Library of Congress Cataloging in Publication Data

Hewlett-Packard Company.
Optoelectronics Division.
Applications Engineering Staff.
Optoelectronics applications manual.
Includes index.
1. Light emitting diodes.
I. Gage, Stan.
II. Title.
TK7871.89.L53H4819i7
621.3815'42
77-5529
ISBN 0-07-028605-1

Copyright 1977 by Hewlett-Packard

Com:>any. All rights reserved. Printed in the United States of
America. No part of this publication may be rt~produced, stored in a retrieval system, or transmitted, in
any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the
prior written permission of Hewlett-Packard.

Hewlett-Packard
assumes no responsibility for the use of any circuits described
herein and makes no representations
or warranties, express or implied, that such
circuits are free from patent infringement.

1.1

1.2
1.3
1.4

1.5

2.1

2.2

2.3

The Theory of P-N Junction

Electrolumineseence.

1.1.1
Semiconductor Energy Gap
1.1.2
Semiconductor Doping
1.1.3
The P-N Junction
1.1.4
Recombination
1.1.5
Materials Available for LED Devices
1.1.6
Direct and Indirect Band-Gap Materials
1.1.7
Enhanced Photon Emission in Indirect G,lp Materials
Quantum Efficiency of LED Devices
Relative Efficiency . .
Material Processing .
1.4.1
LED Structure
1.4.2
Transparent vs. Opaque Substrate
The Effect of Temperature Variation on LED Par,imeters
1.5.1
Forward Voltage as a Function of TempE!rature
1.5.2
Change in Peak Wavelength as a Function of Temperature.
1.5.3
Change in Output Power vs. Temperature
. . . . . .

Physical
2.1.1
2.1.2
2.1.3
2.1.4
2.1.5
2.1.6
2.1.7

2.1.8
2.1.9
LED
2.2.1
2.2.2
2.2.3
2.2.4
2.2.5
2.2.6
LED
2.3.1
2.3.2
2.3.3
2.3.4

Properties of an LED Lamp Device

Plastic Encapsulated LED Lamp
Fresnel Loss . .
Critical Angle Loss . . . .
Optical Efficiency
. . . .
External Quantum Efficiency
Internal Quantum Efficiency
Calculating Radiated Flux
2.1.7.1
Luminous Efficacy and PowE'r Per Unit Solid Angle.
2.1.7.2
Calculating Total Power
Magnification and Luminous Intensity
Diffused and Undiffused LED Lamps.
Lamp Packaging . . . . . . . . .
Lead Frame Packaging.
. . . . .
The Industry Standard T-1 3/4 and T-1 LED Lamps.
The Subminiature LED Lamp
The Rectangular LED Lamp.
. .
The Hermetic LED Lamp.
. . .
LED Lamps that Include Other Components
Lamp Characterization Information
. . . .
Light Output and Color Matching.
. . .
Maximum Temperature Derated Operating Limits.
Pulsed Operating Conditions.
.
Time Average Luminous Intensity.
. . . . .

2.1
2.1
2.1
2.2
2.3

2.3
2.3
2.3
2.3
2.3

2.5
2.6
2.6
2.6
2.7
2.8

2.8
2.9
2.9
2.9
2.10
2.11
2.13
2.14

2.4

Visual Applications of LED Lamps

2.4.1
Introduction...

2.14
2.14

2.4.2
2.4.3

2.16
2.18
2.18
2.19
2.19
2.20
2.21
2.24
2.24
2.25
2.28
2.30
2.35
2.36
2.39
2.42
2.42
2.42
2.44

2.4.4

2.4.5

2.5

3.1

3.2

3.3

Relative Efficiency . . .
Driving an LED Lamp.
.
2.4.3.1
LED Electrical Characteristics
2.4.3.2
Resistive Current Limiting
2.4.3.3
Constant Current Limiting
2.4.3.4
LED-Logic Interface
2.4.3.5
Worst Case Design
LED Arrays
. . . . .
2.4.4.1
Introduction..
2.4.4.2
Designing an X-V Addressable LED Array
2.4.4.3
Design of a Microprocesior Controlled LED Array
2.4.4.4
Analog Bar Graph Arrays.
. . . . .
Backlighting
. . . . . . . . . . . . . .
2.4.5.1
Fundamental Backlighting Requirements
2.4.5.2
ON Sterance Design Considerations
2.4.5.3
Backlighting Construction for Small-to-Medium Legend Areas
2.4.5.4
Backlighting Construction for Very Large Legend Areas

Communications and Signalling Applications

. . . . . .
2.5.1
Device Characterization for Communications and Signalling
2.5.2
Flux Properties in Signalling .
2.5.3
Lens System with LEOs . .
2.5.4
No-Lens Signalling . . . .
2.5.5
Signalling Over Long Distance
2.5.6
Film Exposure
. . . .

Optoisolator Theory
3.1.1
Photo Emitter.
3.1.2
Optical Medium
3.1.3
Photodetector
3.1.4
Amplifier Options
Parameter Characterization
3.2.1
Isolation
3.2.2
Insulation...
3.2.3
Speed of Response
3.2.4
Reverse Coupling.
3.2.5
CTR (or Gain)
Isolator
3.3.1
3.3.2
3.3.3
3.3.4
3.3.5

Packaging
Packaging of Plastic 01 P Isolators
Packaging of High Reliability Isolator:;
Compatibility of Six and Eight Pin Isolators.
Layout Considerations for Optically Coupled Isolators
Bypass Capacitor Requirements.
. . . . . . .

2.44
2.45
2.46
2.51
2.55

3.1
3.1
3.1
3.2
3.4
3.5

3.5
3.9
3.10
3.14
3.14
3.15
3.15
3.15
3.15
3.15
3.17
3.17

3.4

3.5

3.6

CTR Degradation. . . . . .
3.4.1
Introduction....
3.4.2
Cause of CTR Degradation
3.4.3
Worst Case Circuit Design to Allow for Cl'R Degradation
3.4.4
CTR "Degradation Proof" Isolator Circuit Design Techniques
Analog Applications of Optically Coupled Isolator!;
3.5.1
Introduction....
....
3.5.2
Analog Model for an Optically Coupled Isolator
3.5.3
Types of Analog Circuits . .
3.5.4
Servo Isolation Amplifier . . .
3.5.5
Differential Isolation Amplifier .
3.5.6
AC Coupled Isolation Amplifier.
3.5.7
Digital Isolation Techniques
Digital Applications.
. .
3.6.1
Line Receivers. . . . .
3.6.2
Common Mode Rejection (CMR) Enhancement.
3.6.3
Data Rate Enhancement
. . . . . . . .
3.6.4
Party Line Operation (Bussing, Current l.ooping)
3.6.5
Telephone Circuit Applications
3.6.6
Microprocessor Applications

4.1

Theory and Characterization.

4.2

4.1.1
Photodiode Design and Construction
4.1.2
Photodiode Characterizaton
Photodiode Operation . . . . . .
4.2.1
Circuit Model.
. . . . .
4.2.2
Basic Amplifier Arrangements

5.1

5.2

3.20
3.20
3.20
3.24
3.30
3.30
3.30
3.32
3.32
3.32
3.34
3.35
3.36
3.39
3.40
3.44
3.53
3.61
3.66
3.68

5.0.1
Types of LED Dispalys .
5.0.2
Display Fonts.
. . .
5.0.3
The Display Subsystem.
5.0.4
Data Handling in Display Systems
Numeric Displays with an On-Board Integrated Circuit (OBIC) .
5.1.1
The HP 5082-7300 OBIC Display Family. .
5.1.1.1
Character Font . . . . . .
5.1.1.2
The On-board Integrated Circuit.
5.1.1.3
Temperature Considerationl; ..
5.1.2
Intensity Control for Hexidecimal Displays Using Pulse Width Modulation .
5.1.3
Interfacing a Microprocessor to an OBIe: Numeric Display
Strobable-Seven-Stretched Segment Displays.
5.2.1
Description.....
5.2.1.1
Construction...
5.2.1.2
Data Sheet Parameters

5.1
5.1
5.1
5.3
5.3
5.5
5.5
5.5
5.7
5.8
5.8
5.13
5.13
5.13
5.14

Determining Display Drive Condition:;.

. . . . . . . . . . . . . . . . . . . . 5.14
5.2.2.1
Maximum DC Current (I DC)' Junction Temperature (TJl. and Package Resistance (9JA) 5.14
5.15
5.2.2.2
Forward Voltage (V F) and LED Dynamic Series Resistance (Rs)
5.16
5.2.2.3
Variation of Forward Voltage with Change in Temperature .
5.16
5.2.2.4
Operational Curves for Strobing an LED Device. . . . . .
5.16
5.2.2.5
Maximum Dc Current (lDC MAX) and Temperature Derating .
5.16
5.2.3
Sample Calculation of a Typical and Worst Case Design, Strobed Operation
5.18
5.2.4
Relative Efficiency and Light Output . . . . . . . . . . . . .
5.19
5.2.4.1
Sample Calculation of Time Average Luminous Intensity
5.19
5.2.4.2
Digit-to-Digit and Segment-to-Segment Luminous Intensity Ratio.
5.2.5
Driving a Seven Segment Display
. . . . .
5.20
5.2.5.1
Direct Dc Driving a Display
5.20
5.2.5.2
High Speed Counter Usin{1Dc Drive
5.21
5.2.5.3
Concept of Strobed (Multiplexed) Operation
5.21
5.2.6
Interfacing Microprocessors to Seven Segment Displays .
5.25
5.29
5.2.7
Detection and Indication of Segment I=ailures in Seven Segment LED Displays
5.29
5.2.7.1
Seven-Segment Self-Test Circuit for Common Anode Displays.
.
5.2.7.2
Self Test Circuit for Seven-Segment Displays and Associated Decoder Driver
5.30
5.2.8
Suggested Drive Currents for Stretched Seven Segment Displays Used in Various Ambient
Light Levels
5.31
Monolithic Displays.
. . . .
5.32
5.3.1
Introduction....
5.32
5.3.2
Effect of External Lens
5.34
5.3.3
Construction of Monolithic Displays
5.35
5.3.4
Electrical-Optical Characteristics.
.
5.36
5.3.5
Driving Monolithic Displays . . .
5.39
5.3.6
Interfacing Microprocessors to Monolithic Seven Segment Displays
5.39
Alphanumeric Displays.
. . . . . . .
5.42
5.4.1
The 5x7 LED Array.
. . . . . . . . . . .
5.42
5.4.2
Character Generation in 5x7 Arrays
. . . . . .
5.42
5.4.3
Implementation of a 16 Character Row Scan Display .
5.46
5.4.4
Alphanumeric Displays with On-Board Data Storage
5.46
5.4.4.1
Drive Circuit Concept . .
5.49
5.4.4.2
Interface Circuits for HP HDSP-2000
5.50
5.2.2

5.3

5.4

CONTRAST ENHANCEMENT FOR LED DISPLAYS

6.1

6.1
6.2
6.3

6.1
6.2

6.4

6.5

Contrast and Contrast Ratio . . . . . . .

Eye Response, Peak Wavelength and Dominant Wavelength
Filter Transmittance.
. . .
6.3.1
Plastic Filters.
. .
6.3.2
Optical Glass Filters.
Wavelength Filtering.
. . .
6.4.1
Filtering Standard Red Displays (Ap = 655 nm) .
6.4.2
Filtering High-Efficiency Red Displays ~Ap= 635 nm)
6.4.3
Filtering Yellow Displays (Ap = 583 nml . .
6.4.4
Filtering Green Displays (Ap = 565 nm)
Reduction of Reflected Ambient Light as Providl!d by a Contrast Filter.
6.5.1
Effectiveness of a Wavelength Filter in an Ambient of Artificial Lighting
6.5.2
Effectiveness of a Wavelength Filter in Daylight Ambients . . . . .

6.2
6.4

6.4
6.5
6.7
6.7
6.7
6.7

6.9
6.10
6.12

6.6

Special Wavelength Filters and Filters in Combination

6.13

6.7
6.8
6.9

6.6.1
The Purple Contrast Filter for Red LED Cisplays
6.6.2
Filters in Combination
Louvered Filters.
. . . .
. . . . .
Circular Polarizing Filters .
. . . . .
Anti-Reflection Filters, Mounting Bezels and Other Suggestions

6.13
6.13
6.14
6.15
6.16

7.1
7.2
7.3

Spectral Relationships . . . . . . . .
Geometrical Relationships.
. . . . . .
Photometric and Radiometric Measurements
7.3.1
Spectral Effects . . .
7.3.2
Intensity Measurement.
7.3.3
Sterance Measurement
7.3.4
Flux Measurement
. .

8.1
8.2

Reliability Aspects of the LED Semiconductor Chip

Reliability Aspects of LED Packaging . . . . .

