvanometers, 533
games, 764-765
gamma of the amplifier, 859
gate, 67
geometry, hard disks,
598-600
Gibson, Garth, 611
GIF (Graphic Interchange
Format), 712
Gigabit Ethernet, 431
gigahertz, 94
glare, anti-glare treatment
for displays, 844-846
glass-epoxy, 71
gold CD-R discs, 650
Golden Ratio, 849
Gouraud shading, 719
granddaughter boards, 976
Graphic Interchange
Format (GIF), 712
graphics, 27
adapters, 28-29, 819
attributes, 708
commands, 712-715
coprocessors, 829
creating, 720-722
input devices, 696
primitives, 713
raster graphics, 707
three-dimensional, 715
718-720
depth effects, 717
image creation, 720-722
recession, 717-718
tessellation, 715-716
texture-mapping, 716-717
two-dimensional, 705-708
green CD-R discs, 650

ground planes, 72, 912
group standards, fax
machines, 961
Group 1 (analog), 961
Group 3, 961-964
Group 4, 964
control, 257-258
frames, 254-256
guide number (flash), 801

guns
electron, 837-838
inline, 837
Gutenberg, Johannes, 923
H
half-duplex modems, 452
hand scanners, 773-774
hand-wired wiring, 71
handedness (trackballs),
760-761
handling errors, chipsets,
158-159
handshake packets, USB,
359-361
handshaking keyboards, 742
hard copies, 29
hard disks, 24-25, 584. See

also disk drives
drive arrays, 609-616
FAT file system, 601-606
geometry, 598-600
history, 584-585
interface chipsets, 172
mechanisms, 585
data-transfer rate, 590
latency, 588-589
read/write heads, 590-595
rotation, 586-587
speed, 587-588
standby mode, 589
medium, 595-598
platters, 585, 589-591
constant angular velocity
equipment, 586
constant linear velocity
recording, 586-588
determining size of drive
mechanisms, 595
magnetic field, 591
servo surface, 593
Winchester disk drive, 585
hard errors, memory,

525-526
hard stop, 730
hard-wired coding, displays,
865
hard-wired wiring, 88
hardware, 969
audio systems, 904
BIOS links to, 197-198
Bluetooth, 375-377
fax machines, 959-961
flow control, 257, 385
interfaces, 49
APIs, 51
BIOS, 53
control hierarchy, 50-51
device, 54-56
device drivers, 52-53
INDEX
hot-swapping
interrupts, 50, 80
modems, 462-466
motherboards. See motherboards
networks, 441-447
panning, 714-715
amplifiers, 906
control, 910-911
drivers, 907-908
quality, 911-913
transducers, 915-921
windowing, 714
Hardware Emulation Layer

(HEL), 828
Hardware Retriggerable
One-Shot mode (8253
Timer/Counter Chip),
164
Hardware Retriggerable
Strobe mode (8253
165
haze, 718
HCA (Host Channel
Adapter), 280
HDSL (High data rate
469
HDTV (High Definition
Television), 813
head actuators, read/write
heads, 593-594
head control, floppy disks,
629
head indexing, floppy disks,
629
head parking, 594
health concerns, radiation,
1074-1075
height, 851
HEL (Hardware Emulation

Layer), 828
helical recording, 633-634
helical-scan recording tape
systems, 638-641
Henry, Joseph, 64
Hewlett-Packard LaserJet,
932
Hi-Performance mode
(PowerNow!), 108
hibernation (APM), 1043
hierarchies (networks),
426-428
High data rate Digital
HDSL
High Definition Television
(HDTV), 813
high memory area (real
mode memory), 553
high signals, 63
high speed indicators, 466
high-current chargers,
1040
high-level programming
languages, 44-46
compiled, 46-47
48
interpreted, 46
libraries, 47-48
scripting languages, 49
history
Bluetooth, 372-373
computers, 5-7
hubs, 620
liners, 620
magnetic properties,
618-619
medium, 618
shell, 619-620
hard disks, 584-585

HomePNA, 437
equipment, 438
Home Phoneline Networking
Alliance, 438
version 1.0, 437
version 2.0, 438
HyperTransport, 282
277-278
Internet, 481-482
IrDA, 367-368
Holmes, Sherlock, 13
Home Phoneline
Networking Alliance, 438
HomePlug, 438
equipment, 441
HomePlug Powerline
Alliance, 441
Powerline Alliance, 441
protocols, 440-441
transmission methods,
439-440
HomePNA, 436
history, 437-438
Alliance, 438
HomeRun (Tut Systems),

437
horizontal frequency, 815,
859
horizontal retraces, 816
host bridge, 155
Host Channel Adapter
(HCA), 280
HostClk, 409
hosts
keyboard interface, 744-745
USB, 356
hot boots, 179

hot-plugging, PCI
Interconnect), 271
hot-swapping, 286, 611
1099
1100
hotspots
hotspots (Web pages), 499
Howtek Pixelmaster, 931
HTML (Hypertext Markup
Language), 503, 698-699
hubs (network hardware),
443-444
FireWire, 362
floppy disks, 620
switches, 444
USB, 355
Hursley Labs, 585

hybrid discs, 661
hydrides, 1034-1035
hyper-pipelined technology,
135
hyper-pipelining, 96
hyperlinks, 499
Hypertext Markup
Language. See HTML
HyperTransport, 281
bus signals, 284-285
history, 282
standards, 285
structure, 283-285
Hursley Labs, 585

Micro Channel Architecture,
264
MMX (MultiMedia
Extensions), 92
PC Network, 421
IDE (Integrated Drive

Electronics), 300
identification
device, 311-312
displays, 864
SCSI wiring, 329-332
Idle mode, Attachment

power management,
309-310
IEC (International ElectroTechnical Commission),
824
IEEE (Institute of
Electrical and Electronic
Engineers), 515
1284-compatible devices, 394
1284-compliant devices, 394
802.3 specification, 429
1284-1994, 394
1394. See FireWire
images
I
I. link. See FireWire
I/F clock, 162
I/O
bridge, 155
Channel Ready signal, 317
controller hub, 156
mapped, 197
modes, 303
port assignments, 197-198
IA64, 129
IANA (Internet Assigned
Numbers Authority),
487-488
IBA fabric, 279
IBM
Advanced Keyboard layout,
735-736
DisplayWriter, 617
codecs, 823
filters, 823
JPEGs, 823-824
MPEGs, 823-826
CRT characteristics
aspect ratios, 849
controls, 851-852
distortion, 850-851
over scans, 848-849
sizing, 850
digital cameras, 805
sensors, 789-791
impact dot-matrix printers,

925-927
IMVP (Intel Mobile
Voltage Positioning), 107
incremental writing, 674

indexing floppy disks, 629
indicators, 465
Auto Answer, 466
Carrier Detect, 465
High Speed, 466
Modem Ready, 466
as modem hardware, 466
Off-Hook, 466
Receive Data, 465
Send Data, 465
Terminal Ready, 466
inductive read/write heads,

591
Industrial-ScientificMedical band. See ISM
band
Industry Standard
Architecture. See ISA
276-277
communications, 278-279
history, 277-278
standards, 281
structure, 279-280
Infrared (IR) technology,

767
Infrared Data Association.
See IrDA
Infrared Developers
Association. See IrDA
infrared light, 369
Initial Program Load. See
IPL
initialization functions,
BIOS, 179-180
cold boots, 179-180
extensions, 182-183
hot boots, 179
IPL, 183-184
Plug-and-Play, 184-188
POST (Power-On Self Test),
180-182
warm boots, 179-180
INDEX
interfaces
initialize printer, 403-404
initiation key, 187
inkjet printers, 29, 927-929
phase-change inkjets, 931
piezoelectric crystal, 930-931
thermal, 930
inline guns, 837

input devices, 26
Advanced Keyboard, 735-736
audio, 29
digital cameras, 28
faxes, 29
graphic, 696
keyboards, 27, 725-726
capacitive, 729
control keys, 735-737
debouncers, 729
dome-based, 730
electrical functions, 738-750
ergonomic designs, 737-738
function keys, 735-740
layouts, 732-734
legends, 731-732
linear travel, 730
modular keyboard connector,
749
PS/2-style, 747-749
short-travel, 731
soft-touch, 730
stiff, 731
switches, 727-728
touch, 729-731
wireless, 750-751
pointing devices, 27
printers, 29
scanners, 27
input ports, 22, 173

input/output mapping, 54
input/output unit (microprocessors), 85-86
installation, memory, 572
contact materials, 574
memory speed, 573
mixing modules, 573-574

orientation, 575
Institute of Electrical and

Electronic Engineers. See
IEEE
instruction sets (commands), 44, 77
microprocessors, 88
instructions, microprocessors, 76-77

insulators, 66
integer unit, 86
integrated circuits, 20,
68-69
Integrated Drive
Electronics. See IDE
Integrated Services Digital
Network. See ISDN
integration, video display
systems, 818-822
Intel
1004
ATX12V, 1020
Audio Codec, 890
Networking Riser), 1004
Enhanced SpeedStep, 107
Itanium, 142-143
mobile, 137-142
Pentium III, 129-132
Pentium III Xeon, 132-133
Pentium 4, 134-136
Xeon, 136-137
Miniature Cards, 576
Mobile Voltage Positioning
(IMVP), 107
Pentiums. See Pentiums
QuickStart, 106
SpeedStep, 107-108
intelligence, computers,
8-10
inter-record gaps, secondary storage sequential
blocks, 530
interactive media, 825

interfaces, 22, 39, 49,
299-300
ACPI (Advanced
Interface), 1041
APIs, 51
APM (Advanced Power
ACPI (Advanced
Interface), 1045-1047
states, 1043-1044
structure, 1044
AT Attachment, 300
addressing, 304-307
capacity, 304-307
connectors, 313-315
device identification,
311-312
operation, 313
packet interface, 308-309
power management,
307-310
security, 310-311
signals, 313-317
standards, 300
wiring, 318-320
ATA interface, 22
BIOS, 53
classes, fax modems, 964-965
compared to ports, 237
control hierarchy, 50-51
Control Method Battery
Interface, 1041
definition, 39
device, 54-56
device drivers, 52-53
displays, 869
Port), 869-874
connectors, 874-879
1101
1102
interfaces
879-880
UMA (Unified Memory
Architecture), 874
Fibre Channel, 345-346
functions, BIOS, 188-190
entry points, 196
linking to hardware,
197-198
parameter passing, 195-196
software interrupts, 191, 195
hard disk arrays, 616
host, keyboards, 744-745
logical
parallel ports, 410
RS-232C ports, 389-390
printers, 954-955
SCSI interface, 22
SCSI parallel, 322
addressing, 326-327
ANSI, 344-345
bus width, 326
cables, 336
capacity, 326-327
clocking, 324-326
connectors, 337-341
operation, 343-344
signaling, 327-328
standards, 322-324,
341-343
wiring, 328-336
serial, 354, 361-367
Serial ATA, 320-322
software, PC cards, 290-291
UPSs, 1056-1057
user, 35
interference
cases, 1073-1074
suppression, IrDA, 371
interlaced scanning, 861