9.1
9.2
9.3

Similarity in LED Packages

The Bending of Leads . .
The Silver Plated Lead Frame.
9.3.1
The Silver Plating.
.
9.3.2
The Effect of Tarnish
9.3.3
Storage and Handling
Solders, Fluxes, and Surface Conditioners
9.4.1
Solders.....
9.4.2
Fluxes.....
9.4.3
Surface Conditioners
Soldering Process.
. .
9.5.1
Wave Soldering
9.5.2
Hand Soldering
9.5.3
Cutting the Leads
9.5.4
Printed Circuit Board
Post Solder Cleaning.
. . .
9.6.1
Types of Cleaners
9.6.2
Bulk Cleaning Processes
9.6.3
Special Cleaning Instructions for Monol thic PC Board Displays
Socket Mounting
. . . . . . . . . . . .
9.7.1
Special Socket Assemblies for LED Dis~llays. . .
Heat Sinking . . . . . . . . . . . . . . . .
9.8.1
OBI C Display Assembly with On-Board Heat Sink .
9.8.2
A Display Assembly with Heat Pipe
9.8.3
List of Manufacturers

9.4

9.5

9.6

9.7
9.8

9.1
9.1
9.1
9.2
9.2
9.2
9.3
9.3
9.3

9.5
9.5
9.5
9.6
9.6
9.6
9.6
9.6
9.7

9.7
9.8
9.9
9.10
9.11
9.12
9.12
1-1

Over the last decade the commercial availability of the Light Emitting Diode has provided electronic system
designers with a revolutionary component for application in the areas of information display and photocoup'ers.
As the use of the LED and associated derivative rroducts has become more common, many electronic engineers
have encountered the need for a resource of information about the application of and designing with 1ED
products. This book is intended to serve as an engineering guide to the use of a wide range of solid state
optoelectronic products. Wherever appropriate, tl.roughout the book, use has been made of approximation; to
simplify the formulas and design equations. As su,~h, some formulas will differ substantially in form from tJ ose
found in a physics book or the subject; however, the end result will be an excellent approximation of the
behavior of real devices.
To treat the various major aspects of LED products the book is divided into chapters covering each of the
generalized LED product types. Additional chapters treat such peripheral information as contrast enhancement
techniques, photometry and radiometry, LED reliability, mechanical considerations of LED devices, photodiodes
and LED theory.
The authors are grateful to Bill Jensen and Mike A bbey for their preparation of artwork and Barbara Lee for her
accuracy and patience in typing of the manuscript Thanks is also due to Al Petrucello and George Liu for tl elr
effort in breadboarding and performance checking of the circuits detailed in the book

Electroluminescence in solids is a phenomenon which has

been well known and intensively studied for many years.
Perhaps
the most commonly utilized application of
electroluminescence is in the screen of a television set.
Better
known as cathodluminescence,
this form of
electroluminescence
is caused by the collision of high
energy electrons with the phosphor coating which lines the
inside surface of a cathode ray tube. Though numerous
other less common types of electroluminescence have been
observed, one particular type has given rise to an entirely
new field of technology. This is the phenomenon of p-n
junction injection electroluminescence.
The emission of
light (photons) from a p-n junction was first noted in
naturally occurring junctions by Lossew in 1923. More
recently, (circa 1962) studies of GaAs revealed that it was
qui te f easib Ie to achieve relatively high levels of
electroluminescent emission from p-n junctions. Starting in
1962, an intensive development program was undertaken at
Hewlett-Packard
and elsewhere to produce a useful,
manufacturable, visible-light emitting p-n junction device.
An electroluminescent device of this type was thought to
be attractive because of:

It is possible, however, to introduce impurities into the

semiconductor which will produce electronic states in the
forbidden gap. Impurities which add electrons to the
conduction band are called donors and give raise to n-type
conductivity. Impurities which produce electron vacancies,
holes, in the valence band are called acceptors and give raise
to p-type conducitivity. The energy bands and the relative
donor and acceptor levels are shown in Figure 1.1.3-la.

p-TYPE
~
~BAND

CONDUCTION
BAND
DONOR
LEVEL

ACCEPTOR
LEVEL
VALENCE
BAND

VALENCE
BAND
(a.)

"TYPE

1)

2)

3)

The low currents and voltages required to produce

useful light output.
The precision with which the light emitting area
could be defined through the use of semiconductor
photo lithographic processes.
The high speed at which the device could be
switched.

CONDUCTION

pTYPE

~~
~

~~~~t ii ij
r------_~

EO

~~~:-JUNCTION
REGION

............

ACCEPTOR
LEVEL

(b.)

As a background to the theory of electroluminescent

diodes, it is useful to review the basic concepts of current
flow in p-n junctions.
DONOR
LEVEL

In semiconductor materials, as in all crystalline solids,

electrons can assume only certain levels of energy. The
Valence band and the Conduction band are the terms
assigned to the two bands having the highest energy levels
for electrons in normal semiconductor materials. The
separation between the top of the valence band and the
bottom of the conduction band is called the energy gap. In
a pure semiconductor material, electrons cannot exist in
(i.e., assume energy levels within the range of) this
forbidden gap.

ACCEPTOR
LEVEL

JUNCTION
REGION

(c.)

A p-n junction can be formed in a semiconductor mat(:rial

by doping one region with donor atoms and an adjacent
region with acceptors. In this situation, electrons and h'Jles
will flow in opposite directions across the junction (without
any applied bias) until an equilibrium is reached. This -vill
give rise to a built-in potentional barrier, EB, which is
slightly less than the energy gap. Figure 1.1.3-1a and -lb
represent schematically the material before and after rhis
equilibrium is reached. Note that the natural potential
difference across the junction makes it difficult for an
electron, for instance, to move from the n to the p region.
If, however, an external electric bias is applied across the
junction in a manner such as to counteract the built-in
potential, additional electrons and holes will flow (be
injected) across the junction boundary. See Figure 1.1.3-1c.
These
carriers will then recombine via one of rhe
mechanisms described in Section 1.1.4. This recombination,
may give rise to photon emission as well as current flew.
Within a few diffusion lengths of the junction, virtually all
of the injected carriers will have recombined and the bulk
current flow will occur by means of majority carrier
diffusion.

The injected electrons or holes are annihilated by a

carrier of opposite type through one of several possitle
recombination processes. These recombination processes
are either "radiative" or "non-radiative". The energy release
in a radiative process is in the form of photons (ligh1);
whereas, the energy release in a non-radiative process is in
the form of phonons (heat). The two important radiative
processes in LEDs are shown schematically in Figu.~e
1.1.4-1. Band-to-band recombination,
1, is the direct
recombination
of an electron near the bottom of tl-e
conduction band with a hole near the top of the valence
band. The photon energy is then approximately equal to
the band-gap energy of the crystal. This mechanism c.f
radiative recombination is predominant in direct band-gap
materials such as GaAs.

The second type, 2, of radiative recombination occurs by

the formation and annihilation of a bound exciton at an
"iso-electronic"
center. Iso-electronic centers (associated
with specific impurities in the crystal) are normally neutral,
but introduce a local potential which is attractive to
electrons. In p-type material, an injected electron is first
trapped at the center. The negatively charged center then
captures a hole from the valence band to form the bound
exciton. The subsequent annihilation of this hole-electron
pair yields a photon with an energy equal to the band gap
minus an energy approximately equal to the binding energy
of the center. This mechanism of radiative recombination is
predominant in indirect band-gap materials such as GaP.
The photon energy can be converted to wavelength by
equation 1.1.4-1:
~ = 1240

where L'1Eis the energy transition in electron volts.

Equation 1.1.4-1 relates the expected wavelength of photon

emission to energy differential experienced by an electron
undergoing recombination. The maximum possible energy
of the emitted photons is determined by the band-gap
energy of the solid in which the p-n junction is formed.
There are numerous elements and elemental compounds
which have band-gap energies that lie in the region which
could produce emission ranging from ultraviolet to infrared.
However, very few of these materials are viable candidates
for practical LED devices. Table 1.1.5-1 lists a few of the
available materials and the associated band gap and
emission wavelength. Some of these materials are not useful
because they cannot be doped to form a p-n junction, some
do not have emission at a useful wavelength, some have too
low a conversion efficiency to be useful. At the present
time, the only commercially available LED devices are
manufactured using GaAs, GaP or the ternary compound
Ga(As,P).
MATERIAL

CONDUCTION
BAND
SHALLOW DONOR
IMPURITY LEVEL

SHALLOW ACCEPTOR
IMPURITY LEVEL

(nrn)

L'1E

BAND
GAP ENERGY

EMISSION
Anrn

TRANSITION
TYPE

1880
1140
910
560
650
775
6900
563-413

INDIRECT
INDIRECT
DIRECT
INDIRECT
DIRECT
INDIRECT
DIRECT
INDIRECT

eV
Ge
Si
GaAs
GaP
GaAs60P40
AISb
In Sb
SiC

0.66
1.09
1.43
2.24
1.91
1.60
0.18
2.2-3.0

In Section 1.104, the concepts of recombination were

presented. In this section, the concepts of direct and
indirect
band-gap
transitions
are considered. The
conduction band edge and the valence band edge as a
function of momentum for allowed electronic states in the
GaAsI_xP x compound system (where x = mole fraction) are
plotted in Figure 1.1.6-1 for different values of x. As
indicated, there are two wells or minima; one designated as
direct and the other as indirect. Electrons in the conduction
band will generally occupy states in the lowest energy
minimum, while holes will occupy states near the valence
band maximum. Electrons in the direct minimum and holes
at the top of the valence band have equal momentum;
ENERGY

1.'.7
GaAs1_x Px

DIRECT MINIMUM

!
GaP GREEN
GaAsP AMBER
GaAsP RED
GaAs
INFRARED

MOMENTUM

Figure 1.1.6-1

whereas, electrons in the indirect minimum have different

momementum.
Since momentum
is conserved,
band-to-band transitions may occur with high probability
for electrons in the direct minimum. The probability of a
band-to-band transition for an electron in an indirect
minimum is nearly zero, since a third component (phonon)
must participate in the process in order to conserve
momentum. Note that GaAs and GaAs1_xPx up to x '=" .4
are primarily direct gap materials in contrast to GaAs1_xPx
with x > .4 and GaP which are primarily indirect gap
materials. (Also note that the band gap energy increases
with increasing x.) Thus, without the incorporation of
special recombination centers, indirect gap materials, like
GaP, are very inefficient light emitters, since the dominant
recombination processes are non-radiative.

ED
eV

EI
eV

o.

1.43

1.86

910

0.40

1.92

1.97

650

Enhanced
Materials

Photon

Emission

in

Indirect

Gap

More recent developments in LED technology have led to

significantly enhanced radiative recombination in indirect
gap materials, such as GaP. These have been achieved by the
incorporation
of appropriate
impurities to form
iso-electronic trapping centers, as discussed in Section
1.1.4. Because the trapped electron is highly localized at
the center, its momentum is diffused. Thus, momentum can
be conserved and the probability of direct recombination is
greatly enhanced.
In GaP, two types of iso-electronic centers are utilized. One
is formed by replacing a Phosphorous atom with an
equivalent Nitrogen atom in the lattice. Another type is
formed by replacing an adjacent Gallium and Phosphorous
pair of atoms with a Zinc-Oxygen pair having the same total
valence electrons. The resulting trap states will lie in the
band gap at energies somewhat below the edge of the
conduction band. This structure is illustrated schematically
in Figure 1.1.7-1.
For N doped GaP, bound exciton recombination gives rise
to emission at 565 nm (green). For Zn-O doping, the
corresponding emission wavelength has a peak at about 700
nm (red). Nitrogen doping can be utilized in indirect gap
GaAsl_xP x to similarly produce yellow and red emission
with relatively high quantum efficiencies.

(nm)

0.85

1.55

2.17

580

1.0

2.78

2.26

560

Plot of Momentum vs. Bandgap Energy for Various

Compounds of the GaAs/GaPSystem

The efficiency with which an electroluminescent material

can convert current flow into detectable photon emission is
of paramount importance in determining whether usable
devices can be manufactured
from that material. The
percentage
of the current
flow which results in
recombinations which give rise to photons of the desired

efficiency of the diode is largest. This maximum occurs at

655 nm and x = .4. This relationship is illustrated in Figure
1.2-1. The plots of YREL and 1JextREL are normalized to
the peak values. 1JLUMINOUS represents the product of the
two values.

GaP
/

CONDUCTION BAND

For diodes having shorter peak emitting wavelength, such as

N doped Ga(As,P) emitters at 565 nm (green) and 585 nm
(yellow), 1Jext is substantially lower but YREL is higher.
This has resulted in a spectrum of red through green
emitters which all produce about the same 1JLUMINOUS
(within an order of magnitude). The relative emission bands
and responsivities of several different emitters and detectors
are depicted in Figure 1.2-2. When trying to achieve
matching with a silicon detector, factors other than
achieving maximum responsivity may come into play.
These factors are covered in Section 3.1.1.
/

Figure 1.1.71

VALENCE BAND

Plot of Momentum

vs. Bandgap Energy for Indirect

GaP Materials Showing Special Trapping

Levels

wavelength
is a measure of the internal conversion
efficiency, 1Jint of the diode. Obviously, a material which
has a very low 1Jint would be of little interest as a practiGal
electroluminescent device. However, even a material which
has an 1Jint of 100% may not be useful if the emitted
photons cannot be efficiently coupled from the devices to a
detector. Two major factors control the internal to external
coupling coefficient. One factor is direct reabsorbtion of
the emitted photon in the bulk material, basically a
measure of the opacity of the material. The other factor is
internal reflection at the crystal/air interface which causes
the photon to be reflected back into the crystal alld
subsequently reabsorbed.
No matter how efficient a device is in terms of 1Jext and
1Jint' the output power cannot be detected unless the
wavelength is matched to available detectors. In the va,t
majority of LED applications, the detector of interest is
either the human eye or a silicon photodetector.
Tf.e
nominal range of spectral sensitivity of the silicon detector
is from about 300 nm to 1100 nm. The human eye has a
much narrower range of sensitivity with useful responsivity
only in the range from 400 nm to 700 nm. It is generally
desirable to optimize the total coupling efficiency between
the emitting device input signal and detector output
response.
For the case of matching a standarc.,
non-Nitrogen doped GaAs 1-xPx emitter to the human ey,
this is accomplished by picking the wavelength at which th~
product of the relative response of the eye and the relatiw

The mechanism of current flow in an LED as discussed in

Section 1.1.3 represents the function of an ideal diode. Just
as in silicon semiconductor technology, a p-n junction in
Ga(As,P) material exhibits numerous discrepancies between
the ideal and the actual diode. Surface recombination,
tunneling phenomena, space charge recombination, current
crowding and bulk recombination
due to anomalous
impurities tend to reduce the efficiency of photon emission
of an LED. The relative effects of these non-radiative
phenomena tend to be dependent on the current density
(amps per square centimeter of junction area) and the
perimeter to area ratio of the diode. At low forward
currents 1 A/cm2), nearly all of the current flowing in a
practical diode may result from one of these non-ideal
mechanisms.
As the current densi~ in the junction is increased to the
value of 10 amps/cm , the non-ideal mechanisms will tend
to have saturated, so that non-radiative current becomes a
progressively smaller function of total current. As the
junction
current
density
becomes
very large
(>500 amps/cm2) current crowding in the junction will
tend to reduce emitting efficiency. The net effect of these
mechanisms is to cause an LED to have a peak operating
efficiency at a current which is dependent on the area and
geometry of the junction and the size of the electrical
contact. Figure 1.3-1 is a plot of normalized operating
efficiency vs. current density for an .011" square diode.
Note that at low current densities, a doubling of current
may lead to a factor of 5 increase in luminous flux whereas
at high current densities, doubling the current may result in
slightly less than a doubling of light output.
The practical benefits of this aspect of LED devices are
discussed in Sections 2 and 5.

>
u
z
w
u
u..
u..

.1

w
w

>
~

<t:
...J

.01

900

Figure 1.2-1

C.I.E.

>
u
z
w

5:1
u..
u..
w
w

I-

<t:

CC

GaAsP
Hw ~ 16 nm

GaAs

1.0
.8
.6
.4

...J

Relative Luminous Efficiency of the Human Eye and

GaAs _xP x as a Function of Wavelength
1

.2

Si PIN DIODE

\/

"""

565

600

635
65!i

WAVELENGTH,

Figure 1.2-2

Normalized
Detectors

A (nm)

ReSIJOnsivities of Different

Emitters

and

>-

u
z 1.4
w

!:?
u..
u..
w

(::l

z
i=
<l:

a:
w
a..

o
w
>
i=
<l:

-I

a:

1 x 101
J AMPS/cm2
Figure 1.3-1

Normaliziid Operating
for an LED

The processes used to manufacture LED devices are

basically an outgrowth of the techniques perfected for the
manufacture
of silicon semiconductor devices. Crystal
growing,
epitaxial
deposition,
controlled impurity
diffusions, photolithography, and vapor deposition of thin
mms all play an important role in the production of LEDs.

Since Ga(As,P) is a non-congruently melting material, a

single crystal having proper GaP concentration is extremely
difficult, if not impossible, to produce. The Ga(As,P) .s,
therefore, produced as an epitaxial layer grown on a
substrate of either GaAs or GaP. The substrates are wafers
sliced from ingots grown by the Czochralski method. The;e
wafers are mechanically and chemically polished to provide
a crystal
substrate which is nearly free of lattice
imperfections and unwanted chemical impurities. During
the first stage of the epitaxial growth, the crystal layer has
the same composition (GaP or GaAs) as the substrate. As
the growth of the layer proceeds, the composition is
changed gradually so as to maintain monocrystallinity in
the epi layer. During the fmal mil (25 ,urn) of growth, tl-.e
PI As ratio is held at the desired level so that later when the
p-type material is diffused into the epitaxial layer, the
resulting p-n junction will exist in homogeneous GaAs 1-xP x
of the desired composition (typically x = .4 for direct-gap
655 nm RED LEDs). The n-type doping and any special
impurities required to generate the bound exciton states are
also introduced in the gas stream.nf the vapor-phase epitaxy'
system used to grow the layer.

Efficiency ys. Current

Density

Following growth of the epi layer, the wafer is coated with

silicon nitride which acts as a barrier to the p-type diffusant
and passivates the resulting p-n junction. To produce
shaped p-n junctions, photolithography is used as in silicon
wafer processing to shape the openings in the silicon
nitride. Through the openings in the silicon nitride, p-type
diffusant (generally Zinc enters the Ga(As,P) to form the
p-n junction. It is not possible to use a layer of silicon
dioxide
because the p-type diffusant used in LED
processing penetrates silicon dioxide too rapidly. Figure
1.4.1-1 depicts the crossectional and top view of a typical
LED device.

The photons
generated
at the junction of a p-n
electroluminescent diode are emitted in all directions. If the
diode substrate is opaque, as in the case with GaAs, only
those photons which are emitted upward within a critical
angle defmed by Snell's Law will be emitted as useful light.
All other photons emitted into or reflected into the bulk
crystal will be absorbed. This phenomena is illustrated in
Figure 1.4.2-la. GaP is nearly transparent by comparison
with GaAs. Diodes formed in an epitaxial layer grown on a
GaP substrate will exhibit improved efficiency due to the
emission of photons which would be absorbed in the GaAs
substrate structure. The resulting structure is depicted in
Figure 1.4.2-1b. In actual manufacturing practice, indirect
gap devices are generally fabricated on GaP substrates while
direct-gap devices are fabricated on GaAs substrates.

I'
/

AuGe BACK CONTACT

SHAPED JUNCTION
AREA

EDGE
OF NITROX

(.1251

Figure 1.4.2-1

Effects of Transparent and Opaque Substrates on

Photons Emitted at the Junction

The Effect
Parameters

of Temperature

Variation

on LE D

The physical parameters of light emitting diodes, as in all

semiconductor devices, exhibit a dependence on absolute
temperature.
The forward
voltage/forward
currer;t
relationship, quantum efficiency, and emitted wavelengtl
are the temperature variant parameters of greatest interest
to the LED user.

The forward voltage/forward current relationship

LED can be expressed by equation 1.5.1-1

for all

Figure 1.5.1-1

where n is a function of temperature, IF' and the nature of

the recombination mechanism. Empirical results for both
direct
and indirect
gap LEDs exhibit temperature
coefficients of -1.3 mVtC to -2.3 mVtC depending on
forward current. Figure 1.5.1-1 depicts this relation.
1.5.2

Change in Peak Wavelength

Temperature

as a Function

of

The effective energy gap in both direct and indirect gap

semiconductors tends to become slightly smaller with
increasing temperature. This will result in slight increases in

Temperature Coefficient of Forward Voltage as

a Function of Forward Current

the emitted wavelength. For direct gap emitters wavelength

will increase by 0.2 nmtC. The wavelength of Nitrogen
doped indirect gap emitters shows a somewhat lower
dependency with typical positive variations of about
0.09 nmtC.

The radiant power of an LED decreases as a function of

increasing
temperature.
Variations on the order of
-1 % per 0 C are typical for both direct and indirect gap
materials. F()r LED devices in applications using the eye as
a detector, the variation in responsivity of the eye as a
function of wavelength must be added to the temperature

dependent variation of the LED radiant emission. In the

650 nm Red region, the eye response is changing at about
-4.3%/nm. In the 565 nm Green region, the responsivity is
changing by about -.86%/nm. If the change in wavelength as
a function of temperature is .2 nmtC for a 655 nm direct
gap device, then apparent optical intensity will be
decreasing by about:
(1.5.3-1)
t.IV ( %)

t.T

(-4.3% ) (.2 nm) _1%

nm
c
c

-1.86 %fC
IVTEMP 0 = luminous intensity at reference
temperature

Conversely, an indirect gap device in the 565 nm green

region will exhibit:
(1.5.3-2)
t.IV (~)

t.T

(-.86%)
nm

(.09 nm)_ 1%

The change in luminous intensity of an LED exhibits a

logarithmic relationship over large changes in temperature.
Calculation of expected changes should be done using the
following relationship:

= -1.08

%fC

medium. The transmission coefficient

interface is:
"2-"1

T = 1 + R = 1 -(--)

Incandescent, flourescent and neon lamps have been used

for a long period of time in a wide variety of applications.
As a result, their physical properties are well understood.

4"2 "1
"2 + 2 (12"1 + "1

Most commercial
LED lamps are manufactured
by
encapsulating an LED chip inside a plastic package with a
lens surface directly above the LED junction. If the plastic
is undiffused, this configuration forms an immersion lens.
The effect of this immersion lens construction is to enlarge
the apparent size of the emitter. The magnification is a
d ire c t fun ction of the index of refraction of the
encapsulating
material. The immersion construction
increases the light output from the LED chip by reducing
fresnelloss and increasing the critical angle.
Not every photon generated within the LED's p-n junction
emerges from the surface of the junction to reach the eye
of an observer. Three separate loss mechanisms contribute
to reduce the quantity of emitted photons: 1) loss due to
absorbtion within the LED chip material, 2) fresnelloss and
3) critical angle loss.

"2+"1

In recent years, the semiconductor LED lamp has been

replacing these earlier devices in many applications. Also,
new applications, designed specifically for the LED lamp,
are being developed almost every day. Therefore, in order
to effectively utilize an LED lamp in an application, a
designer should be familiar with the physical properties of
an LED device.

across the medium

The Fresnel Loss Efficiency Factor, f]Fr' is obtained by

dividing numerator and denominator of equation 2.1.2-2 by
n2n1 :

For an unencapsulated GaAsP LED chip with an index of

refraction of 3.4 emitting directly into air with index of
refraction of 1.0, the fresnelloss efficiency factor is 0.702:
f]

Fr

4
2 + 1/3.4 + 3.4/1

= .702

Thus, only 70.2% of the light reaching the chip surface is

transmitted across the chip/air interface.
If the LED chip is coated with an intermediate material
having a suitable index of refraction, f]Fr can be improved.
Ideally, this material should have an index of refraction
nx = Vn1 n2 = V(3.4)(l) = 1.84. Then f]Fr = T 1T2, where:

GaAsP/GaAs LED devices are opaque to light and absorb

approximately 85% of the photons emitted at the junction,
allowing
a material efficiency factor of f] R: .15. A
significant improvement is observed with GaAsP/GaP
devices where f] R: .76.

=T =

T
1
When light passes from a medium whose index of refration
is n 1 to a medium whose index of refraction is n2' a
portion of the light is reflected back at the medium
interface. This loss of light is called fresnel loss. The
reflection coefficient is:

rn;

2+y~-

f"2

= .912

+Y-E.
__

"2
"1
and the overall
transmission
is: f]Fr =
(.912)2 = .832. This is an improvement of 18.5%.

Substantial improvement is obtained when a LED is

encapsulated in plastic having an index of refraction of 1.5.
The fresnel efficiency factor is calculated to be 0.816:
f]Fr = T 1 T 2

Since the numerator (n2-n1) is squared, the same reflectior

loss occurs whether the light is passing from a low-index tc
a high-index medium or from a high-index to a low-inde>.

T 1T2

=~
=

+ 3.4/1~ + 1.5/3.J
[.850] [.960] = .816

~ + 1/1~ + 1.5/J

This is an improvement of 16.2% over that

unencapsulated LED and just 2.3% less than ideal.

Light rays from within the crystal reaching the surface at an

angle greater than 17.1 are totally reflected back into the
crystal.

of an

The Critical Angle Efficiency Factor for an LED emitting

into air is 0.0865:

The third efficiency loss is due to total internal reflection

of photons incident to the chip surface at angles greater
than the critical angle. This effect is shown diagramatic.ally
in Figure 2.1.3-1. As is depicted in Figure 2.1.3-1a, a ray of
light passing from the interior of the crystal to the outer
surface is refracted according to Snell's Law:

where:

2)2 = ( ---)
1 \2 =.0865
~Cr= (n-n1
3.4
Encapsulating an LED in a medium with a high index of
refraction, nx' increases the amount of flux that can escape
from the crystal. However, if a flat surface is used, this
increase in flux is lost because the refraction angle, ()x' at
the crystal-to-encapsulant interface becomes the incidence
angle a t the encapsulant-to-air
interface, see Figure
2.1.3-1 b. The value of the critical angle is not improved.

()1 = The angle of incidence inside, at the surfal:e

of the crystal.
()2 = The angle of refraction outside, at the
surface of the crystal.
n1 = The index of refraction of the crystal.
n2 = The index of refraction of the medium
outside of the crystal.

By shaping the encapsulant into a dome lens, the incidence

angle at the encapsulant-to-air interface is less than the
refraction angle at the crystal to encapsulant interface. As
shown in Figure 2.1.3-1c, flux that would have been
trapped by a flat surface is permitted to pass through the
dome lens. If no flux is trapped within the encapsulant, the
critical angle efficiency for a plastic dome lens become
0.195:

The angle of incidance, () 1, at which the angle of refraction

()2 is equal to 90, is called the critical angle, ()c' The
critical angle is calculated for an LED chip as follows:
n1 sin ()c = n2 sin 90

(2. U:-2)

sin ()c = n2/n1

()c = sin1 (n2/n1) = sin-1 (1/3.4)
()c=17.1

"I

Oc
FRESNEL LOSS
DUE TO DIFF. OF
INDEX OF
REFRACTION
~ 30%

sin

0I

= sin-I

="2

sin O2

OlIn]

17

Figure 2.1.3-1

at ',A:: "lSnO.
at B: n.snOx

=n.s~nO.