interlacing displays, 861
interleaving, CMOS
advanced setup, 212
internal access, cases,
1066-1067
internal bays, 1067
internal CDs, 902
internal computer modems
computer interfaces, 462
packaging hardware, 464
internal devices, SCSI cable

internal resistance, 1026
International ElectroTechnical Commission
(IEC), 824
International Standards
Organization (ISO), 824
International
Telecommunications
Union. See ITU
Internet, 23, 479
addressing, 482
CIDR, 486
DHCP (Dynamic Host
488-489
IPv6, 487
content, 481
DNS (Domain Name
System), 480, 489, 499-500
subdomains, 499
top-level domains, 489-500
Locators), 499
history, 481-482
ISPs (Internet Service
Providers), 504-506
routing, 502-503
sound systems, 895
Web pages, 503
WINS (Windows Internet
Name System), 502
Internet Assigned Numbers

Authority. See IANA
Internet model (networks),
422
Internet Protocol Version
6. See IPv6
Internet Service Providers.
See ISPs
interpolated resolution
(scanners), 779
interpreted programming
languages, 46
interpreters, 46
Interrupt 5C(hex) routines,
421
Interrupt 13(hex) addressing, ATA, 306
Interrupt on Terminal
Count mode (8253
164
interrupts, 499
BIOS interface, 191, 195
chaining interrupts, 195
control, 165-169
digital electronics, 80
handlers, 191
hardware, 50
level-sensitive, 168
RS-232C, 390
serialized, PCI, 268-269
service routines, 191
vector table (BIOS), 191-195
video, 698
INDEX
landing zone
intersymbol interference,
592
Intr, 408
inverters, 1023, 1057
178, 183-184
IPv6 (Internet Protocol
Version 6), 487
IR (Infrared) technology,
767
IrDA (Infrared Data
Association), 22, 353,
367-369
data rates, 369
formats, 371
frames, 371
history, 367-368
infrared light, 369
modulation, 370-371
pulse width, 369-370
speed, 370
iris blades, 791

iron, as magnetic material,
534
ISA (Industry Standard
Architecture), 30, 975
ISDN (Integrated Services
Digital Network),
472-473
BRI (Basic Rate Interface),
472
PRI (Primary Rate Interface),
472
ISM band (IndustrialScientific-Medical band),

376, 435
ISO (International
Standards Organization),
824
isochronous communications, 248
isochronous mode
(FireWire), 364
Providers), 499, 504
performance limits, 504-505
security, 505-506
Itanium microprocessors,
129, 142-143
ITU (International
Telecommunications
Union), 458
J
JEITIA (Japan Electronics
and Information
Technology Industries
Association), 286-287
Joint Photographic Experts
Group. See JPEGs
Joint Technical Committee
(JTC), 824
Joliet CD-ROM Recording
Specification, 667
JPEGs (Joint Photographic
Experts Group), 805,
823-824
JTC (Joint Technical
Committee), 824
jukeboxes, near-line storage, 528
K
K56flex, 461
kardia, 916
Katmai. See Pentiums
Katz, Randy H., 611
Key Tronic Corporation,
729-731
keyboards, 27, 695,
725-726
Advanced Keyboard, 735-736
capacitive, 729
chipsets, 173
control keys, 735-737

controllers, 739
debouncers, 729
dome-based, 730
electrical functions, 738-739
host interface, 744-745
ports, 739-742
software commands,
742-744
USB port, 750
ergonomic designs, 737-738
function keys, 735-740
layouts, 732
Dvorak-Dealey, 733-734
QWERTY, 732-733
legends, 731-732
linear travel, 730
modular keyboard connector,
749
PS/2-style, 747-749
short-travel, 731
soft-touch, 730
stiff, 731
switches, 727-728
touch, 729-731
transmitter, 694
wireless, 750-751
keycaps, 731-732
keypresses, 694
Kleinschmidt, Edward, 694
Krum, Charles, 694
Krum, Howard, 694
L
lag, digital camera shutters,
797-798
landing zone, read/write
heads, 594-595
1103
1104
LANs
LANs (local area networks),
23, 417-418
Connect module, 162
hardware, 441
cabling, 445-447
hubs, 443-444
NICs, 442-443
routers, 445
Internet model, 422
OSI model, 418-421
standards, 428
Ethernet, 428-431
HomePlug, 438-441
HomePNA, 436-438
431-436
LAPB (Link Access

Procedure, Balanced), 457
LAPM (Link Access
Procedure for Modems),
457
large storage units, 516
Large-scale integration. See
LSI
laser printers, 29, 931-935
laser servo, 628
laser-linked keycaps, 732
latches, 76
latency, 606
bridges, 246
CAS, 217
hard disk mechanism, 588-589
layers, PCI Express,

275-276
layouts
keyboards, 732-734
LBA (logical block addressing) addressing, 306-307

LCC (Leadless Chip
Carrier), 111
LCDs (liquid crystal displays), 28, 834, 852-853

cholesteric technology, 855
color issues, 857
double-scanned displays, 856
nematic technology, 853-855
response time, 856
shutter printers, 932
yields, 856-857
lead-acid cells (batteries),

1030-1032
lead-in area, 668
lead-out area, 668
Leadless Chip Carrier
(LCC), 111
leaf, FireWire, 362
Leclanchs, Georges, 1029
LEDs (light-emitting
diodes), 932
legacy connectors, 875
legacy data, BIOS storage
functions, 199
ACPI tables, 203
data area, 199-200
date, 200
disk parameter tables, 201-202
DOS tables, 202-203
system identification bytes,
200-201
legal issues, capturing

sounds, 904
legends, keyboards,
731-732
lenses, digital cameras, 791
aperture, 791-793
conversion, 795
Letter quality mode, 940

Letterbox mode, 680
level-sensitive interrupts,
168
LFE (low frequency

effects), 899
libraries, 47-48
light meters, 798
light-emitting diodes
(LEDs), 932
lighting effects, 718
Lightning Data Transport.
See HyperTransport
lines
compensation, 454
interfaces, as modem hardware, 463-464
printers, 938-939
rates, 815
width, CRTs (cathode ray
tubes), 842
line-interactive power systems, 1054

linear power supplies, 1014
linear recording tape systems, 634-636
QIC-EX, 637-638
Travan, 636-637
linear topologies (network

architecture), 424-425
linear travel keyboards, 730
liners, floppy disks, 620
lines, 102
lines of memory, 515-516
Link Access Procedure for
Modems. See LAPM
Link Access Procedure,
Balanced. See LAPB
links, Bluetooth, 374-375
Linux, 42-43
liquid crystal displays. See
LCDs
Liquid Crystal Institute of
Kent State University, 855
Lisa (Apple), 695
Lithium batteries,
1035-1036
INDEX
manual focus
lithium solid polymer cells,
1036
lithium-iron disulfide cells,
1036-1037
loaders, 220
local area networks. See
LANs
local loops, 450
Locked mode, 311
logic, 13-14
logic boards, 974
logic circuits, 62-64, 1015
logic design, digital electronics, 74
interrupts, 80
logic gates, 75-76
memory, 76
76-77
programs, 79-80
registers, 77-79
logic gates, digital electronics, 75-76

Logic Link Control layer
(Ethernet Data Link
layer), 429
logical block addressing
(LBA), 306-307, 669-670
logical format, CD-ROMs,
663-664
addressing, 666
capacity, 666
data coding, 664-665
formats, 667
sessions, 665
logical interface
parallel ports, 410
RS-232C ports, 389-390
logical organization, memory, 552-555

LongRun, 107-109

lossless compression systems, 544
lossless linking, 687
Lotus SmartSuite, 37
loudspeakers, 906-907,
918-921
low frequency effects
(LFE), 899
low signals, 63
low-level formatting, 600
low-level languages, 45
low-profile computers, 988,
1063
low-voltage differential signaling (LVDS), 285, 328
low-voltage microprocessors, 141
lower memory (real mode
memory), 553
LPX motherboards, 988
LS-120 Disk Format, 627
LS-120 system, 625
LSI (large-scale integration), 69
luminance, 819
lunchbox computers, 1065
LVD (low-voltage differential), 328
LVDS (low-voltage differential signaling), 285
M
machine languages, 44
magnetic properties, floppy
disks
hubs, 620
liners, 620
magnetic properties, 618-619
shell, 619-620
magnetic storage, 583

formats, 622-625
history, 617-622
mechanisms, 628-629
SuperDisks, 625-628
hard disks, 584
drive arrays, 609-616
FAT file system, 601-606
geometry, 598-600
history, 584-585
mechanisms, 585-595
medium, 595-598
tapes, 630-631
formats, 634-641
streaming, 631-632
magnetic storage (memory), 532-539

coercivity, 537-538
535-536
magnetic materials, 534
permanence, 533-534
saturation, 536-537
magneto-resistive
Magnetophon, 532
main boards, 974
maintenance-free batteries,
1031
management of power,
1042
APM (Advanced Power
Energy Star, 1048
mantissa, 87
monitors, 868-869
manual exposure control,
800
manual focus, 803
1105
1106
mapping
mapping
cache, 102-103
input/output, 54
memory, 54-55
mark, serial signal, 379

mask ROM, 523, 556
masking (memory), 517
mass storage. See secondary
storage (memory)
master file table (MFT),
604
masters, expansion buses,
243-244
mathematics
calculations, processing, 19-21
calculators, 6
Matsushita-Kotobuki
Electronics Industries,
Ltd., 625
maximum cable length,
SCSI, 336
maximum data rates, digital
telephone services, 467
Maximum Performance
Mode (SpeedStep), 107
maximum usable memory,
addressing limits, 550-551
MCM (Multi-Cavity
Module), 111-112, 121
MCNS (Multimedia Cable
Network System Partners,
Ltd.), 474
MDFAT (Microsoft
DoubleSpace FAT), 603
measurements
memory, 514-516
printers, 939
mechanical anti-glare treatments for displays,