"lSmOI

=n2sin02

Oc

=0251002

sin-I "}/n1

17<>

SOLVING
SNELL'S LAW
FOR PLASTIC
COATING

Be = sin~lo./nl

The Effects em Optical Efficiency Produced by an

Optical Coal ing.

26

CRITICAL ANGLE
FOR SPHERICAL
DOME LENS

This improvement of 2.25: lover an unencapsulated LED

chip. The critical angle at the LED chip to dome interface is
increased from (J c = 17 to (J c = 26.
(J

= sin-1

ratio of the radiative component to the total current. The

internal quantum efficiency may be calculated by dividing
the external quantum efficiency by the optical efficiency:
1/

(n /nl) = sin-1 (1.5/3.4) = 26.2

x

qext

1/optical

The optical efficiency of an LED lamp is the product of the

absorbtion efficiency, fresnel efficiency and critical angle
efficiency:

The optical efficiency for a T-l 3/4 clear high-efficiency red

lamp is 0.121:

The ratio of luminous flux (lumens) to radiant flux (watts)

is called luminous efficacy, 1/v (lm/w). For a complete
discussion of luminous efficacy, see the section on
photometry. The value of luminous efficacy is given on
each lamp data sheet. These values for luminous efficacy
are:
LED
Luminous Efficacy. 77v (Im/W)

External quantum efficiency,

following expression:

77

r/>e (W)

~xtl

(A) [1240

LX(nm)

eV ~

77qext' is defined by the

60
photon
electron

photonJ

The value 1240 is the product of Plank's Constant and the

speed of light:
X

J!!;
Eg
34 joules'

= (6.626 x 10-

sec)(2.998 x 108 meters/see)

Eg (1.602 x 10-19 joules/eV)

= 1.240 x 10-6

X (nm) = 1240
Eg(eV)

~xt (.010A)(

1240 )
635 nm .

= 4.95 x 10-3

Yellow
460

Green
630

In some applications, such as providing a light source for a

lens focusing system which subtends a specified solid angle,
the amount of radiated power in microwatts per steradian,
Ie CIlW /sr), is important. The value of radiated power per
unit solid angle may quickly be calculated from the
following equation:
Iv (mcd = m Im/srl
Ie (j1W/sr) = ....;;...-----1/v (Im/w))
(
1000

= (12 m 1m/srI (1000) = 89 j1W/sr

A T-l 3/4 high-efficiency red lamp with 2(J~ = 35 and

produces 12 mcd at 10 mA radiates 96.6 j1W. The external
quantum
efficiency
for this lamp is 4.95 x 10-3
photons/electron.
6

Hi-Eff. Red
135

For a T-l 3/4 clear high-efficiency red lamp with 2(J~ = 35

and Iv = 12 mcd, the radiated power per sterdian is
89pW/sr.

m - eV

Eg(eV)

1/ = 96.6 x 10-

Std.Red

photons
electron

The current flowing through an LED is composed of two

components, a radiative component and a non radiative
component. The internal quantum efficiency, (J qint' is the

1351m/W

In applications where the total radiated flux is to be

utilized, the lamp radiation pattern must be known. The
section on photometry provides a detailed discussion on
calculating the total radiated flux using the method of
zonal integration. In this section, an illustrative example of
this method is presented which calculates the total flux for
a T -1 3/4 undiffused high-efficiency red lamp with
2(J~ = 35 and producing 12 mcd.

20
30
40

,....."

/'

",

~ 0.8

50
r-0.6

60

"

0.4

r--- r--

...
~ ~

.......

0.2

r--- ~
l----- >---

--

r-.....

.......

......

r.......

10

,20

0.2

30

50

40

60

70

80

90

10

Figure 2.1.7.2-1 Radiation Pattern for an Undiffused T-1 3/4 Hi

Efficiency Red Lamp.

0
0.4

ep e =

120
0.6

1000 Ivo cp
=/lW
flv

where:

130
140

0.8

10

Ivo = On-axis luminous intensity (mcd = m 1m

cp= Total solid angle into which the flux is
radiated (sr).
flv = Luminous efficacy (Im/W).

150

(2.1.7

ep = 1000 (12) (1.0869) = 96.6/lW

e
Figure 2.1.7.2-1 is the polar/linear plot of the radiation
pattern for the above lamp. The linear graph is used to
determine the relative luminous intensity points for use in
the calculation. Since the grid is marked off in 5
increments, we shall take our summation over 5 intervals.
!::,= 5 = 180

then N = 36

'

cp(mA) = 1. I (M!::')Cz (M!::')+ ~ Ir (m!::')C (m!::')

z
Io
2 r
m=1
(2.1.7.2-1)

where: Cz (M!::,)= ~
N

Ir (m!::')

1
2
3
4
5

5
10
15
20

6
7

30
35
40
45
50
55
60
65

.981
.938
.801
.516
.435
.379
.342
.298
.248
.199
.149
.099
.056

.0478
.0952
.1419
.1875
.2317
.2742
.3145
.3524
.3877
.4200
.4492
.4749
.4969

10
11
12
M-1 ~13

25

sin (M!::')= .5483 sin (M!::,)

Ir (M!::,)= Relative Luminous Intensity from linear graph

at the angle m!::'.
The calculation is set up in a tabular format. The
calculation needs to include angles only up to m!::'= 70 ,
M= 14.

m!::'C)

m-1

135

The total radiated

following equation:
cp(M!::') =
10

(m!::')

Ir (m!::')Cz (m!::')

.0469
.0893
.1137
.0968
.1008
.1039
.1076
.1050
.0961
.0836
.0669
.0470
m-1 .0278
L 1.0854
m=1

power is now calculated

1. (.0031) + 1.0854 = 1.0869 sr

2

from the

The luminous intensity of an LED lamp is a function of the

magnification provided by the immersion lens. Using the
parameters of Figure 2.1.8-1 and assuming paraxial light
rays, the magnification may then be defined as the ratio of
the image size to the object size and approximated by the
following formula:
M agm'f"IcatlO"

b, Y2

m=

Y,

""

The effect of magnification on luminous intensity and

radiation pattern for an undiffused package is illustrated in
Figure 2.1.82. If the LED is encapsulated very close to the
dome surface, curve A, the dome surface is effectively flat.
There is no magnification and the radiation pattern remains
the lambertian pattern of the unencapsulated LED chip. In
the lambertian radiation pattern, the luminous intensity
varies as the cosine of the off-axis angle, ().

, _ ~ (,, _ "2)
r

",

For a T1 3j4lamp, the focal length is: Xl =

4.70 mm, r = 2.44 mm and n1 = 1.53. The magnification is
approximately 3:

, _ 4.70 mm
2.44 mm

('--'-)
'.53

A1R

",

'\;OBJECT

PLASTIC
IMMERSION
LENS

~t

IMAGE

b__
ON-AXIS LUMINOUS
INTENSITY.IVO

LUMINOUS
FLUX.V

RADIATION
PATTERN

",0

IV: ""VOof
unencapsulated LED

n IV1

Lambertian

0/2=,

n21V1

n n21V1

Lambertian

>0/2

>n21

<n2 n IV1

Non Lamberian

CURVE
A

fl

The magnification may also be calculated from the focal

length of the immersion lens. If the surface of the dome
lens is regarded as the principal plane, a focal point may be
defined as that point on the lens axis from which paraxial
light rays emerge from the dome lens parallel to the lens
axis. The principal plane is the locus of equivalent
refracting points for a lens where extended entering and
emerging rays intersect. The focal length of the lamp's
spherical dome lens may be determined from the following
formula:

,_("2 )

",

2.44mm

,-(._,

\ '.53
The magnification
length:

,-(:n

may then be calculated from this focal

DISTANCE
X

Figure 2.1.82

The luminous
intensity.

V1

Effect of Magnification on Luminous Intensity and

Radiation Pattern for an Undiffused Plastic Lamp.

flux is equal to

1f

times the luminous

At distances from the dome surface of o<X<r, the LED

chip is magnified and the luminous intensity is directly
proportional to the square of the magnification. When
x = r, the magnification is equal to the index of refraction
of the encapsulant and the luminous intensity is then equal
to that of the unencapsulated chip multiplied by the index
of refraction squared. For a spherically shaped dome lens,
the radiation pattern is still lambertian, curve B. The total
luminous flux increased by the factor of index of refraction
squared.
At distances greater than the radius of the dome lens
further increases the magnification above that value equal
to the index of refraction. The luminous intensity increases
but the radiation pattern narrows reducing the viewing
angle, curve C. The maximum practical magnification is
that at which the image of the LED's emitting area equals
the diameter of the dome lens.

BOTH LAMPS HAVE THE SAME

LED CHIP AND THE SAME SHAPED
DOME LENS

An aspheric dome lens can be used to achieve a high value

of on-axis luminous intensity along with a wider radiation
pattern than is obtainable with a spherical dome lens. An
aspheric dome lens is used as the package of T-l 3/4 low
profJle lamps. The comparison of the radiation pattern of
an aspheric dome lens with a spherical dome lens of equal
magnification is illustrated in Figure 2.1.8-3.
At this point, the reader is reminded that when evaluating
the specifications for an LED lamp, it is necessary to take
into account both the on-axis luminous intensity and :he
lamps radiation pattern. Even though two lamps may h.lve
the same luminous intensity at a specified forward current,
one may have a wider radiation pattern. The lamp with ":he
wider radiation pattern may have a more efficient dome
lens or a more efficient LED chip.

The immersion lens concept applies to a lamp which has the

LED encapsulated in undiffused plastic. The result i~ a
beam of light with a high value of luminous flux which is
concentrated in a narrow radiation pattern. This lamp
configuration
is especially
useful for backlighting
applications and for applications requiring a concentrated
light source.
A front panel indicator lamp requires a very wide off-axis
viewing angle. To achieve this wide viewing angle, diffusant
is added to the lamp to disperse the light rays emitting frem
the LED. The result is a lamp with a wide radiation pattern
and a reduced value of on-axis luminous intensity. Figure
2.1.9-1 pictorially illustrates the differences between m
undiffused and diffused lamp. Dye coloring is added to tint
the diffused lamp to enhance on/off contrast.

=-EE---)
T-' 3/4 LOW PROFILE LAMP
WITH AN ASPHERIC DOME LENS
Figure 2.1.8-3

Radiation Patterns for Undiffused Lamps with

Spherical and Aspheric Dome Lenses.

Figure 2.1.9-1

Comparison Between Undiffused and Diffused

Plastic LED Lamps.

The function of a lamp package is to utilize the various

physical properties described in Section 2.1 to effect the
best coupling of the emitted light from an LED to an
observer or electronic detector. The desired package
configuration depends upon the specific requirements of
each application. These requirements will determine such
parameters as the size of the device, if the lamp is to be
tinted or untinted and whether the encapsulating epoxy is
to be diffused or undiffused. Some requirements may
dictate a lamp device where epoxy encapsulating is not
desired, such as in a high-reliability application.
It is obvious that there could well be as many lamp package
configurations as there are applications. However, certain
lamp packages have been developed by the optoelectronics
industry which cover a wide range of the more common
applications. Prior to discussing the commonly available
package configurations, it is a benefit to the reader to first
understand the basic lamp packaging process.

Immensely popular because of its low cost, the lead frame

techniqw is used for the packaging of LED lamp devices.
Basically, in lead frame packaging, LED dice are die
attached to one element of a metal frame and a wire bond
is made from the top of each die to another element of the
frame. During the die-attach and bonding operations, the
elements are joined for mechanical support by metal straps

called "dam bars". Plastic encapsulation applied around the

devices surrounds also the elemen ts of the frame while the
dam bars remain outside of the plastic. When cured, the
plastic provides mechanical support to the elements and the
dam bars may then be sheared away. Shearing of the dam
bars leaves the opposite ends of the lead frame protruding
from the plastic; these then become the external leads of
the fmished device. Only after the dam bars are removed
can the devices be directly energized for testing.
The material of the lead frame is thick enough that the dice
can be attached on the edge (rather than the face) of the
frame material. This places the die on the end of an
element, called a "die-attach post", which is canted,
allowing the die to be centered with respect to those
portions of the lead frame which later become the external
leads, as shown in Figure 2.2.1-1. The bonding wire is then
connected from the top contact of the die to the end of the
adjacent element, called a "bonding post". After the
bonding wires are in place, the lead frame, bearing a
number of devices is clamped into a fiXture which controls
precisely the distance by which the posts of the frame are
lowered into cavities that have been previously filled to a

LED CHIP, CONDUCTIVE

EPOXY DIE ATTACH,
BALL WIRE BOND ONTO
TOP CONTACT
EPOXY ENCAPSULATION
MAGNIFYING DOME LENS

CATHODE POST
REFLECTOR LEAD
FRAME USED ON
HIGH-EFFICIENCY
RED,
YELLOW AND GREEN.
REFLECTOR DISH
COINED INTO
CA THODE POST
(NO REFLECTOR
ON STANDARD
RED DEVICES)

WEDGE
WIRE
BOND

SI LVER PLATED
COPPER ALLOY
.-/'
LEAD FRAME.HER.
YELLOW & GREEN
(SOME STD.-RED
LAMPS HAVE SOLDER
DIPPED KOVAR
LEADS)

FLAT ON SIDE OF DOME

INDICATES CATHODE
LEAD

UPPER
DAM BAR

CATHODE LEAD IS
SHORTER THAN
ANODE LEAD
LOWER
DAM BAR

Figure 2.2.2-1

Construction

Features of a T-1 3/4 Plastic LED

Lamp.