844-845
mechanical mouse, 754-755
mechanical switches (keyboards), 727
mechanisms
hard disks, 585
data-transfer rate, 590
latency, 588-589
rotation, 586-587
speed, 587-588
standby mode, 589
media
magnetic storage, 597
optical, 643-644
CD-RWs, 652
Digital Versatile Disc. See
DVDs
drives, 654-659
DVD-Audio, 682-689
history, 643-645
standards, 659-682
printer paper, 948-951
Media Access Control layer

(Ethernet Data Link
layer), 429
medium
floppy disks, 618
hard disks, 595
areal density, 596
contaminants, 597-598
media, 597
substrates, 595-596
megabits, memory chips,

567
megahertz, 94
megapixels, 790
membrane switches (keyboards), 728
memory, 24, 511, 547-548
access, 517
518-519
aperture, 156
attribute, PC cards, 292-293
caching, 100
cards, 575
CompactFlash, 576-577
Memory Stick package
(Sony), 578-579
MultiMedia Cards, 579
Secure Digital cards, 579
SmartMedia, 577-578
xD Picture Cards, 580
chipsets, 156-159
CMOS assignments, 205-208
common memory, 288, 293
computers, 8-10
controller hub, 155
data compression, 543-545
digital electronics, 76
display, 700
errors, 523
detection and prevention,
526-528
hard errors, 525-526
soft errors, 524-525
flip-flop, 76
floppy drives, 24-25
hard disk drives, 24-25
I/O, 54
installation, 572
contact materials, 574
memory speed, 573
mixing modules, 573-574
orientation, 575
logical organization, 552-555
magnetic storage, 532-535
coercivity, 537-538
535-536
magnetic materials, 534
permanence, 533-534
retentivity, 538-539
saturation, 536-537
INDEX
mixing modules
mapping, 54-55, 197
mass storage, 24
measurements, 514-516
module cards, 24, 547
optical drives, 24-26
optical storage, 539-541
packaging, 567-572
PC cards, 288-289
primary storage, 511-512
printers, 942
requirements, 548-551
screen, 700
secondary storage, 512, 528
data organization, 529-532
near-line storage, 528
offline storage, 528
online storage, 528
set-reset memory, 76
solid-state memory, 519-520
read-only memory, 523
read-write memory,
520-522
tape drives, 24-26
technologies, 555
read-only memory),
557-558
EPROM (erasable programmable read-only memory),
557
mask ROM, 556
PROM (programmable
read-only memory),
556-557
read-only memory, 555-556
read-write memory,
559-567
tests, CMOS advanced setup,
210
virtual memory, 541-543
volatility, 513-514
Memory Stick package

(Sony), 578-579
mesh, anti-glare treatment
for displays, 844
meta-languages, 48
metal-in-gap heads, 591
metal-oxide varistors. See
MOVs
Metcalf, Robert, 428
MFM (Modified Frequency
Modulation), 253
MFT (master file table),
604
MH (Modified Huffman)
encoding, 963
Micro Channel
Architecture, 264
micro-ops, 91-92
microATX motherboards,
990-991
Microcom Networking
Protocol Standards Levels
1 through 9, 461
microde microprocessors,
88-90
microfarads, 912
micronets, 372
microphones, 903, 915-917
microprocessors, 17-19,
83-85, 93-94,114-115, 128
486 family, 119-120
ADM, 143-149
arithmetic/logic unit, 86
branch prediction logic, 96
caching, 100-103
clock speed, 94-95
control unit, 86
electrical characteristics,
103-109
EPIC, 98-99
floating-point unit, 87-88
front-side buses, 99-100
influence of PCs on development, 116-118
initial development, 115-116
input/output unit, 85-86
instruction sets, 88
instructions, 76-77
Intel, 128-143
micro-ops, 91-92
microde, 88-90
mobile, 123-124, 137-142
nanoprocessors, 89
out-of-order execution, 97-98
Pentium II, 122-124
Pentium II Celeron, 124-127
Pentium III, 129-133
Pentium 4, 128, 134-136
Pentium family, 120-121
Pentium Pro, 121-122
pipelining, 95-96
register renaming, 98
registers, 77-79
RISC, 90-91
SIMD, 92-93
speculative execution, 96
superscalar architecture,
96-97
testing, CMOS advanced
setup, 210
TLB, 100
Transmeta, 149-151
VLIW, 93
Xeon, 136-137
Microsoft
DoubleSpace FAT (MDFAT),
603
Natural Keyboard, 737
Office, 37
OnNow initiative, 1045
Windows Media Player, 895
midrange speakers, 919

mini-ATX motherboards,
990
Mini-LPXs, 988
mini-tower cases, 1065
Miniature Cards
(CompactFlash), 576
miniature phone plugs, 918
mixed devices, SCSI cable
connections, 334
mixers, 910
mixing modules, memory
installation, 573-574
1107
1108
MMCD
MMCD (Multimedia
Compact Disc), 644
MMR (Modified Modified
Read) encoding, 963
MMX (Multimedia
Extensions), 92, 516
mobile processors,
121-124, 137-148
model byte, 201
Modem Ready indicators,
466
Modem-on-Hold feature,
461
modems, 23, 450-451
background, 450-451
channels, 452-454
connection-enhancing technologies, 454-458
fax, 959-960
hardware, 462-465
modulation, 230
Shannons Limit, 451-452
speed standards, 458-461
telco-return modems, 475
modes, printer speed,

940-941
Modified Frequency
Modulation (MFM), 253
Modified Huffman (MH)
encoding, 963
Modified Modified Read
(MMR) encoding, 963
Modified Read (MR)
encoding, 963
modular keyboard connectors, 749
modulation, 458
Bluetooth, 377
communication channels, 229
amplitude, 233
carriers, 229-232
complex, 236
demodulation, 230
frequency, 233-234
frequency shift keying, 232
modems, 230
phase, 234-235
pulse, 235-236
quadrature, 235
requirements, 230-231
IrDA, 370-371
modulators, 230, 463, 813

modules
LAN Connect, 162
memory module cards, 547
monitors, 28-29, 813,

833-834. See also displays
compared to displays, 833
CRT (cathode ray tube),
834-835
anti-glare treatment,
844-846
aspect ratios, 849
dot-pitch, 841
image sizing, 850
line width, 842
over scans, 848-849
phosphors, 835-836
purity, 839
resolution, 843-844
dot-clocks, 862
electronics, 858-859
bandwidth, 861-863
865-868
identification, 864
interlacing, 861
868-869
range, 859-861
flat-panel display, 834,
852-853
interfaces, 869
Port), 870-874
Port), 869
connectors, 874-879
879-880
UMA (Unified Memory
Architecture), 874
834, 852-857
portrait displays, 848
standards, video, 826-827
video integration, 818-822
video technology, 814-818
monochrome displays, 836

Moore, Gordon E. (Moores
Law), 69
Morse, Samuel, 693
motherboards, 30, 983
ATX, 988-989
backplanes, 974
baseboards, 974
design, 969-970
expansion boards. See expansion boards
FlexATX, 991-992
logic boards, 974
LPX, 988
main boards, 974
microATX, 990-991
mini-ATX, 990
NLX, 988, 992-994
nomenclature, 973-974
physical requirements, 984-987
planar boards, 973
standards, 987-988
system boards, 973
WTX, 988, 994
INDEX
nomenclature
Motion Picture Experts
Group. See MPEGs
Mount Rainier, 674-675
Mount Rainier drives, 618
mouse, 27, 753-758
MOVs (metal-oxide varistors), 1051-1052
MP3 players, 914-915
MPC (Multimedia PC), 658
MPEGs (Motion Picture
Experts Group), 655,
823-826
MR (Modified Read) encoding, 963
Multi-Cavity Module. See
MCM
multi-word DMA transfers,
303-304
multichannel sound, 895-896
multifunction fax printers,
959-961
multilayer circuit boards, 72
multimedia, software, 36
Multimedia Cable Network
System Partners, Ltd.. See
MCNS
MultiMedia Cards, 579
Multimedia Compact Disc
(MMCD), 644
Multimedia Content
Description Interface, 826
Multimedia Extensions. See
MMX
Multimedia PC (MPC), 658
multiple zone recording
(MZR), 587, 599-600
multiple-carrier modems,
455
multiple-image mode, 807
multiplexing
memory chips, 559-560
Interconnect), 266
multipliers, 216
multiprocessors
accumulators, 54
logic gates, 75-76
multitasking, 79
MZR (multiple zone
recording), 587, 599-600
N
N-channel Metal Oxide
Semiconductor. See NMOS
N-type semiconductors, 67
nAckReverse, 409
name servers, 500
names, RS-232C, 389-390
NAND (Not AND), 75
nanoprocessors, 89
narrow SCSI, 342
nAStrobe, 408
National Business
Education Quarterly, 734
National Television
Standards Committee
(NTSC), 819-820
Natural Keyboard, 737
nAutoFd, 404
nDStrobe, 407-408
Near letter quality mode,
940
nematic technology, LCDs
(liquid crystal displays),
853-855
NetBIOS (Network Basic
Input/Output System),
421
network interface cards. See
NICs
Network layer (OSI
model), 420
networks, 417-418
classes, 485-486
connectivity, 21
hardware, 441-445
Internet model, 422
LANs, 23
lnternet. See Internet
OSI model, 418-421
prefixes (CIDR), 486
standards, 428-436
WANs, 23
wireless, 23
New Process technology,