0/

precise level with the uncured liquid epoxy. The shape of

these cavities, called "mold cups", determines the shape of
the lenses of the lamp packages, and the distance to which
the posts are inserted in the mold cups determines the
magnification and hence the radiation pattern of the
finished product. Dye is used to tint the epoxy to absorb
ambient light and make the lamp appear darker when it is
off; since the dye does not absorb appreciably the light
from the LED, this raises the ON/OFF contrast ratio. In
some devices, a diffusant material is added to the epoxy to
cause light scattering, thereby increasing the effective
viewing angle.

Curing of the epoxy is done in two stages. A pre-cure

hardens the plastic to a point which permits removal from
the mold cups and shearing of the lead frame and dam bars.
The post-cure is then applied, which further develops and
stabilizes the mechanical properties of the epoxy, enabling
the lamps to withstand the abuse they may later receive
when being installed and connected into circuits. Except
for testing, manufacture is now complete.

When LED lamp devices were first introduced,

two
established miniature sized packages were borrowed from
the incandescent lamp. These are the T-l and T-l 3/4 sizes.
In the TX designation, the number indicates the lamp
diameter in 1/8ths of an inch. These two sizes have now
become standard in the optoelectronic industry.

The construction features of a T-l 3/4 lamp are illustrated

in Figure 2.2.2-1. The lead frame for a GaP transpannt
substrate LED has a dish shaped reflector coined into the
top of the cathode post. A GaP transparent substrate LED
emits light from the sides of the chip as well as from the
top surface. The reflector directs this side emitted li,~t
towards the dome lens in an analogous fashion to :he
common hand held flasWight. The result is a lamp with a
front emitter equal to the larger area of the reflector dish.
.
.
A standard red lamp does not requIre a reflector, smc,~ a
GaAsP/GaAs LED emits light only from the top surface of
the junction.

TRUNCATED CONE
WITH
SPHERICAL LENS

LED CHIP, CONDUCTIVE

EPOXY DIE ATTACH,
BALL WIRE BOND ON
TOP CONTACT

WEDGE WIRE BOND

TO ANODE LEAD

A variation of the standard T-l 3/4 lamp is the low proJile

T-l 3/4 lamp. The construction is identical to the standad
lamp except for the height and shape of the dome lens. The
dome is 2/3rds the height of the standard lamp and is an
aspheric lens which provides high magnification coupled
with a wide viewing angle. For a visual comparison, ~,ee
Figure 2.1.8-3.
The construction of the T-l lamp is essentially the same as
for T-l 3/4 lamp. However, the reflector dish is too large to
be successfully encapsulated inside this small lamp package.
It is for this reason, that a more compact lead frame,
without a reflector, is used in all of the T-llamp products.
Initially, the two standard lamp sizes were used only as
indicators on front panels and printed circuit boards.
However, as the variety of uses expanded into other areas,
such as backlighting and various array configurations, the
T-l and T-l 3/4 packages could not be easily adapted to
meet the shape and size requirements for all of these more
sophisticated applications. As a result, new lamp packages
have been developed to conform to many of these needs.

Many applications require a very small, low cost LED lamp

to be used in a location where space is at a premium. Such
might be the case for the colon in a small desk clock, for an
indicator on a small hand held instrument or for use in an
array where a high packing density is required. To meet this
need, the subminiature lamp package has been develope i.
The subminiature device offers the customer a lamp that is
low in cost, smaller in size than a T-llamp and has superior
optical consistency device-to-device. The small size allov!s
for a very high packing density in an array, see Secticn
2.4.4 on Arrays. The optical consistency of each lamp is
derived
from the transfer molding process used t,)
encapsulate the device.
Figure 2.2.3-1 illustrates the construction features of th
subminiature lamp package. The subminiature package uses
a radial, flexible, rectangular lead frame suitable for

REFLECTOR
LEAD FRAME,
ALL COLORS
SILVER PLATED,
COPPER ALLOY
LEAD FRAME
Figure 2.2.3-1

Construction
LED Lamp.

Features

of a Subminiature

Plastic

handling in the transfer molding process. From a user's

point of view, this lead frame offers two important
benefits. First, unlike the T-l lead frame which does not
have a reflector, a small reflector cup is formed by bending
up the sides of the die attach and bonding pads. The optical
effect is similar to the T-l 3/4 lamp, producing a small lamp
with very high sterance. The second advantage is a choice of
installation procedures; the leads may be left in the radial
position for soldering to pads on the face of a printed
circuit board, or the leads may be bent 90 into the axial
position for insertion into plated through holes and
subsequent flow soldering or for insertion into a socket.
The package dome is constructed as a truncated cone with a
spherical lens to obtain high magnification
for high
intensity combined with a very wide viewing angle. The
package base is square shaped to provide firm support for
the lead frame, resulting in a high degree of mechanical
reliability, and to aid in positive alignment in high density
array applications.

The rectangular package is designed for use in those specific

applications where a cylindrically shaped device is not

effective. For instance, the rectangular LED lamp is most

effective for illuminating a legend by directly backlighting a
transparent character so that it will stand out more vividly.
As another example, a bar graph is easily implemented with
rectangular LED lamps, as they may be either end stacked
or side stacked to form a continuous bar of light.

LIGHT RAYS FROM LED TRAVEL THROUGH

LIGHT PIPE TO DIFFUSING LAYER. THE
DIFFUSING LAYER SPREADS LIGHT TO
FROM AN EVENLY LIGHTED RECTANGULAR
SURFACE

The construction of the rectangular lamp is an extention of

the T-l 3/4 package, as both use the same reflector lead
frame. As shown in Figure 2.2.4-1, the difference is in the
encapsulation. The encapsulating epoxy is formed into a
rectangular light pipe with an integral diffusing layer at the
top. The light pipe is the optical path through which the
LED light travels to diffusing layer. The diffusing layer
spreads the light to form an evenly lighted rectangular
source.

Although plastic devices are used in the majority of LED

lamp applications, they become vulnerable when placed
into adverse environments. The problem is the limit to
which the encapsulating epoxy can withstand temperature
extremes, moisture or other detrimental environmental
conditions without losing its optical properties or exerting
failure causing stresses on the LED die attach and wire
bonds. Therefore, a hermetically sealed device is used for
many military lamp applications and applications where the
lamp may be exposed to an adverse industrial environment.
The hermetic lamp is assembled in a TO-18 package as
illustrated in Figure 2.2.5-1. After die attach and wire
bond, a coating of silicone jell is applied. At the top of the
metal cap is an optical window with a hermetic seal at the
glass to metal interface. An epoxy dome is cast on top of
the optical window to increase viewing angle and to provide
good on/off contrast. The metal cap is welded to the TO-18
header to complete the assembly. The exterior of the
package is gold plated to resist corrosion.

The preceding sections have described the LED lamp

package as a device containing only one component.
Electrically, an LED lamp is a two terminal device and is
capable of utilizing a limited number of other components,
in addition to the LED chip, to perform certain functions.
Such devices are termed "integrated LED Lamps".
A commonly used integrated lamp is the resistor LED lamp.
In this device, the package contains an integral current
limiting resistor, that is die attached to the anode post and
wire bonded to the LED top contact. This integral resistor
is a nominal 215S1, allowing the lamp to be driven directly
from 5.0 volt supply when controlled by a TTL gate.

REFLECTOR CATHODE
POST WITH LED CHIP
TINTED EPOXY
LIGHT PIPE PACKAGE

SILVER PLATED
COPPERALLOY
LEAD FRAME

CATHODE LEAD IS SHORTER

THAN ANODE LEAD
Figure 2.2.4-1

Construction
LED Lamp.

Features of a Plastic Rectangular

Another integrated lamp device is the voltage sensing LED

lamp. In this device, an integrated circuit turns the lamp on
or off depending upon the level of the applied voltage. A
reference threshold level, VTR is built into the integrated
circuit. If the applied voltage is above VTR the lamp is ON,
and below VTR the lamp is OFF. This device is commonly
used as a battery test indicator.
Figure 2.2.6-1 shows schematic representations
two integrated lamp devices.

for these

A lamp data sheet contains specific characterization

information to aid the designer in selecting the correct lamp
for his application, determining the maximum worst case
operating
limits and establishing nominal operating

TINTED DIFFUSED EPOXY

DOME PROVIDES WIDE
VIEWING ANGLE AND
GOOD ON/OFF CONTRAST

Once the lamp package configuration has been selected, the

designer is ready to consider those characteristics of most
importance:

Other considerations may also be of importance such as

reverse
breakdown voltage, capacitance or speed of
response, depending upon the application.
For a designer to be able to effectively utilize the
information contained in a data sheet, he needs an
appreciation of what the numbers mean and on what basis
they have been derived.
BLACK GLASS
INSULATOR

conditions. The data sheet contains this information in the

following basic sections:

Of initial concern to a designer is the selection of that lamp

which is of the proper color and has the most light output
at a specified current. The electrical/optical characteristics
contain four parameters that quantitatively aid the designer
in making this selection. These are peak wavelength,
dominant wavelength, luminous intensity and the included
angle between half intensity points.
The color of the lamp, as perceived by the eye, is
dependent upon the radiated spectrum of the LED. Not
only is the radiated
spectrum of importance
for
determining color in visual applications, but it is very
important in determining the coupling to a detector in
non-visual applications. Peak wavelength and dominant
wavelength are the two quantitative parameters which
describe the radiated spectrum.

E
I

IZ

c::
c::

LED
OFF

40

---.r-

LED,
ON
I

30

::::>

u 20
I-

::::>
Q.
Z

10

I
GND
RESISTOR LED LAMP,
STANDARD RED

VOLTAGE SENSING
LED LAMP
STANDAF:D RED
Figure 2.2.6-1

3
VTH

VI -INPUT

Schematic Representation of Two Common

Integrated LED Lamps.

VOL TAGE - V

Peak wavelength, Ap' is that wavelength at the peak of the

radiated spectrum. It is maintained within narrow limits
during the growth of the epitaxial layer, with each wafer
being inspected by a photo-luminescence measurement on a
production
basis. To the designer of a non-visual
application, Ap becomes important when determining the
coupling efficIency of the LED light to a photodetector,
signal loss through a fiber optic conductor or photographic
film sensitivity. In visual applications, the total amount of
LED emitted light passing through an optical filter, used for
contrast enhancement, is approximately equal to the filter's
relative
transmission
at Ap. Therefore, the relative
transmission at Ap is a quantitative measure of the optical
density of a contrast filter.

Luminous Intensity, Iv, is a measure of axial light output.

The unit of measure is the candela, which is luminous flux
per unit solid angle. Thus, the measure of luminous
intensity gives no information concerning the radiation
pattern. The axial luminous intensity of each device is
measured
on a production
basis, using a calibrated
photometer, to insure that it meets a minimum value at a
specified drive current.

The color of an LED device is especially important to the

designer if lamps from various manufacturers
might
possibly be installed in the same array. Not only does the
eye detect intensity differences, it also detects color
differences. The dominant wavelength, Ad' is a quantitative
measure of the color of an LED device as perceived by the
eye. Two devices of somewhat different radiated spectra
will appear as the same color if they both have the same Ad.
The dominant wavelength is not necessarily dependent
upon peak wavelength. Conceptually, Ad may be envisioned
as that wavelength near the centroid of the radiated
spectrum. Specifically, Ad is that wavelength when mixed
with an equal amount of the light from a 65000K lamp will
be perceived by the eye as the same color as is produced by
the radiated spectrum of the LED.

The importance of the radiation pattern to a designer is

that it defines the apparent luminous intensity of the device
when viewed at some off-axis angle. The radiation pattern is
quantitatively described at a single point, which is the
included off-axis angle where the luminous intensity is
equal to one half of the axial intensity. This included angle
is referred to as 28~. The value of 8 ~ is controlled by the
distance of the LED from the focal point of the dome lens.
Larger values of 28~ result in a lamp which may be viewed
at greater off axis angles. Also, a lamp which has a greater
28~, but the same axial intensity will actually have a much
greater total flux. A good example of this is the comparison
of two undiffused T-1 3/4 LED lamps that have the same
axial luminous intensity but differ in the value of 28 ~_ The
first lamp has 28~ = 35 and the second lamp has
28 ~ = 25. Depending upon the exact shape of the two
radiation patterns with respect to each other, the first lamp
could produce up to three times the luminous flux that is
produced by the second lamp. An examination of the data
sheet curve of the radiation pattern reveals the intensity at
any specific off-axis angle, as illustrated in Figure 2.3.1-1.

It has been common practice of many designers to use the

data sheet luminous intensity as the only parameter
necessary to evaluate light output. This is not the only
important consideration in the determination of the total
available luminous flux. A designer needs to consider both
the axial luminous intensity and the radiation pattern in
making a determination of total light output. It is the
radiation pattern that has the most influence.

Relying only on a minimum or a typical intensity value for

lamps used in an array does not guarantee intensity
matching. It is best to install lamps that have been
categorized for light output. Each light output category
should have an Iv MAX/Iv MIN ratio of 2: 1 or less. A
single array is then assembled with all lamps from the same
category. To an observer, the lamps will appear evenly
illuminated throughout the array.
In summary, peak wavelength, dominant wavelength, axial
luminous intensity and radiation pattern are necessary to
completely specify the optical characteristics of an LED
device.