932
nFault, 403
ni-cad cells (batteries),
1032-1034
nibbles, 393, 405-406, 515
nickel, as magnetic material, 534
nickel-cadmiums. See nicad cells
nickel-metal hydrides (batteries), 1034-1035
NICs (network interface
cards), 442-443
NiMH. See nickel-metal
hydrides
nine-pin D-shell jacks, 875
nine-wire serial cable, 386
nInit signal, 408
NLX motherboards, 988,
992-994
NMOS (N-channel Metal
Oxide Semiconductor),
67-68
serial channels, 249
No Return-to-Zero
Inverted (NRZI), serial
channels, 249-250, 359
No-ID Format, 601
noise interference, 1073
noise protection, 1050-1051
nomenclature
1109
1110
non-bussed signals
non-cache coherent (data
transfers), 273
non-contact keyboards,
728-729
nonresident attributes, 605
nonshielded connectors
alternative 1, 338-341
nonvolatile memory, 513,
575
NOR (Not OR), 75
normal mode, 1049
north bridge, 155
Not AND (NAND), 75
Not OR (NOR), 75
notebook cases, 1065-1066
Noyce, Robert N., 68
nozzles (printers), 928-930
nPeriphRequest, 409
nReverseRequest, 409
NRZ (No Return-to-Zero),
249
NRZI (No Return-to-Zero
Inverted), 249-250, 359
nSelectIn, 404
nStrobe, 401
NTFS (Windows NT File
System), 604-606
NTSC (National Television
Standards Committee),
819-820
null modem cables, 388
Num Lock, CMOS
advanced setup, 213
numbers, expansion slots,
985
numeric processors, CMOS
advanced setup, 210
nWait, 408
nWrite, 407-408
Nyquist frequency, 888
Nyquist, Harry, 888
O
O. R. Technology, 625
object code, 47
OCR (optical character
recognition), 958
octets, IP addresses,
482-483
Oersteds, 538
OFDM (orthogonal frequency division multiplexing), 439
Off state (APM), 1044
off-axis sounds, 916
Off-Hook indicators, 466
Office, 37
1053-1054
old technology DIMMs,
569
omni directional microphone, 916
OMTP (over molded thin
package), 577
On signal, 1019
on-axis sounds, 916
onboard ROM, 188
one-pass printing, 935
1054-1058
online storage, 528
OnNow initiative, 1045
OPC (optical photoconductors), 933
open loops, 593
open reels, 631
Open Systems
Interconnection
Reference Model. See OSI
model
open-reels, 634
OpenCable, 287
operating systems, 39-43

operating voltages, microprocessors, 104-106
operations
Port), 871-873
ATA Attachment, 313
CD-Rs, 670-676
PC cards, 288-290
343-344
opposite tracking path

(PTP), 648
Opteron, 149
optical character recognition (OCR), 780-781, 958
optical media, 643-644
CD-RWs, 652
Digital Versatile Discs. See
DVDs
drives, 24-26, 654-655
access time, 658
compatibility, 659
controls, 658-659
DVD-Audio, 682-689
speed, 655-658
standards, 659-682
history, 643-645
optical mouse, 755-757

optical photoconductors
(OPC), 933
optical resolution (scanners), 779
optical storage (memory),
539-541
Optical Storage Technology
Association (OSTA), 678
optical viewfinders, 806
optical zooming, 794
INDEX
PBCC-22
optimizing compilers, 47,
91
option boards, 975
option retaining bracket,
977
Orange Book
CD-R standards, 667-677
CD-RW standards, 677-678
DVD-ROM standards,
678-679
DVD-Video standards,
679-682
orientation, memory installation, 575

orthogonal frequency division multiplexing. See
OFDM
OSI model (Open Systems
Interconnection
Reference Model),
418-421
OSTA (Optical Storage
Technology Association),
678
out-of-order execution,
97-98
output, 697
ports, 22, 173
systems, digital audio,
913-915
over-voltage protection,
1049-1052
overclocking, 216
over molded thin package.
See OMTP
over scans, CRTs, 848-849
P
P cable, SCSI, 336
P-type semiconductors, 66
P5, 120
P54C, 120
packaging
memory, 567-572
modems, 464-465
packed storage units, 516

packet binary convolutionary coding system. See
PBCC-22
Packet Identifier byte, 359
packet interface, AT
Attachment, 308-309
packetized communications, 257
packets
OSI model Data Link layer,
420
TCP/IP, 502-503
USB, 359-361
writing, 674
pad-printed keycaps, 731

paddles, game, 764
page faults, 541
page interleaves, CMOS
advanced setup, 212
page mode, CMOS
advanced setup, 212
page printers, 939
page scanners, 772
page-description language,
955
painting software, 36
Palo Alto Research
Laboratory of Xerox
Corporation, 695
PAM signal, 460
Pan mode, 681
panning, 714-715
paper, 946-954
paper empty parallel port
signal, 403
parallel access arrays, 615
parallel channels, 236-237
parallel ports, 353, 390-391

adapters, 398
cable, 398-399
connectors, 395-398
control, 410-414
byte mode, 406-407
compatibility mode, 400-404
(EPP) mode, 407-408
(ECP) mode, 408-410
nibble mode, 405-406
history, 391-394
logical interface, 410
PS/2, 406
parallel recording, 632

parallel tracking path
(PTP), 648
parameters, passing,
195-196
parity, 210, 238, 526
park-and-lock, 594
Partial Response Maximum
Likelihood. See PRML
pass-through devices, 362
Passed Diagnostics signal,
317
passing parameters,
195-196
passive convection cases,
1070
passive cooling, 104
passive matrix LCDs, 855
passive speakers, 921
passive systems, 921
passwords
AT Attachment, 311
CMOS advanced setup, 214
Patterson, David A., 611

PBCC-22 (packet binary
convolutionary coding
system), 435
1111
1112
PBXs
PBXs (private business
exchanges), 426, 468
PCI (Peripheral
Component
Interconnect), 30,
165-169, 237, 240,
262-264
arbitrated bus-mastering, 270
boards, 998-1000
bridge, 155
burst mode, 267
clock, 161
data integrity signals, 267-268
Express, 274-276
hot-plugging, 271
layers, 275-276
multiplexing, 266
PCI-X, 272-274
serialized interrupts, 268-269
setup, 272
speed, 265-266
standards, 276
PCL (Printer Control

Language), 955
PCM (pulse code modulation), 236, 451
PCMCIA (Personal
Computer Memory Card
International Association),
262, 285-287
cards, 1006-1008
CardBus, 293-294
connectors, 1009
PC cards, 287-288
card interface, 291
operation, 288-290
setup, 291-293
signals, 288-290
software interface, 290-291
slots, 464
standards, 286-287, 295-296
zoomed video, 295
PCR (Phase Change

Recordable), 677
PCs. See also computers
cards
hot-swapping, 286
JEITIA, 286-287
modems, packaging hardware, 464
PCMCIA, 262, 285-296
influence on microprocessor
development, 116
286 family, 117-118
386 family, 118
8086 family, 116-117
peer buses, 246

peer-to-peer systems (networks), 427-428
Pentiums, 120-121
Dynamic Execution, 121
Mobile Pentium II, 123-124
OverDrive, 120
Pentium II, 20, 122-123
Pentium II Celeron, 124-126
Pentium II OverDrive,
127-128
Pentium II Xeon, 126-127
Pentium III, 129-132, 140
Pentium III Xeon, 132-133
Pentium III-M, 140-141
Pentium 4, 128, 134-136
Pentium 4-M, 141
Pentium Pro, 121-122
Processor with MMX, 120
Xeon, 136-137
performance
BIOS storage functions,
214-216, 220
BIOS shadowing, 219
clock speed, 216-217
cycle shaving, 217
system integrity, 217-218
hard disks, 606-609
HyperTransport, 282-283
ISP limits, 504-505
Performance tab
(Windows), 550
PeriphAck, 409
Peripheral Component
Interconnect. See PCI
peripherals
chipsets, 172-173
permanence
magnetic memory storage,
533-534
printers, 945-946
perpendicular recording,
619
PError, 403
Personal Computer
Memory Card
International Association.
See PCMCIA
personal computers. See
PCs
perspective divide, 718
Peterson, David, 90
PFM (pulse frequency modulation), 235
PGA (pin grid array), 110
phase, 851
Phase Alternating Line, 820
Phase Change Recordable
(PCR), 677
phase modulation, 234-235
phase-change inkjet printers, 931
phase-change media, 652
phones, protection from,
1058
phosphors, CRTs (cathode
ray tubes), 835-836
photo detectors, optical
memory storage, 540
photo scanners, 775
photo-resist, 71
photography, digital, 787
physical design, read/write
heads, 590-591
INDEX
ports
physical format
CDs, 653-654
DVDs, 653-654
hard disks, 600-601
Physical layer
Ethernet, 430-431
OSI model, 420
PCI Express, 275
physical requirements,
motherboards, 984
bus standards, 984
expansion slot layout, 986-987
expansion slot numbers, 985
expansion slot spacing,
985-986
piconets, 372
picture elements, 835
Piezo inkjets, 930
piezoelectric crystal, 930
piezoelectric crystal inkjet
printers, 930-931
pin grid array. See PGA
pin jacks, 875-877
pin-in hole technology,
72-73
pincushion distortion,
850-851
PIO (Programmed
Input/Output), 302
pipeline nibble mode
DRAM. See BEDO (Burst
EDO DRAM)
pipelining, 86, 95-96
pipes, 357
pixelization, 696
pixels, 696, 944
Plain UltraDMA, 304
planar boards, 973
Plastic Leaded Chip
Carrier (PLCC), 111
Plastic Quad Flat Package
(PQFP), 111-112
plated-through holes, 72
platens, 947
platters, 585, 589-591

equipment, 586
recording, 586-588
determining size of drive
mechanisms, 595
magnetic field, 591
servo surface, 593
players, MP3, 914-915

PLCC (Plastic Leaded Chip
Carrier), 111
Plug-and-Play, 56, 184-186
boot operation, 187
expansion board support, 186
onboard ROM, 188
USB, 357
plugs, miniature phone

plugs, 918
pocket modems, packaging
hardware, 464
point devices, 695-696
point sticks, 27
point-to-point links,
Bluetooth, 375
point-to-point wiring, 71
pointing devices, 27, 753
drivers, 767-768
game controllers, 764-765
interfaces, 765-767
mouse, 753-759
protocols, 767-768
TouchPads, 761-763
trackballs, 759-761
polarization, anti-glare
treatments for displays,
845-846
port bypass circuits, Fibre
Channel, 346
portable power supplies,
1021-1022
airplane adapters, 1023-1024
external, 1022-1023
portrait displays, 848

ports, 22, 54, 353-354
bidirectional parallel port, 392
Bluetooth. See Bluetooth
chipsets, 173
237-239
compared to interfaces, 237
FireWire, 361-363
cable, 366-367
configuration, 364
connectors, 365-366
daisy-chaining, 362
leaf, 362
signaling, 363
I/O port assignments,
197-198
IrDA, 367-369
data rates, 369
formats, 371
frames, 371
history, 367-368
infrared light, 369
interference suppression, 371
modulation, 370-371
pulse width, 369-370
speed, 370
keyboards, 739-741
handshaking, 742
serial byte protocol, 741
USB port, 750
mouse, 765-766
parallel, 353, 390-391
adapters, 398
cable, 398-399
connectors, 395-398
control, 410-414
history, 391-394
logical interface, 410
PS/2, 406
PS/2 parallel port, 392
RS-232C, 378-379
cables, 385-388
connectors, 380-382
electrical operations, 379
1113
1114
ports
register functions, 389
signals, 382-385
UARTs, 388
serial, 766
USB, 354-355, 766
cable, 358-359
connectors, 357-358
function, 355
hosts, 356
hubs, 355
Plug-and-Play, 357
protocol, 356-361
speed, 356
standards, 361
POST (Power-On Self

Test), 180-182
PostScript 3, 955
power, 30-31, 979, 1011
AC (alternating currents),
1012
capacity, 1027
cell types, 1025
charging, 1038-1040
chemical systems, 1028-1037
currents, 1026
energy, 1027
safety, 1041-1042
shelf life, 1028
1040-1041
storage density, 1027-1028
voltages, 1025-1026
Bluetooth, 377
chipsets, 171
device demands, 1017
management, 1042
APM (Advanced Power
Energy Star, 1048
PC cards, 290
power supplies, 1012-1013
desktop, 1013-1021
fans, 1013
portable, 1021-1024
voltages, 1013
protection, 1048
blackouts, 1053-1058
phones, 1058
power line irregularities,
1048-1052
power bricks, 1022

power calibration area, 668
Power Good signal, 1019
power lines, irregularities,
1048
noise, 1050-1051
over-voltage, 1049-1052
under-voltage, 1050

AT, 1019
ATX, 1019
ATX12V, 1020
desktop, 1013
ATX, 1019
linear, 1014
logic circuits, 1015
ratings, 1015-1017
switching, 1014-1015
voltages, 1015-1018
fans, 1013
PC, 1019
portable, 1021-1022
airplane adapters,
1023-1024
external, 1022-1023
SFX12V, 1020
TFX12V, 1020
voltages, 1013
WTX, 1021
XT, 1019
Power-On Self Test

(POST), 180-182
PowerNow!, 107-109
PPM (pulse position modulation), 236
PQFP (Plastic Quad Flat
Package), 111-112
precompression, 456
presence detect pins (old
technology DIMMs), 569
Presentation layer (OSI
model), 421
prevention, memory errors,
526-528
PRI (Primary Rate
Interface), 472
primary batteries, 1025
primary cache, 101
Primary Rate Interface. See
PRI
primary storage (memory),
511-512
print engines, 924-925
direct thermal, 935-936
dye-diffusion, 937-938
impact dot-matrix, 925-927
inkjet, 927-931
lasers, 931-935
speed, 940
thermal transfer, 936-937

Printer Control Language
(PCL), 955
printers, 29, 923-924, 938
colors, 943-946
command set, 954
consumables, 951-952
cartridges, 952-953
paper, 953-954
refilling, 953
control register, 411
data register, 411
fully formed character printers, 924
interfaces, 954-955
INDEX
publications
line, 938-939
memory, 942
multifunction, 959-961
page, 939
paper handling, 946-947
capacity, 948
collation, 951
duplex operation, 951
friction feeds, 947
media-handling, 948-951
print engines, 924-925
direct thermal, 935-936
dye-diffusion, 937-938
impact dot-matrix, 925-927
inkjet, 927-930
lasers, 931-935
phase-change inkjets, 931
piezoelectric crystal, 930-931
thermal transfer, 936-937
quality, 941-943
speed, 938-939
engine speed, 940
measurement units, 939
modes, 940-941
throughput, 940
status register, 411
printing, one-pass, 935

priorities, buses, 244
private business exchanges.
See PBXs
private networks, addressing, 487-488
592-593
processing, 19
BIOS, 19-21
chipsets, 19-20
microsprocessors. See microprocessors
profiles
40-millimeter, 1020
60-millimeter, 1020
program area, 668

program memory area, 668
programmability (computers), 14
programmable interrupt
controller. See PCI
programmable read-only
memory. See PROM
Programmed Input/Output.
See PIO
programming languages,
34, 43-44
assembly languages, 45
high-level, 44-46
compiled, 46-47
48
interpreted, 46
libraries, 47-48
low-level languages, 45
machine languages, 44
meta-languages, 48
programs, 8, 34-36
applets, 38
databases, 36
digital electronics, 79-80
disassembling, 35
email, 36
Microsoft Office, 37
multimedia, 36
spreadsheets, 36
utilities, 37-38
Web browsers, 36
Word processors, 36
progressive scanning, 861

PROM (programmable
read-only memory),
556-557
propagation delay, 325
properties, floppy disks

hubs, 620
liners, 620
magnetic properties, 618-619
shell, 619-620
protected mode memory,

553
protection, 1048
blackouts, 1053
1053-1054
1054-1058
phones, 1058
power line irregularities, 1048
noise, 1050-1051
over-voltage, 1049-1052
under-voltage, 1050
writing floppy disks, 621
protocols
addressing, DHCP (Dynamic
Host Configuration
Protocol), 488-489
Group 3 fax, 961
HomePlug, 440-441
serial byte protocol (keyboards), 741
split-transaction protocol, 122
TCP/IP, 502-503
USB, 356-361
Windows network model, 423
PS/2, 747-749
mouse ports, 766
parallel ports, 392, 406
PTP (parallel tracking

path), 648
publications
A Case for Redundant Arrays of
Inexpensive Disks, 611
Electronics, 70
National Business Education
Quarterly, 734
1115
1116
pulse code modulation

pulse code modulation
(PCM), 236
pulse frequency modulation
(PFM), 235
pulse modulation, 235-236
pulse position modulation
(PPM), 236
pulse width, IrDA, 369-370
pulse width modulation
(PWM), 236, 294, 906,
1014
pulse-coded modulation.
See PCM signal
purity, CRTs (cathode ray
tubes), 839
pushbuttons, 736
PWM (pulse width modulation), 236, 294, 1014
Q
QAM (quadrate amplitude
modulation), 475
QDR memory (quad datarate memory), 564
QIC-EX, 637-638
QoS (Quality of Service),
474
quad-words (memory measurements), 515
quadraphonic recording,
897
quadrate amplitude modulation (QAM), 475
quadrature modulation,
235, 820
quality
printers, 941-943
Quality of Service (QoS),

474
quantization distortion, 452
quantization levels, 460
QuickConnect, 460-461
QuickStart, 106
QuickTime, 895
QWERTY (keyboard layout), 732-733
R
racks (cases), 1068
radiation, cases, 1072-1075
radio
Bluetooth, 371-373
components, 378
hardware, 375-377
history, 372-373
software, 374-375
standards, 378
technology, 767
radio frequency interference (RFI), 1072-1073

RAID (Redundant Array of
Inexpensive Disks),
610-616
RAM (random access memory). See memory
Rambus (memory addressing), 560
DRAM, 564-566
Inline Memory Modules. See
RIMMs
RAMDACs (random access

memory digital-to-analog
converters), 832
random-access media,
531-532
RAS (Row Address Strobe
signal), 560
raster graphics, 707
raster image processor
(RIP), 934
raster memory, 942
raster scanning, 814-815
rasterizers, 51
Rate Generator mode (8253
164
ratings, 1015-1017
raw capacity, CD-Rs, 669
ray-tracing, 718-719
read-only memory. See
ROM
read-only recording material, 649
read-write memory, 520,
559-560
BEDO (Burst EDO DRAM),
562
dynamic memory, 521-522
EDO (extended data out
memory), 561-562
FPM RAM (fast page-mode
RAM), 560-561
multiplexed memory chips,
559-560
Rambus DRAM, 564-566
SDRAM (synchronous
DRAM), 562-564
static memory, 522
video memory, 566-567
read/write heads, hard disk

mechanism, 590
electrical design, 591-592
head actuators, 593-594
landing zone, 594-595
physical design, 590-591
592-593
reading CDs/DVDs, 657

real mode memory, 552-553
real-time clock chipsets,
162-164
RealAudio, 895
reasonable facsimiles, 957
receive data
indicators, 465
RS-232C ports, 383
serial port signal, 383
recession, 717-718
recordable recording material, 649-652
INDEX
rollers
recordings
recording, 586
recording, 586-587
helical, 633-634
multiple zone recording
(MZR), 587, 599-600
parallel, 632
perpendicular recording, 619
serpentine, 632-633
zoned constant angular velocity (ZCAV), 600
records, secondary storage

sequential data organization, 530
rectifiers, 65
Red Book, 660
Reduced Instruction Set
Computer (RISC), 90-91
redundancy, 610-611
Redundant Array of
Inexpensive Disks. See
RAID
refilling consumables
(printers), 953
reflective disc sides, 649
reflex viewfinders, 806
refreshing, 521
memory
chipsets, 158
concurrent, CMOS advanced
setup, 212
refresh rates, 859
register functions,
RS-232C ports, 389
register-to-register transfers, 273
registered SDRAM, 562
registers, 77-79, 98
regulators, voltage,
1015-1016, 1022
relaxed ordering, 273
relay switches, 522
reliability, 610-611
remapping sectors, 609
Request signal (DMA), 316
request to send (RTS)
RS-232C ports, 384
serial port signal, 384
requirements
memory, 548-551
modulation, 230-231
resolution, 888
CRTs (cathode ray tubes),
843-844
digital audio, 890
digital cameras, 790
Group 3 standards, 961-962
mouse, 758-759
printers, 941
scanners, 779
still cameras, 788
trackballs, 761
two-dimensional graphics,
707-708
Resolution Enhancement
Technology (ReT), 942
resolvers, 500
resource allocation, 56
Resources command (View
menu), 167
response time, LCDs (liquid crystal displays), 856
ReT (Resolution
Enhancement
Technology), 942
retaining brackets, 977
retentivity, magnetic memory storage, 538-539
retracing display systems,
816-817
(RTZ), 249
(RTI) coding, 249
reverse data transfer mode,
406
reverse-spiral dual-layer
(RSDL) discs, 648
rewritable DVD formats,
685
Blu-ray, 688-689
DVD-RAMs, 685-686
DVD-RWs, 686-688
RFI (radio frequency interference), 1072-1073