Once the optical characteristics of an LED device have been

established, the next step is to determine the maximum
temperature derated limits at which the device may be
operated. Forward current, power dissipation, thermal
resistance and LED junction temperature are all interrelated
in establishing absolute maximum ratings.
Figure 2.3.1-1

Radiation Pattern for a T-1 3/4 High-Efficiency

Red LED Lamp.

The absolute maximum ratings of an LED device have been

determined theoretically and by extensive reliability testing

and are those limits beyond which reliable operation cannot

be assured. These maximum ratings cannot be taker as
limits in concert, as operation at one maximum limit may
preclude the operation at another maximum limit. The hest
example of this exclusion rule is that a device cannot be
reliably operated at maximum power dissipation in the
maximum allowed ambient temperature. Operation at the
maximum ambient temperature is only allowed with proper
power derating.

120
110
;;:100
E

z 90
0

i=
<t

80

0..

gj
0 70

a::
w
;;:

60

0
0..

Maximum ratings are based on package temperature

limitations and the current density limit within the LED
junction. The package temperature limitation may be ba,ed
on the glass transition
temperature,
TG, of the
encapsulating epoxy. Above TG' the cross linkages between
the epoxy molecules change allowing the molecules to
move with respect to each other. The epoxy changes fwm
an ordered structure to an amorphous structure, analogous
to that of glass, and the coefficient of thermal expanSlOn
drastically increases. The result is that thermal stresses
above TG may be sufficient to cause a catastrophic failure.
The maximum current rating is based on (1) the level of
current density within the LED junction that produces no
more than
an acceptable
amount of light output
degradation, and (2) the amount of power dissipation
generated by the forward current that maintains the LED
junction at an acceptable temperature level below TG,
The limiting
factor for operation in any ambient
temperature is the LED junction temperature, TJ' For a
hermetic device, TJ should be less than 12Soe to realize In
acceptable rate of light output degradation, and for a
plastic device, TJ should be at least 100e below the TG of
the encapsulating epoxy to prevent catatrosphic failure.
The encapsulating epoxies currently being used in the
manufacture of LED lamps have a nominal TG = 120C.
The two parameters that effect TJ' which a designer has
control of, are forward current and thermal resistance. TIle
maximum forward current is derived from the maximum
allowed temperature derated power dissipation and the
thermal resistance is dependent upon the method used to
install the lamp in the circuit.
The maxim um allowed temperature
derated power
dissipation is obtained by derating the absolute maximum
power rating at the rate of -1.6 mwte above an ambient l)f
SOe. This is a linear derating from maximum power at an
ambient of SOe to zero power at an ambient of 12Soe, IS
shown in Figure 2.3.2-1.
Average power dissipation is the product of average forwaJ d
current and peak forward voltage. An equivalent circuit for
an LED, consisting of a dc voltage source in series with a
dynamic resistance, may be derived from the diode's
forward characteristics. Also, a lamp data sheet lists a vallJe

<:J
<t
a::

50

>

40

\,
\

<t

<:J 30

>
<t

0..

20

10

\
10

20

30

40

50

60

TA' AMBIENT TEMPERATUREC

Figure 2.3.2-'

Maximum Average Power Derating for a Plastic

LED Lamp.

of forward voltage at a specific current. From these two

observations, the following equation may be derived to
calculate the power dissipation in an LED lamp:

where:

'AVG

Average forward current, amperes.

Ip = Peak forward current, amperes.

Rs

LED dynamic resistance, ohms.

V F(DS)
IF(DS)

=
=

Data sheet forward voltage, volts.

Data sheet forward current where

V F(DS) is specified, amperes.

For standard red devices, Rs = 1.6[2 typical and S[2

maximum; for GaP transparent substrate devices Rs = 21[2
typical and 3S[2 maximum.
Once the maximum tolerable operating conditions have
been established for pulsed operation, as described in
Section 2.3.3, equation 2.3.2-1 may be used as a check to
insure that the average power dissipation does not exceed
the maximum derated limit. For dc operation, the dc
power
dissipation
may be calculated
by setting
IDe = IAVG = IpEAK
LED junction temperature is the sum of the operating
ambient temperature and the temperature rise above this
ambient.
T J (oC) = T A + fiT J

TJ

(0 C) =

TA + e JA pAVG

where:

= Ambient

TA

temperature immediately
surrounding the lED lamp, DC.
PAVG = Average power dissipation, Watts.
e JA = Thermal resistance lED junction-to-ambient
of the lamp installed into the circuit, C/W.

loglog scale to form the family of curves shown in Figure

2.3.3-1.
The curve for <iny specific refresh rate is the locus of
maximum tolerable operating conditions which maintain
the limitation of TJ' Any combination of operating

The value of e J A is the sum of the device thermal

resistance, LED junction-to-lead, e JC' and the thermal
resistance to ambient of the supporting structure, e CA:

For most printed circuit boards, e CA ranged between

3SoC/W
and SOC/W, depending upon metallization
pattern. Figure 2.3.2-2 illustrates this concept.

10 KHz
30 KHz
100 KHz

A numerical example utilzing these equations is presented

in Section 2.4.3.
8JA CC/W) = 8JC +OCA
TJ ("C)
LED JUNCTION
TEMPERATURE,

TA +8JA PAVG (W)

TJ (OC)

8 JC (OCIW)

THERMAL RESISTANCE
JUNCTION- TO-LEAD
(DATA SHEET VALUE)
Maximum Tolerable Peak Current vs. Pulse
Duration for a T-' 3/4 High-Efficiency Red LED
Lamp.

eCA
PRINTED/
CIRCUIT
BOARD

(oC/W)

THERMAL RESISTANCE
PC BOARD TO AMBIENT,
LAMP SOLDERED
INTO BOARD
(MEASURED BY USER)

AMBIENT AIR
TEMPERATURE, T A (OC)
Figure 2.3.2-2

A Schematic Representation
of the Thermal
Resistance Paths for a Plastic LED Lamp.

When a design requires an LED lamp to be operated in the

pulsed mode, maximum tolerable operating limits need to
be established. These maximum tolerable limits should not
raise the LED junction temperature above that which
would be obtained by operating the lamp at the maximum
dc current. This limitation on TJ imposes a definite
interrelationship between peak current, pulse duration, and
refresh rate. This interrelationship is most easily obtained
by establishing combinations of peak current and pulse
duration
for various refresh rates, maintaining the
maximum TJ at that value obtained by operating at the
maximum dc current. These data points are plotted on a

conditions at, or below, a line of constant refresh rate is

permissible. Operation above a line of constant refresh rate
violates the limitation on junction temperature.
One procedure
for determining a set of operating
conditions from Figure 2.3.3-1 consists of the following
five steps:
Step 1. Determine the desired duty factor, DF. Example:
DF= 30%.
Step 2. Determine the desired refresh rate, f. Use DF to
calculated pulse duration, tp. Example: f = 1 KHz; tp =
DF /f = .30/1 KHz = 300 Ilsec.
Step 3. Enter Figure 2.3.3-1 at the calculated value of tp.
Move vertically to refresh rate line and record the
corresponding value of IpEAK MAX/IOC MAX' Example:
At tp = 300 Ilsec and f = 1 KHz, IpEAK MAX/IDC MAX =
2.4.
Step 4. Calculate IpEAK MAX using the data sheet
maximum average forward current for IOC MAX. Calculate
IAVG from IpEAK and DF. Example: from data sheet, the
maximum average current = 20 mA.

is the mathematical representation of this curve, with the

value of n having been derived directly from the curve.

IpEAK MAX = (2.4)(20 mA) = 48 mA

lAVe = (.3)(48 mA) = 14.4 mA
Step 5. Refer to Equation 2.3.2-1 and calculate PAVG as
a check to insure that the above operating conditions are
within the required power derating corresponding to the
operating ambient temperature. Should PAVe fall above
the required power derating, decrease tp to reduce IAVe
(or reduce IpEAK) in order to lower PAve to an
acceptable level.
An alternate procedure is presented
numerical example in Section 2.4.3.5.

in the form of a

The axial luminous intensity value listed on the data sheet

is a dc measurement. The measurement is made by driving
the lamp at the specified current for a time duration of 20
msec to 50 msec and measuring the intensity with a
calibrated photometer. The luminous intensity of e<ch
lamp will be at least equal to the specified minimun, but on
the average will be close to the typical value listed on the
data sheet. This measurement does not give a designer any
insight as to what the intensity will be at some other dnve
condition.
Of interest to a designer is the light output, as perceived by
an observer, when the lamp is driven at some other dc
current or is operated in the pulsed mode. Since the eye i~ a
time averaging detector, it is the time average lumino lS
intensity that is of specific interest.
For dc operation, the time average intensity is a supralinear
function of forward current.
I

v _(

Ivo

I
n
F )
IFO

Ivo

An initial luminous intensity at a reference

dc current, IFO'
Iv = The luminous intensity at the operating dc
current, IF'
n = A supralinear exponent that ranges between
1.1 and 1.4, depending upon the LED product
and the forward current.

The determination of time average intensity, observed with

pulsed
operation,
involves the concept of relative
efficiency. Once the concept of relative efficiency is
understood, a designer is able to calculate the time average
intensity for any given set of pulsed operating conditions.
The efficiency of an LED device may be stated as intensity
per unit current, such as millicandelas/milliampere. Relative
efficiency is the ratio of the efficiency at one peak current
to the efficiency at another peak current for the same
average current.
A graph of relative efficiency is obtained in the following
manner. An LED device is measured for luminous intensity
at the test current specified in the data sheet. Then the
device is strobed at different peak currents, but with the
same average current as specified in the data sheet, and
measured for luminous intensity. The resulting curve of
luminous intensity vs. peak current is normalized to 1.0 at
the data sheet test current. This is the curve of relative
efficiency for that device. The data sheet relative efficiency
curve represents the average of a large sample of devices
from many different lots.
To obtain the time average intensity of an LED lamp being
operated in the pulsed mode, a designer need only multiply
the ratio of the operating average current to the data sheet
average current by the product of the relative efficiency at
the operating peak current and the luminous intensity
specified on the data sheet.
Section 2.4.2 on relative efficiency illustrates the use of the
two data sheet graphs, Relative Luminous Intensity vs.
Forward Current and Relative Efficiency vs. Peak Current,
in determining time average intensity.

Equation 2.3.4 is derived in the following manner. A

measurement of luminous intensity at various drive currents
is obtained from a large sample of devices selected from
many different lots. A curve of relative luminous intensity
vs. forward current is derived from this data, normalized to
1.0 at the current where the luminous intensity is specified,
and presented as a graph on the data sheet. Equation 2.3.4

The largest usage of LED lamps is in visual applications.

LED lamps have long been used as panel mounted
indicators, printed circuit status indicators, and both x-y
addressable and linear arrays. With the introduction of the
High Efficiency LED lamps, applications that previously
could only use neon and incandescent lamps, now in many
instances, can also use LED lamps. These high intensity
applications include backlighting a legend or illuminating a
push button. LED lamps offer many advantages to the
designer. They are small, light weight, and mechanically
rugged. Since they operate at low voltages and currents,
they can interface directly to most digital logic families.
Because LEOs are solid state devices, they have a projected

operating life of over 100,000 hours. These features benefit

the end user by substantially reducing field maintenance
costs due to lamp replacement.
Figure 2.4.1-1 shows some of the traditional uses for LED
lamps, such as panel indicators and printed circuit status
indicators. The designer has the flexibility of soldering the
lamps directly into a printed circuit board and positioning
the board behind the front panel as shown, or by using a
clip and ring to attach the lamp to the front panel and
soldering or wirewrapping leads to the LED lamp. When
several lamps are driven by a common LED driver, they are
commonly called an array. While it is not important that
the lamps be positioned together, special purpose displays
can be formed by specific patterns of lighted lamps in a
specific grouping. An x-y addressable array is a group of
LED lamps that are connected so that one particular lamp
is illuminated through the application of proper signals to
an x and a y coordinate. Numeric and alphanumeric
information can be displayed with 35 LED lamps arranged
in a 5x7 matrix. Fewer LEDs are required if only numeric
information is to be displayed (an example of a modified
4x7 matrix to display hexidecimal information is illustrated
in Figure 5.1.1.1-1). Analog information can be displayed

with a linear display. Two types of linear displays are the

bar graph display and the position indicator display In the
bar graph display, all LED lamps that represent a value
smaller than the input quantity are turned on. Only the
single LED lamp that represents a value closest to the input
quantity is turned on in the position indicator display.
Another application of LED lamps is to highlight or
backlight a printed legend. Traditionally, LED lamps have
been used to attract the attention of a viewer to a message
printed near the LED lamp. The introduction of the High
Efficiency LED lamps, now allows LEDs to backlight a
legend printed on translucent or diffused mm.
Electrically, LED lamps behave similarly to silicon or
germanium diodes. LED lamps emit light only when they
are forward biased. Because the LED lamp has a very small
dynamic resistance above the turn-on voltage of the device,
LEDs are normally driven by a current source. For most
applications, a battery in series with a resistor can be
considered as a current source if V BAT ~ VF and R ~ RS'
Figure 2.4.1-2 shows how the current flowing through an
LED lamp can be solved graphically by superimposing a
load line over the forward characteristics of the LED
device. The forward current through the LED should be

CIRCUIT BOARD STATUS

OR MAINTENANCE
LIGHTS

REGISTER READOUT
LIGHTS
BACKLIGHTED
ANNUNCIATOR
LEGEND LIGHTS

POWER SUPPLY
STATUS LIGHTS

BACKLIGHTED

MTG. SCREW

PUSHBUTTONS
CLlNCH\

SPACER
FRONT PANEL
P.C.CARD

TYPICAL LED MOUNTED IN A PANEL

CLIP WITH WIRE WRAP LEADS

PANEL MOUNT
Uses of LED Indicator
Instrument.