RI (ring indicator), 384
ribbon cables, 25
Rice, Kellogg, 918
rights management, 892
RIMMs (Rambus Inline
Memory Modules),
568-570
ring indicator (RI),
RS-232C ports, 384
ring topologies (network
architecture), 425, 918
RIP (raster image processor), 934
ripping CDs, 901-902
RISC (Reduced Instruction
Set Computer), 90-91
riser boards, 976, 993,
1002-1003
ACR (Advanced
Communications Riser),
1005-1006
1004
Networking Riser),
1004-1005
RLL (Run Length

Limited), 253
robust mode, HomePlug
transmission, 440
rollers, 933
1117
1118
ROM
ROM (read-only memory),
523, 555-556
character ROM, 701
read-only memory), 557-558
557
mask ROM, 523, 556
PROM (programmable readonly memory), 556-557
shadowing, CMOS advanced
setup, 211-212

Root System Description
Table, 1047
root zone files, 501
rotation, hard disk mechanism, 586-587
routers, 445
routines
addresses, entry points, 196
hardware links to BIOS,
197-198
routing, Internet, 502-503

Row Address Strobe signal
(RAS), 560
RS-232C ports, 378-379
cables, 385-388
connectors, 380-382
signals, 382-385
UARTs, 388
RSDL (reverse-spiral duallayer), 648

RTS (request to send), 384
RTZ (Return-to-Zero coding), 249, 370
rubber dome switches (keyboards), 728
Run Length Limited (RLL),
253
Russell, Bertrand, 75
Rutledge, Joseph D., 763

RXD, 383
RZI (Return-to-Zero
Invert), 370
S
S-video (separate video),
819-821
S/N ratio (signal-to-noise
ratio), 451
S/W layer (PCI Express),
275
safety, batteries, 1041-1042
salable screens, 857
sampling rates, digital
audio, 888-890
sanctioning organizations,
networks
Alliance, 438
HomePlug Powerline
Alliance, 441
WiFi Alliance, 436
SANs (system area networks) InfiniBand. See

InfiniBand
satellites
digital services, 476-477
speakers, 919
saturation, magnetic memory storage, 536-537

SCA (Single Connector
Attachment), 337
SCAM (SCSI Configured
Automagically), 329
scan codes, 739
Scan mode, 681
scanners, 27, 771-772, 776
application interfacing,
782-784
D-max, 778
drum, 772-773
dynamic range, 777-778
electrical interfacing, 782
feature matching, 781

flatbed, 773
hand, 773-774
optical character recognition,
780-781
page, 772
photo, 775
resolution, 779
sheet feeders, 781-782
slide, 775-776
speed, 776-777
transmissive, 775
transparency adapters, 780
video, 775
scanning, 814-815, 861

scatternets (Bluetooth), 376
screens
curvature, CRTs (cathode ray
tubes), 842-843
memory, 700
sizes, CRTs, 846-848
screen-pitch, 840
scroll wheel (mouse), 758
SCSI (Small Computer
System Interface), 238
SCSI Configured
Automagically (SCAM), 329
SCSI parallel interface, 22,
322
addressing, 326-327
ANSI, 344-345
bus width, 326
cables, 336
capacity, 326-327
clocking, 324-326
Common Command Set, 323
connectors, 337-341
operation, 343-344
SCSI-2 standards, 323
SCSI-3 connector, 340
SCSI-3 standards, 323-324
signaling, 327-328
standards, 322-324,
341-343
wiring, 328-336
INDEX
shading, three-dimensional graphics

SDDS (Sony Dynamic
Digital Sound), 681, 900
SDR (single data-rate
memory), 563
DIMMs, 570
memory, 563
SDRAM (synchronous
DRAM), 562-563
CAS latency, 563
DDR memory (double datarate memory), 563-564
QDR memory (quad data-rate
memory), 564
registered SDRAM, 562
SDR (single data-rate memory), 563
unbuffered SDRAM, 563
SDS 56 (Switched Data

Services 56), 471-472
SDSL (Symmetrical Digital
SEC (Single Edge Contact)
cartridge, 112-113
SECAM (Sequence
Couleur Memoire), 820
secondary batteries, 1025
secondary cache, 101
secondary storage (memory), 512, 528
data organization, 529-532
online storage, 528
sectored-erase Flash memory, 558

sectors
boot, 184
hard disks, 599-600
remapping, 609
Secure Digital cards, 579

security
Providers), 505-506
seek time, 606

select input, 404
select signal, parallel ports,
402-403
Selker, Ted, 763
Send Data indicator, 465
sensing digital cameras,
798-799
sensitivity, imaging sensors,
791
sensors, 765
SEP (Singe-Edge
Processor), 112-113
separate sync, 816
separate video (S-video),
821
Sequence Couleur
Memoire (SECAM), 820
sequential media, secondary storage data organization, 529-531
Serial ATA, 320-322
serial byte protocol (keyboards), 741
serial channels, 246
bit-coding, 248-249
clocking, 247-248
No Return-to-Zero (NRZ)
coding, 249
(NRZI) coding, 249-250
(RTZ), 249
(RTI) coding, 249
Serial Infrared Interaction

Pulse (SIP), 371
serial interfaces, FireWire,
361-363
cable, 366-367
configuration, 364
connectors, 365-366
daisy-chaining, 362
leaf, 362
signaling, 363
serial modems, computer

interfaces, 462
serial ports, RS-232C,
378-379, 766
cables, 385-388
connectors, 380-382
signals, 382-385
UARTs, 388
serial presence detect pins

(old technology DIMMs),
569
serialized interrupts, PCI
Serialized IRQ Support,
268
serpentine recording,
632-633
servers, blade, 973
services
card, PC cards, 291
socket, PC cards, 291
Windows network model, 423
servo patterns, 599

servo surface (platters), 593
servo-voice coil actuator,
594
Session layer (OSI model),
421
sessions, CD-ROMs, 665
set-reset memory, 76
setup
PC cards, 291-293
Interconnect), 272
SFX12V, 1020
shading, three-dimensional
graphics, 719
1119
1120
shadow masks
shadow masks, CRTs (cathode ray tubes), 839-840
shadow memory, 554-555
shadowing
BIOS, 219
ROM, CMOS advanced
setup, 211-212
Shannon, Claude L.,

451-452
shared memory architecture. See UMA (Unified
Memory Architecture)
sharing, 54
sheet feeders (scanners),
781-782, 946
shelf lives, batteries, 1028
shell floppy disks, 619-620
shield, audio connector, 918
shielded connector alternative 2, 338
shielding, twisted-pair
wires, 447
shifts, 377
Shockley, William, 66
Sholes, Christopher, 732
short slots, 1069
short-travel keyboards, 731
Shugart Associates, 617
shutdown inverter, 1057
shutters, digital cameras,
796-798
sideband signals, 872
signal-to-noise ratio. See
S/N ratio
signals
ATX design, 1019
fault, 403
FireWire, 363
ground, RS-232C ports, 384
HostAck, 409
HostClk, 409
initialize printer, 403-404
interference, 1073
nAckReverse, 409
nAutoFd, 404
nPeriphRequest, 409
nReverseRequest, 409
PC cards, 288-290
PeriphAck, 409
PeriphClk, 409
PError, 403
RS-232C ports, 382-385
327-328, 342
select input, 404
select parallel ports, 402-403
steering, 898
SIMD (Single Instruction,

Multiple Data), 92-93
SIMMs (Single Inline
Memory Modules), 568
Singe-Edge Processor
(SEP), 112-113
Single Connector
Attachment (SCA), 337
single data-rate memory.
See SDR memory
Single Edge Contact (SEC)
cartridge, 112-113
Single Inline Memory
Modules (SIMMs), 568
Single Inline Pin Packages
(SIPPs), 568
Single Instruction, Multiple
Data (SIMD), 92-93
single-board computers,
970-971
single-density recording,
253
single-ended SCSI, 327-328
single-image mode, view
panels, 807
single-layer discs, 661
single-word DMA transfers,
303
SIP (Serial Infrared
Interaction Pulse), 371
SIPPs (Single Inline Pin
Packages), 568
sizes
cache, 101
cases, 1063-1066
CRT images, 850
imaging sensors, 790
Sleep mode, AT
Attachment power management, 309-310
slide scanners, 775-776
slideshows, view panels, 807
slots, expansion
cases, 1069
layout, 986-987
numbers, 985
spacing, 985-986
Small Computer System

Interface. See SCSI
small footprint computers,
1063
small form-factor designs,
1020
Small Outline DIMMs. See
SoDIMMs
1040-1041
SmartMedia memory cards,
577-579
SMBus clock, 162
SMM (system management
mode), 106
smooth shading, 719
socket services, PC cards,
291
SoDIMMs (Small Outline
DIMMs), 568-572
soft errors, memory,
524-525
Soft Off mode (ACPI), 1046
soft-touch keyboards, 730
software, 33-36
applets, 38
Bluetooth, 374-375
INDEX
SSE
commands, keyboard,
742-744
databases, 36
disassembling programs, 35
email, 36
flow control, 385
interface, PC cards, 290-291
interrupts, 80, 191, 195
multimedia, 36
spreadsheets, 36
utilities, 37-38
Web browsers, 36
Word processors, 36
Software Retriggerable
Strobe mode (8253
165
solder-ball grid array, 112
solid inkjet printers, 931
solid-state electronics, 66
Solid-State Floppy Disk
Cards. See SSFDCs
solid-state memory,
519-520
read-only memory, 523
read-write memory, 520-522
Sony Corporation
Dynamic Digital Sound
(SDDS), 681, 900
floppy disks, 617
Memory Stick package,
578-579
sound, 884-885
analog audio, 885
amplitude, 885-886
decibels, 886-887
distortion, 887-888
frequencies, 885
voltage, 886
digital audio, 888
bandwidth, 890-891
capturing, 901-904
resolution, 890
transmission, 891
sound pressure level (SPL),