Lamps in a Complex

selected to give the desired time averaged luminous

intensity for worst cased values of power supply voltages
and circuit tolerances. Maximum forward current is also
constrained by the maximum allowable average p )wer
dissipation
based on ambient temperature
and the
maximum tolerable peak current for a specified pulse
duration and repetition rate. Section 2.4.3 outlines some of
the techniques to operate an LED lamp in dc or pulsed
mode and shows several worst cased circuit designs using
these techniques.

Efficiency vs. IpEAK" that are shown in the data sheet for
that
particular
device. Section 2.4.2 describes these
calculations in more detail.

Traditionally, LED lamps have been characterized for light

output (IV) vs. dc forward current (IF)' An example of this
characterization is shown in Figure 2.4.2-1. The abcissa is
dc current in milliamperes and the ordinate is luminous
intensity in millicandelas or luminous intensity normalized
to one at a particular value of input current. The use of
such a figure is relatively simple if one variable is known.
For example, if an LED lamp emits 4.0 mcd at 10 mA IF'
then typically it will emit 9.9 mcd at 20 mA IF'

Since an LED lamp is an optoelectronic

device, the
luminous intensity, radiation pattern, and visual appearance
of the lamp should dictate the choice of LED packagef and
whether the lamp should be diffused, non diffused, tinted
or non tinted. The luminous intensity of the lamp should
be large enough to achieve a desired contrast ratio between
the lamp and the background around the lamp. A lamp
mounted on a reflective surface will require a higher
luminous intensity than the same lamp mounted on a dull,
non reflective surface. If the ambient luminous incidence is
increased, the luminous intensity of the lamp will ha\'e to
increase
proportionately.
Once the desired luminous
intensity is known, the required drive current to obtain that
luminous intensity can be calculated using either the figures
of "Relative Luminous Intensity vs. IF" or "Relc.tive

Two other representations of Figure 2.4.2-1 are useful in

strobed applications. An LED lamp will emit a certain
number of photons per milliamp of input current. This is
known as the efficiency of the LED. This efficiency varies
according to the peak current through the LED. The
relative efficiency curve such as Figure 2.4.2-2 is used to
show this relationship. The abcissa is peak current in
milliamps and the ordinate is millicandelas per milliamp
normalized to one at a particular value of input current.

NOMINAL LOAD LINE, 220n 10% CURRENT ~I

LIMITING RESISTOR, 5V SUPPLY
2:2

HI

I-

z
w

II:
II:

=>

U
Cl
II:

13.2 mA.;;; IF';;; 14.1 mA, R = 220n

12.2 mA.;;; IF';;; 15.7 mA, R = 220n 10%

Hi
1L\

1"~.

II:

0
u..

10

u..

AS SHOWN IN FIGURE

8
6
4

.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

VF - FORWARD VOLTAGE - V
Figure 2.4.1-2

Graphical Solution of the Current

a High Efficiency Red LED.

Flowing through

5.0

5.5

6.0

Figure 2.4.2-2 can be used to calculate the time averaged

luminous intensity for any particular peak current and duty
cycle with equation 2.4.2-1.

(2.4.2-1)
[lPEAK]
IV TIME AVG =

[DUTY CYCLE] [T/(IPEAK)] [IV SPEC'

[lSPEC] [T/(IPEAK)]

where IpEAK is the desired peak current; DUlY CYCLE is

the ratio of time the LED is "ON" to total time; T/ is the
relative efficiency of the LED at IpEAK or at ISPEC;
ISPEC is the dc current at which IV SPEC is tested.; and
IV SPEC is the luminous intensity of the LED accordlllg to
the test conditions.

1/
/

For example, using Figure 2.5.2-2, if the desired IpEAK is

60 mA, the desired duty cycle is 1/8, and the LED has a
luminous intensity of 4.0 mcd at 10 mA dc, then the time
averaged IV can be calculated as follows:

/
/

= (60 mAl (.125) (1.54) (4.0 mcd)

"

./

(10 mAl (1.00)

IV TIME AVG (mcd)

= 4.6

mcd at 60 mA IpEAK- 1/8 DUTY CYCLE

10

12

14

16

IF-FORWARDCURRENT-mA

One final representation of Figure 2.4.2-1 is a time averaged

luminous intensity curve. An example of this representation
is shown in Figure 2.4.2-3. The abcissa is average current in
milliamps and the ordinate is time averaged luminous
intensity. Time averaged luminous intensity is then plotted
for different values of peak current or duty factor. With
this representation, time averaged luminous intensity can be
read directly from the curve. For example, if the desired
IpEAK is 60 mA, the desired duty cycle is 1/8 and the LED
has a luminous intensity of 4.0 mcd at 10 mA dc, then the
time averaged IV can be read directly off Figure 2.4.2-3 as
4.6 mcd (IAVE=7.5mA).
For lamps with a different
luminous intensity than 4.0 mcd at 10 mA dc, the final
result will need to be linearly scaled to reflect the
difference. Figure 2.4.2-3 shows the advantages of strobing
the LED as a means for achieving a higher luminous
intensity at the same average current or by reducing the
required average current and still maintaining the desired
luminous intensity. For example, a typical device will emit
1.55 mcd at 5 mA as shown by Figure 2.4.2-3. The same
device will emit 3.1 mcd at 60 mA peak,S mA average
current or 1.55 mcd at 60 mA peak, 2.5 mA average
current.
Ignoring the effect of junction heating due to the average
power dissipation in the LED (the temperature coefficient
of IV is about -1%tC), Figures 2.4.2-1,2.4.2-2, and 2.4.2-3
are equivalen 1. The designer can derive one figure from
either of the other two if he prefers. The relative efficiency

Relative Luminous Intensity

Efficiency Red LED.

vs. Current

......... ~
U
o

1.4

/"

:-1

5~ 0 1.3

tj~
-->5~
"0

,w

./

1.2

--

."

/'

"N

"::;t
,.::;; 1 .1
0::

-'0
"z 1.0
c:_

.....

for High

r/

15

20

25

30

35

40

45

50

IpEAK (mAl

Relative Efficiency vs. Peak Current for High

Efficiency Red LED.

55

curve is normally given on all data sheets. As a final

example assume a lamp is strobed at 20 mA peak current at
a 10% duty cycle and the lamp has a luminous intensity of
3.0 mcd at 10 mA dc:

Figure 2.4.2-2: Iv

(20)(.10)(1.24)(3.0)
(10)( 1.00)( 1.00)

= .74

'B
.s

mcd

>
l-

e;;
Figure 2.4.2,3:

Iv

(1.0)(3.0)

Z
UJ

.75 mcd

~
en
::>

4.0

a
z
::;:

60 mA -'PEAK
I

40 mA - IpEAK

::>
...J
fiJ

t?

<l:

cc

Figure 2.4.3.1-1 shows typical electrical characteristics :'or

standard red, high efficiency red, yellow and green lam :JS.
Above 1.5 volts VF, the current flowing through an LED
increases very rapidly. The dynamic resistance can be
considered to be the slope of the diode characteristic
(t.VF/t.IF) in the forward region. The standard red lanp
has a very low dynamic resistance, while the high efficien~y
red, yellow and green lamps have a somewhat higher

UJ

~
UJ

::;:

i= 1.0
...J
<l:

l1.

.8

>
I-

dynamic resistance. Since the dynamic resistance is so

small, LED lamps should not be connected in parall ~l.
Small variations in V F or dynamic resistance can cause
current hogging by the LED with the lowest VF. This
current hogging can cause variations in luminous intensity
and excessive power dissipation in the lamp. However, LED

Figure 2.4.2-3

Typical Time Averaged Luminous Intensity vs.

Average Current for a High Efficiency Red LED.

<l:

<l:

I
I

IZ
UJ

50

I-

40

cc
cc
::>
<.J

cc
cc
::>
<.J

UJ

Cl

Cl

cc

30

<l:

:::
cc

a
u.

20

,
I
I
I
I

I
I
I
I
I

<l:

:::
cc
a
u.
u.

10

I
I

cc

-1

_VF

.5

1.0

1.5
I

2.0
,

-. FORWARD VOLTAGE - V

2.5
~

I
-2

I
I

.5
-1_

-2.

-3.

1.0
1.5
2.0
2.5
3.0
I
I
I
I
VF - FORWARD VOLTAGE - V

3.5

lamps can be connected in series as long as the combined

VF doesn't exceed the power supply potential.

If VCC is considerably larger than V F + VCE SAT' then

small variations of VF or VCE SAT will have only
negligable effects on IF' However, variations of VCC or R
will cause a corresponding variation of IF' For example,
suppose that a standard red LED lamp is to be driven from
a 5.0V supply by a transistor switch with a VCE SAT of
.2V at 18 mA IF' Then the nominal value of resistance is
equal to:

Negligable current flows through an LED in the reverse

direction until the breakdown voltage is exceeded. Above
the breakdown voltage, BVR, the reverse current increases
very rapidly, such as shown in Figure 2.4.3.1-1. Exceeding
the BVR will not harm the LED lamp as long as the reverse
current is externally limited to prevent excessive power
dissipation in the LED. When several LED lamps are
connected in an array, reverse leakage current can cause
unwanted ghosting of normally off LED's. This can be
prevented by using LED drivers with a high impedance off
state.

R = 5.0 -.2 -1.65

18

175n

Figure 2.4.3 .2-1 shows both numerical and graphical

solutions to this example assuming tolerances of VCC, R,
VCE SAT' and VF
Resistor-LED lamps are also available that are designed to
operate off of a 5 volt supply.

When LED lamps are driven from a regulated power supply,

a resistor can be used to limit the current flowing through
the LED. The LED current, IF' is determined by the
following equation:

In some applications, it may be desirable to drive LED

lamps with a current source. The current source can be used
to regulate the current through the LED regardless of
power supply variations or variations in VF between LED
lamps. Figure 2.4.3.3-1 shows some examples of simple
current sources constructed of npn transistors. For both
circuits, the current through the LED string remains
constant as long as VCC is greater than VCC (min)" The

where VCC is the power supply potential, VCE SAT is the

"on" voltage of the LED driver, IF and VF are the forward
characteristics of the LED, and R is the current limiting
resistor.

,,"-IF
:\
I

:!4

I (
= vCClmax) - VCE SAT lmin)' VFlmin)
F maxI
R(min)

:!2

I ( . ) = VCClmin) - VCE SATlmax) - VFlmax)

F mlO
R(max)

:W

1S0

,
'" \, '\\.\

NUMERICAL SOLUTION:

= 26.7 mA

=~

..

,\.

IS

I-

ASSUME VF(min) = 1.62@20 mA

a:
a:
::>

VF(max) = 1.67 @20 mA

VCC = 5.0V 10%
R = 1S0n 10%

VCC = 4.5V, IF(max) = 4.50 - .20 - 1.62 = 16.5 mA

(1S0)(.9)
. ) = 4.50 - .201.67
mlO
(1S0)(1.1)

14
12

a:

10

u-

0
u-

= 13.3 mA

VCC = 5.0V:. 15.S mA.;;; IF';;; 19.6 mA

VCC = 5.5V:. 1S.3 mA.;;; IF';;; 22.7 mA

IF(max) @ R(min)

-,

;z-

IF(max) = 17.7 mA

'\",

IF(min) = 17.4 mA

\."""li

a:

VCE SAT = .2V@ 20 mA

IF(

16

,~

IF(min) @ R(max)

~
'.\
'..

.--

"\

,
-.-~
"" ~,

~
'.\ I"

_SLOPE = 1= _1R 1S0

'~

~
o

.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

t 5.0

VCC' VCE = 4.SV

Figure 2.4.3.2-1 Numerical and Graphical Solution to Resistive
Current Limiting.

VF - FORWARD VOLTAGE - V

5.5

6.0

first circuit uses an LED lamp as a voltage reference. Since

~VF{T) == ~VBE(T) == -2 mVtC, the current source will
remaIn stable over temperature. In the second circuit, slnce
VBE varies with temperature, the current source will also
vary with temperature. However, this change is typically
about:

748:

IOL";;;; 20 mA

74LS:

74H: IOL";;;; 20 mA

74L:

74:

IOL ,,;;;;
S mA
IOL";;;; 3.6 mA

IOL";;;; 16 mA

If additional current is required, such as in a strobed

application, TTL buffers are also available. CMOS buffers
can also drive many LED lamps directly. Table 2.4.3.4-2
lists some of the commonly available CMOS buffers:
RCA 4049, 4050

+Vee

II
lice

VREF

Vee'

VREF'

1.5V for GaAsP LED

1
R

4009,4010

IOL~SmA, VOL =.5V, V CC=1OV

IOL-SmA,

VSE

Ice

IOL -20mA,

Ice

CAN

Vee

REPLACE

ZENER

VOL=5V, VCC=4.75V
VOL =.5V, V CC=10V

VREF + n VF

National 74C901
74C902

Ice = VREF - VSE

R2
I

VOL =.4V, VCC =5V

VREF
1 mA

Vee(min)"

GaAsP LED
SUCH AS
HP 50824100

IOL~3mA,
}

DIODE

=.4V,

. VREF
R1

LED1
IF

} IOL~3.SmA,VOL

WITH

DESIRED

+Vee

II
R

R
2

VSE

Ice
(Vee(min)

. 2VSE

- n

VF}(~2

1)

Ice

SERIES
SWITCHING

SHUNT
SWITCHING

ACTIVE PULLUP - TOTEM POLE GATES

Figure 2.4.3.3-1

Some Examples of Constant Current LED Drivers

that Regulate ICC Regardless of VCC.