885
sound systems, 904-905,
908-909
amplifiers, 906
control, 910-911
drivers, 907-908
pulse-width modulation, 906
quality, 911-913
transducers, 915
microphones, 915-918
soundboards, 885, 908-909

control, 910-911
quality, 911-913
source code, 47
Source Manager, Twain,
783
Source-Synchronous
Transfer, 134
Space, serial signal, 379
spacing
screen-pitch, 840
speakers
active systems, 921
bass reflec, 920
cones, 918
drivers, 908
ducted, 920
full-range speaker system, 920
loudspeakers. See
loudspeakers
midrange, 919
passive systems, 921
satellite, 920-921
satellite speakers, 919
three-way, 919
tuned-port, 920
two-way, 919
USB, 913-914
Windows, 908
specifications, backup
power devices, 1055-1056
speculative execution, 96
speed
CD-Rs, 671-672
clock, 94-95, 216-217
clocking, SCSI, 342-343
DDR memory (double datarate memory), 563
digital camera shutters,
796-797
floppy disks, 629
hard disk mechanism,
587-588
IrDA, 370
printers, 938-941
scanners, 776-777
USB, 356
speedlight, 801
SpeedStep, 107-108
SPGA (staggered pin grid
array), 111
spikes, 1049
Spindle Sync/Cable Select
signal, 317
spindles, 585, 628
SPL (sound pressure level),
885
split transactions
Interconnect), 274
protocol, 122
spreadsheets, 36
sprites, 713
Square Wave mode (8253
165
SSE (Streaming SIMD
Extensions), 92
1121
1122
SSFDCs
SSFDCs (Solid-State
Floppy Disk Cards), 577
stack machines, 87
staggered pin grid array
(SPGA), 110
stampers, 646
Standard Parallel Port, 394
analog services, 458-461
AT Attachment, 300
Bluetooth, 378
cable digital services, 474-476
cases, 1062-1063
CD-DAs, 660-661
CD-ROMs, 663-667
CD-Rs, 667-669
logical blocks, 669-670
operation, 670-677
CD-RWs, 677-678
CDs, 659-660
communication channel limitations, 228-229
display systems, 818-821,
826-827
DVD-ROMs, 678-679
DVD-Video, 679-682
DVDs, 659-660
group faxes, 961-964
HyperTransport, 285
InfiniBand Architecture, 281
JEITIA (Japan Electronics and
Information Technology
Industries Association),
286-287
ATX, 988-989
FlexATX, 991-992
microATX, 990-991
mini-ATX, 990
NLX, 992-994
WTX, 994
networks, 428
Ethernet, 428-431
HomePlug, 438-441
HomePNA, 436-438
431-436
Interconnect), 276
PCMCIA PC cards, 286-287,
295-296
322-324, 341-343
Super Audio CDs, 661-663
USB, 361
Standby mode
Attachment power management, 309-310
hard disk mechanism, 589
Standby state (APM), 1043

Standby Supply Voltage signal, 1019
star topologies (network
architecture), 425-426
Star workstation (Xerox),
695
start bits, 254
start-stop mode, tapes, 631
states
ACPI, 1046-1047
APM (Advanced Power
static memory, 522

status bytes, 739
stereo, 896
stereoscopic displays, 715
stiff keyboards, 731
still cameras, 788
stop bits, 254
storage
densities, batteries, 1027-1028
functions, BIOS, 199
legacy data, 199-203
performance tuning,
214-220
system configuration,
203-214
image, digital cameras, 805

magnetic, 583
hard disks, 584-616
tapes, 630-641
straight-through cables,
RS-232C ports, 385-386
Streaming SIMD
Extensions (SSE), 92
streaming tapes, 631-632
strobe lines, 363, 401
strokers, 707
structure
APM (Advanced Power
Management), 1044
HyperTransport, 283-285
279-280
sub-action gap, 364

subchannels, 665
subdomains (DNS), 499
submodel byte, 201
subpictures, 680
subroutines, 45
subscriber IDs, 475
substrates, 595-596, 977
subwoofers, 919-921
suites, 37
Super Audio CDs, 661-663
super-pipelining, 95
super-pixels, 944
super-voltage, 558
SuperDisks, 625-628
superscalar architecture,
96-97
supertwist nematic, 854
SuperVGA connector
pinouts, 878
supplemental lenses, digital
cameras, 795-796
supply voltages, 1018
support chips, 154
support circuits, 153-154
INDEX
telecommunications
surface mount technology,
printed circuits, 73
surge protectors, MOVs,
1051-1052
surge suppressors, 31
Suspend state (APM), 1043
swap files, 542-543
switch bounce, 738
Switched Data Services 56.
See SDS 56
switches
keyboards, 727-728
modems, 453
network hubs, 444
trackballs, 759-760
symbolic logic, 75
Symmetrical Digital
SDSL
Sync Field, 359
synch-on-green, 816
synchronization
frequency ranges, 859-861
signals, 815-816
synchronous buses, 243

synchronous DRAM. See
SDRAM
synchronous SCSI, 325-326
synthesis, 884
synthesizers, 908
system area networks
(SANs). See InfiniBand
system boards, 973
system buses, 99-100
system configuration, BIOS
storage functions,
203-214
system control
chipsets, 159
clocks, 161-164
DMA (Direct Memory
Access), 169-171
PCI (programmable interrupt controller), 165-169
power management, 171
timers, 164-165
timing circuits, 159-160
system identification bytes,

BIOS legacy data,
200-201
system integrity, performance tuning, 217-218
system management mode
(SMM), 106
system resource memory,
549-550
system-level errors, 525
system-specific standards,
PC cards, 292
system.ini file, 551
T
T-cal, 595
T1 high-speed telephone
service, 468-469
T3 telephone service, 469
tables
ACPI, 203
disk parameter, 201-202
disk parameter table, 598
DOS, 202-203
Fixed ACPI Description
Table, 1047
Root System Description
Table, 1047
tactile feedback, 729

Taiyo Yuden CD-R discs,
650
Tanner. See Pentium III
Xeon
Tape Carrier Package, 112
tapes, 630-631
drives, 24-26
formats, 634
helical-scan recording,
638-641
linear recording, 634-638
streaming, 631-632
systems, 897
Target Channel Adapter

(TCA), 280
Task File, 308
Taylor, Bob, 481
TCA (Target Channel
Adapter), 280
TCP/IP (Transport
Control Protocol/Internet
Protocol), 502-503
TDMA technology (time
division multiple access
technology), 475
technologies
faxes, 958-959
keyboards, 726-727
memory, 555
read-only memory),
557-558
557
mask ROM, 556
PROM (programmable
read-only memory),
556-557
read-only memory, 555-556
read-write memory,
559-567
telco-return modems, 475

telecommunications, 21-23,
449-450
analog services
modems, 450-458, 462-466
standards, 458-461
1123
1124
telecommunications
cable services, 473-476
satellite services, 476-477
telephone services, 467-473
telephone services
DSL, 469-471
ISDN, 472-473
maximum data rates, 467
SDS 56, 471-472
T1, 468-469
VDSL, 471
protection from, 1058
Telephoto converters, 795

teleprinters, 957
teletypes, 693-694
BIOS support, 698
character technologies, 699-704
color coding, 710-715
graphic input devices, 696
HTML, 698-699
movement to terminals,
697-698
output, 697
three-dimensional graphics,
715
718-720
depth effects, 717
recession, 717-718
two-dimensional graphics, 705
bitmapped graphics, 706-707
block graphics, 705-706
color planes, 708-710
graphic attributes, 708
resolution, 707-708
vector graphics, 707
televisions, 818
Federal Communications
Commission, 818
interference, 1073
invention, 814
temperature cutoff, 1040

terminals
dumb, 698
movement from teletypes,
697-698
Terminal Ready indicators,
466
terminations
cable, SCSI, 334-336
codes, 964
floppy disks, 349
pin jacks, 876-877
tertiary cache, 101

testing
CD-Rs, 676
memory, CMOS advanced
setup, 210
numeric processors, CMOS
advanced setup, 210
TFT (Thin Film
Transistor) technology,
855
TFX12V, 1020
thermal calibration (T-cal),
595
thermal compensation,
594-595
thermal constraints, microprocessors, 103-104
thermal dye-sublimation,
937
thermal inkjet printers, 930
thermal printers, 935-937
thermal wax transfer printers, 936
Thin Film Transistor
(TFT) technology, 855
thin form-factor design
(power supplies), 1020
thin-film heads, 591
ThinkJet, 930
three-color printers, 944
715
718-720
depth effects, 717
recession, 717-718
three-pass machines (scanners), 776

three-way speaker systems,
919
throughput, 590, 940
thumbnail mode, 807
time division multiple
access technology
(TDMA), 475
timers, chipsets, 164-165
timing
circuits, 159-160
ports, 239
tip, audio connector, 918

TLB (translation look-aside
buffer), 100
TMDS (Transition
Minimized Differential
Signaling) signals, 880
token packets, USB,
359-360
tool-free entry (cases), 1067
top-level domains (DNS),
489, 500
country codes, 490-498
organizational versus national,
489
topologies (networks),
375-376, 424-426
touchpads, 27, 729-731,
761-763
towers, 1064-1065
traces, 718
INDEX
Universal ADSL Working Group

track multisession mode
(writing to CDs), 672-673
track pitch (CDs), 653
track-at-once method
(writing to CDs), 672
trackballs, 27, 759-761
tracks, hard disks, 598
traditional parallel ports,
411-412
transaction byte count, 273
Transaction layer (PCI
Express), 275
transcience, 524
transducers, 765, 915
microphones, 915-918
transfer modes, AT
Attachment, 301-303
transfer rollers, 933
transformers. See external
power supplies
transients, 1049
transistor-transistor logic
(TTL), 64, 1015
transistors, 66, 69-70, 522
Transition Minimized
Differential Signaling
(TMDS) signals, 880
translation look-aside
buffer (TLB), 100
Transmeta, 109, 149-151
transmission methods
digital audio, 891
HomePlug, 439-440

transmissive scanners, 775
transmit data, 383
transmitters, 694
transparency
719-720
adapters, scanners, 780
Transport Control
Protocol/Internet
Protocol (TCP/IP),
502-503
Transport layer (OSI
model), 421
trapezoidal distortion, 851
Travan technology, 636-637
tree topologies (network
architecture), 426
trickle chargers, 1040
trilinear filtering, 717
True Color systems, 709
TTL (transistor-transistor
logic), 64, 1015
tuned-port speakers, 920
tuning BIOS storage functions, 214-216, 220
BIOS shadowing, 219
clock speed, 216-217
cycle shaving, 217
system integrity, 217-218
Tut Systems HomeRun,

437
Twain, 783-784
tweeters, 919
twisted nematic (TN) technology, 854
twisted-pair wires, 445-447
two-color systems, 708
two-dimensional graphics,
705
bitmapped graphics, 706-707
block graphics, 705-706
color planes, 708-710
graphic attributes, 708
resolution, 707-708
two-way exchanges of
information. See duplex
communications
two-way speaker systems,
919
typematic rate, CMOS

advanced setup, 213
typewriting behavior, 733
U
UART (Universal
Asynchronous
Receiver/Transmitter),
173, 388
UDF (Universal Data
Format), 678
UHF (Ultra High
Frequency), 376
ULSI (ultra large-scale
integration), 69
ultra-low-voltage microprocessors, 141-142
Ultra160 SCSI, 325
Ultra320 SCSI, 325
UltraDMA modes, 304
UMA (Unified Memory
Architecture), 554, 828,
874
UN-II (Unlicensed
National Information
Infrastructure), 434
unbuffered SDRAM, 563
under-voltage, protection
from, 1050
underrun, 675-676
under scans, CRTs,
848-849
unidirectional microphones, 916
Unified Memory
Architecture (UMA), 554,
828, 874
Uniform Resource Locators
(URLs), 499-500
Univac, 7
Universal ADSL Working
Group, 470
1125
1126
Universal Asynchronous Receiver/Transmitter

Universal Asynchronous
Receiver/Transmitter
(UART), 173, 388
Universal Data Format
(UDF), 678
universal power supplies
(UPSs), 1018
Universal Serial Bus. See
USB
Unix, 41-42
Unlicensed National
Information Infrastructure
(UN-II), 434
Unlocked mode, AT
Attachment security, 311
unshielded twisted pairs
(UTPs), 446
upgrades, BIOS, 220-221
upper memory (real mode
memory), 553
UPSs (universal power supplies), 1018
capacity, 1057-1058
interfaces, 1056-1057

Locators), 499-500
USB (Universal Serial Bus),
22, 353-355
cable, 358-359
clock, 162
connectors, 357-358
function, 355
hosts, 356
hubs, 355
NICs (network interface
cards), 442
Plug-and-Play, 357
ports, 750, 766
protocol, 356-361
speakers, 913-914
speed, 356
standards, 361
very low frequency (VLF),

1074
Very-high-data-rate Digital
Subscriber Line (VDSL),
471
VGA (Video Graphics
Array), 708, 877-878
video
V.42 modem standard, 461

V.42bis modem standard,
461
V.44 modem standard, 461
V.92 modem standard,
459-461
V.PCM upstream, 460
VA (volt-amperes), 1055
vacuum tubes, 65
valves, invention of, 7
variable-speed drives (optical media drives), 656
Vaught, John, 930
vcache, 551
VDSL (Very-high-data-rate
471
VDT (video data terminal),
694
vectors, interrupt, 191-195
version 1.0 (HomePNA),
437
version 2.0 (HomePNA),
438
vertical frequency, display
systems, 815
vertical synchronizing frequency, 859
very large-scale integration
(VLSI), 69
Very Long Instruction
Words (VLIW), 93
video data terminal (VDT),

694, 735
Video Graphics Array
(VGA), 708
video random access memory (VRAM), 567
video-frame grabbers, 822
View menu commands,
Resources, 167
user interfaces, 35, 463

USLI (ultra large-scale
integration), 69
utilities, 34, 37-38
UTPs (unshielded twisted
pairs), 446
amplifiers, 858
attributes, 702-704
boards, 819
cameras, 788-789
capture, 822
composite, 816
controllers, 828-832
display systems, 814
back porches, 817
blanking, 817
front porches, 817
retracing, 816-817
scanning, 814-815
standards, 826-827
synchronizing signals,
815-816
input, 934
integration, 818-822
interrupts, 698
memory, 566-567
mode flag, 701
overlay, 818-822
pages, 704
scanners, 775
viewfinders, 806
Zoomed Video, 287, 295
INDEX
woofers
view panels (digital cameras), 807-808
viewfinders (digital cameras), 806
virtual memory, 541-543
viruses
ISP security, 505-506
protection, 213
VITS (vertical interval test

signal), 818
VLF (very low frequency),
1074
VLIW (Very Long
Instruction Words), 93
VLSI (very large-scale integration), 69
VMM (virtual memory
manager), 542
voice and data combination,
457-458
voice-coils, 918
VoiceSpan, 457
VoiceView, 457
volatile memory, 513-514
volt-amperes (VA), 1055
voltages, 1013-1017
audio systems, 886
1039
cutoff voltage, 1039
over-voltage, protection from,
1049-1052
regulators, 1015-1016, 1022
supply, 1018
under-voltage, protection
from, 1050
voltage and temperature cutoff (VTCO), 1040
von Behren, Robert, 634

VRAM (video random
access memory), 567
VTCO (voltage and temperature cutoff), 1040
W
wafers, 68
wait states, CMOS
advanced setup, 211
wake-up calls (APM), 1044
WANs (wide area networks), 23
warm boots, 179-180
Watson, Thomas J., 763
Watt, James, 1027
watts, 1027
wave-soldering, 72
Web. See WWW (World
Wide Web)
weighting, 799
wheels (mouse), 758
white balance, digital cameras, 804
wide area networks
(WANs), 23
Wide SCSI, 238, 326
wide-angle converters, 796
Widescreen mode, 681
width, 851
WiFi (Wireless Fidelity),
432, 436
Willamette. See Pentium 4
Winchester disk drive, 585
windowing, 714
Windows
minimum memory requirements, 548-549
Performance tab, 550
speaker driver, 908
Windows
Windows
Wizard,
Windows
9x, 40-41
Add Hardware
52
Device Manager
DMA (Direct Memory

Access), 170
I/O port usage, 199
interrupt assignments,
167-168
Windows Internet Name

System (WINS), 502
Windows Media Player, 895
Windows model (networks), 422-423
Windows NT, 41
File System. See NTFS
Windows Select Network

Component Type dialog
box, 422
Name System), 502
wireframes, 722
Wireless Ethernet
Compatibility Alliance,
433
Wireless Fidelity (WiFi),
432, 436
wireless keyboards,
750-751
wireless networks, 23,
431-432
802.11a, 433-434
802.11b, 432-433
802.11b+, 435
802.11g, 435-436
dual-standard systems, 436
WiFi Alliance, 436
wireless pointing devices,

766-767
wiring
hand-wired, 71
hard-wired, 88
local loops, 450
point-to-point, 71
cable connections, 332-334
cable terminations, 334-336
identification, 329-332
maximum cable length, 336
Wittgenstein, Ludwig, 75
wizards, Windows Add
Hardware Wizard, 52
woofers, 919
1127
1128
Word processors
Word processors, 36
words (package of bits), 515
working memory. See primary storage (memory)
World Wide Web. See
WWW
write-back caches, 102
write-once DVD formats,
684-685
write-through caches, 102
writing protection, floppy
disks, 621
WTX motherboards, 988,
994
WTX power supplies, 1021
WWW (World Wide Web),
23, 479
addressing, 482
CIDR, 486
DHCP (Dynamic Host
488-489
IPv6, 487
browsers, 36
cameras, 789
content, 481
DNS (Domain Name
System), 480, 489, 499-500
subdomains, 499
top-level domains, 489-500
Locators), 499
history, 481-482
Providers), 504-506
routing, 502-503
Web pages, 503

Name System), 502
X
x2 standard, 461
xD Picture Cards, 580
Xerox Star workstation, 695
Xircom Incorporated, 393
XLR connectors, 918
XT power supplies, 1019
Y
Yamaji, Keizo, 932
Yasuda, Hiroshi, 824
Yellow Book CD-ROM formats, 667
yields, 856-857
Z
z-buffering, 710, 719
ZCAV (zoned constant
angular velocity) recording, 600
zinc-air cells, 1037
zip disks, 624-625
zoned constant angular
velocity (ZCAV) recording, 600
zoom digital cameras,
794-795
zoom lenses, 793
zoom view panels, 807-808
Zoomed Video, 287, 295

Winn L. Rosch - Hardware Bible

Uploaded by

Copyright:

Available Formats

Winn L. Rosch - Hardware Bible

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Winn L. Rosch - Hardware Bible

Uploaded by

Copyright:

Available Formats

WINN L.

Winn L. Rosch Hardware Bible,

Copyright 2003 by Que Publishing

All rights reserved. No part of this book shall be reproduced, stored

International Standard Book Number: 0-7897-2859-1

Warning and Disclaimer

Sharry Lee Gregory

Functions and Components 17

Channels and Coding

12 Local Area Networking

15 Principles: Freezing Time

Computers in History ..........................5

WINN L. ROSCH HARDWARE BIBLE, SIXTH EDITION

Electronic Design ..............................62

Register Renaming ..................98

WINN L. ROSCH HARDWARE BIBLE, SIXTH EDITION

Signals and Operation ............313

Universal Serial Bus ..........................354

Analog Services ................................450

15 Principles: Freezing Time

Defining Memory ............................511

WINN L. ROSCH HARDWARE BIBLE, SIXTH EDITION

Hard Disks ........................................584

18 Optical Storage (CD and DVD)

5 Human Interface 691

The Big Three ..................................788

WINN L. ROSCH HARDWARE BIBLE, SIXTH EDITION

Video Technology ............................814

Computer Display Standards ..........826

Print Engines ....................................924

Phase-Change Inkjets ............931

Design Approaches ..........................969

PCI Boards ........................................998

Power Management ........................1042

Early Commercial Chipsets

Legacy Labeling Standards

BIOS Performance Penalties

Microprocessor Operating Modes

Microprocessors of the Past

Plug and Play Step-by-Step

Others Networking Standards

Data Packet History

Setting Up a Small Network

Using Network Software

Expansion Bus History

Industry Standard Architecture In-depth

Alternate Drive Technologies

ATA Life Story

Early SCSI Host Adapter Addressing

Legacy Memory Issues

Less Popular Disk Interfaces

SCSI Operating System Support

Choosing and Using a Backup System

Digital Linear Tape

General-Purpose Interface Bus

Disk Capacity Limits

How Hardware Flow Control Works

DriveSpace File Compression

Dual-Actuator Hard Disks

Parallel Port Performance Issues