MAY BE
REQUIRED"
Since LED lamps operate at low voltages and currents, they
can interface to most digital logic families directly. Figu re
2.4.3.4-1 shows some of the ways that an LED lamp can be
,,!sed to interface to digital logic. TTL logic families ai'e
guaranteed to sink a minimum amount of current (1m)
which can drive most LED lamps without additioml
buffering. Table 2.4.3.4-1 lists the guaranteed I
ofmo:;t
OL
common TTL families:

SERIES
SWITCHING

SERIES SWITCHING:
GATE SUPPLIES
CURRENT TO LED

Low cost transistors can be used when larger currents are

required than a logic gate can supply directly. Figure
2.4.4.2-3 shows some of the commonly used LED drive
schemes implemented with NPN, PNP, and FET transistors.

Regardless of whether the LED lamp design uses resistive or

constant current limiting, is DC driven or strobed, the
designer should consider worst case design techniques. The
purpose of worst case design is two-fold. First, the design
should guarantee that the LED lamps will operate within
their recommended operating conditions. This will insure a
long operating lifetime for the LED devices -- probably
longer than the expected lifetime of the instrument.
Secondly, worst case design can determine whether power
supply, resistor and device tolerances will cause a noticeable
variation in luminous intensity from lamp to lamp. In many
a pplications,
this requirement for luminous intensity
matching is not important, but when several LED lamps are
used in a closely packed array, wide variations in luminous
intensity can be very objectionable to the viewer.
When a lamp is dc driven, a worst case design analysis can
determine
whether
the lamp operates within its
recommended
operating
conditions. The maximum
operating conditions for an LED lamp under dc operation
are determined by two factors. Normally each LED device
is derated
over temperature
to keep the junction
temperature
below a specified maximum temperature.
Typically, this maximum junction temperature is 110C for
a plastic encapsulated lamp. Secondly, the current density
through the LED junction should be limited to prevent
rapid degradation of light output. The maximum peak
current for a standard red lamp is typically 1000 mA and 60
mA for high efficiency red, yellow, and green lamps.
Depending on the maximum expected ambient temperature
for the particular application, the designer can determine
the maximum average power dissipation, PAve' from the
power derating curve for the device. An example of a power
derating curve is shown in Figure 2.3.2-1. The average
power dissipation in a lamp is the product of the average
forward current times the peak forward voltage. Equation
2.3.2-1 can be used to calculate the average power
dissipation in an LED lamp:

where IAve is the average forward current, IpEAK is the

peak forward current, IF and VF are the LED lamp test
conditions, and RS is the LED dynamic resistance.
In a dc application, IAVe is equal to IpEAK. Equation
2.3 .2-1 and Figure 2.3 .2-1 can be used to calculate the

maximum
below:

dc current

through

the LED lamp as shown

(2.4.3.5-1 )

PAVG(mW) = 'AVG(mA)

[VF+Rs (IAVG(mA)-'F)/1000]

For a standard red lamp, VF (max) = 2.0V

RS (max) = 5Q
:. PAVG = I

AVG

.
_
.. IAVG(mA) -

[2.0 + 5 (IAVG -20)/1000]

J<(1.9) 2 + (.02) PAVG(mW)

20 mA If'

(2.4.3.5-2)

-1.9

.01

So at 50 C since PAve = 120 mW, then lAVe = 55 mA

o
and at 70 C where PAve = 87 mW, then lAVe = 41 mA.
Since the maximum average current is speCified in the data
sheet as 50 mA, then lAVe should be restricted to less than
50 mA regardless of the results of equation 2.4.3.5-3. For
additional reliability, the designer can operate at currents
0

less than IAve (max)

j
co
<{:;;

20

a::J

':j~1o
~ a:
~ 10
I:;; :;; a: 8
:JO:J
:;; I- U

~~UJ

O:J<{

-I-u
Xzo

u.a:a::
oUa:
-~UJ

~:56
a: <>.1-

100
tp

Figure 2.4.3.5-1

- PULSE DURATION

1000
-I1S

Maximum Tolerable Peak Current vs. Pulse Duration

for a T-1 3/4 Red Lamp (Ip'C MAX per Figure
2.3.2-1 and Eqn. 2.4.3.5-3

For strobed applications, a curve of maximum peak

current, pulse width, and repetition rate can be used to
determine
the maximum
recommended
operating
conditions for an LED lamp. The curve is determined by
comparing the peak junction temperature of a lamp under
strobed conditions to the average junction temperature
under maximum allowable dc conditions. At any operating
point, the peak junction temperature should not exceed the
average junction temperature under maximum allowable dc
conditions. An example of such a curve is shown in Figure
2.4.3.5-1. IOC MAX is the maximum average current
calculated by equation 2.4.3.5-1 or as restricted by the data
sheet. The ordinate of Figure 2.4.3.5-1 is the ratio of
maximum allowable peak current to maximum allowable dc
current. At any specified repetition rate, the relationship
between maximum peak current and pulse width is shown.
For reliable operation, the device should be operated at or

2.50
DESIRED CIRCUIT:
2.25
+VCC

>
f-

2.00

enZ

w 1.75
f-

!:
en 1.50
:>

z
:!:

HP 5082-4855

:>
....I

w
VCC = 5.0V 5%
R = STANDARD

>

1.00

.75

i=
<l:
....I

10% VALUE

1.25

a:
.50

.25

0
10

20

2.

DETERMINE NOMINAL
(see Figure 2.4.3.5-2(a)

3.

CALCULATE MAXIMUM
(see DATA SHEET)

POWER DISSIPATION:

4.

CALCULATE MAXIMUM
(see Equation 2.4.3.5-1)

P.vERAGE

CALCULATE NOMINAL
(see Equation 2.4.3.2-1)

CURRENT:

V/\LUE

OF R:

CALCULATE WORST CASE IF:

(see Equation 2.4.3.5-4)

7.

IF(

max

) = 4.75 - .20-1.6L=
(150)(.9)

:. RELATIVE

PAVG = 'AVG

[VF + RS (IAVG - IF)]

R = VCC - VF - VCE SAT

'F

max

LAMP OPERATES VIELL WITHIN

CALCULATION
OF LUMINOUS
(see Equation 2.4.3.5-4)
(see Figure 2.4.3.5-2(a)

IF(

PAVG = 100 mW -1.6 mW (70-50) = 68 mW

.
j.)(1.9)2+(.02)PAVG}1.9
.. IAVG-.01

:.R = 5.0 ;~.6 - d.. = 14111

I :.

40

If':

VF"2.0V@20mA,RS"EI1

5.

30

IF - FORWARD CURRENT - mA

Figure 2.4.3.5-2(a)

R = 15011 10%

) = 5.25 - .201.61
(150)(.9)

ABSOLUTE

INTENSITY

- 25 5 mA
.

MAXIMUM

RATINGS

RATIO IF DESIRED:

218 mA
.

IV @21.8 mA = 1.07

- )=4.75-.40-1.66-163mA
mm
(150)(1.1)

:. RELATIVE

Figure 2.4.3.5-2

IV@ 16.3 mA = .75

VCC = 5.0

VF = 1.6
VCE SAT = .3V
IF=22mA

(see Equation 2.4.3.5-5)

(see Figure 2.4.3.5-2(a))

'F(

=33mA

VF(max)=1.66V
:. 'V RATIO = 1.07 = 1.43:1.00

.75

Example of a Worst Case Design for a Dc Driven

lED lamp.

UJ
-I

co

a:

LJ
.J
OJ

UJ

-I ....
~~

10 KHz
I 300 Hz
"
30 KHz 11 I
1 KHz II11I1
100 KHz 1/
3 KHz
100 Hz

~ Z LJ ..
UJ .J Z
~a:OUJ

:J a: ~- a:
~ :J c; a:
- u ~5:J
x ~ ~_u
~:u
~ UJ >< Cl
u. a..

o
o

/
/

::

~-

.....

a:

/
II

'\

i~

VCC = 6.0V 5%
R = STANDARD

SELECT LAMP:

5% VALUE

HIGH EFFICIENCY

DETERMINE IpEAK:
(see Figure 2.4.2-2)

3.

CALCULATE

MAXIMUM

RED LAMP - 2.0 mcd at 10 mA DC

IpEAK = (1.5 mcd)(10 mA)(1.00)

(2.0 mcd)(.10)(1) @ IpEAK)

:. IpEAK = 50 mA

:. lAVE = 5 mA

POWER DISSIPATION:

(see DATA SHEET)

CALCULATE

MAXIMUM

AVERAGE

CURRENT:

(see Equation 2.4.3.5-1)

.
.. IAVG-

VF<;;3.0@10mA,RS<;;35n

PAVG"

(.
_ )(2.65)2

IAVG [VF + RS (IAVG - IF)]

CALCULATE
NOMINAL
(see Equation 2.4.3.2-1)

VALUE

R = 6.0 5~7 - .3 = 60n SELECT

7.

OF R:

+ (.14) PAVG~ 2.65

.O~'

R=

CALCULATE
WORST CASE IF:
(see Equation 2.4.3.5-4)

-'.
TO 20 mA ON DATA SHEET

VCC-VF-'JCESAT

IF

R = 62n 5%

- 6.3 - 2.5 - .2 = 61 mA
F(max) - (62)(.95)

LAMP OPERATES SLIGHTLY ABOVE ABSOLUTE MAXIMUM

RATINGS.
DESIGNER CAN CHOOSE LARGER VALUE OF
RESISTORiTiGHTER
TOLERANCES ON COMPONENTS.
DESIGNER

CHOOSES 68n 5% RESISTOR

CALCULATION
OF LUMINOUS INTENSITY RATIO IF DHIRED:
(see Equation 2.4.3.5-4)
- 5.7 - 2.5 - .2
464
A
I
(see Figure 2.4.2-2)
F(max) - (68)(.95)
=
. m
:. I

(
) = (46.4)(.1 )(1.48)(2.0)
V max
(10)(1.00)

(see Equation 2.4.3.5-5)

(see Figure 2.4.2-2)

= 1 37 mcd
.

_5J-3~-.4

F(min) -

(68)(1.05)

("j:--~sr~)+-

";.llsI:- '2. +- ("

= 25 mA, L1MI1ED

6.

0 ~r

322
J
. m,

. I ( . ) = (32.2)(,1)(1.38)(2.0)
.. V mm
(10)(1.0)
= .89 mcd

Example of a Worst Case Desi9n for a Strobed

LED Lamp.

F -It.~

r ~- (Vr=

I (~sr')-Pilv~

:rF ")r - PIIJ"

_It!>r~)

below these conditions. For example, suppose the de~ign

requires a standard red lamp to be driven at 200 mA p ~ak
on a 10% duty cycle at 25C. At 100 Hz repetition rate, the
maximum allowable peak current for a 1000 /lS pulse width
is 3.7 x 50 = 185 mA. However at a 300 Hz repetition rate,
the maximum allowable peak current is 4.6 x 50 = 230 mA
on a 10% duty cycle. Thus, by operating the device at ~~OO
mA peak current, 10% duty factor, at a 300 Hz repetition
rate, the maximum junction temperature will not exceed
the junction temperature obtained by operating the device
at 50 mA dc. For additional reliability, the designer can
derate IDC MAX below the maximum temperature dera ted
IAVG(MAX) as calculated by equation 2.4.3.5-1 or as
restricted by the data sheet.
When LED lamps are to be grouped together in an arr.1Y,
the designer should also consider the requirement or
luminous intensity matching of adjacent LEDs. Variations
in power
supply voltages, resistor tolerances, driver
tolerances, and device tolerances all contribute to variatie,ns
in IF and hence luminous intensity variations. The usual
procedure to determine whether variations in IF will cause
noticeable luminous intensity variations is to calculate the
worst case minimum and maximum values of IF' Then the
relative luminous intensity or relative luminous efficiency
curves can be used to determine the worst case variation in
luminous intensity. In general, the maximum luminous
intensity ratio between LED lamps should be less thin
2.0: 1.0, and ratios
greater
than 2.3:1.0 will Je
objectionable to an observer. When LED lamps are driv<m
by a common power supply, variations in the power supply
voltage will cause only a small change in the maximum ratio
of LED forward currents since the forward currents will all
change proportionally
due to power supply variatiors.
However, the component tolerances will have the greatest
effect on IF at the minimum power supply voltage. When
resistive current limiting is used, the. minimum and
maximum IF can be calculated as shown below:

- VCC - VCE SAT(MIN) - VF(MIN)

F(MAX) R (MIN)
(2.4.3.5-fi)

- VCC - VCE SAT(MAX)

F(MIN) R (MAX)

- VF(MAX)

An array of standard red LED lamps is to be dc driven with

a desired typical luminous intensity of 1.5 mcd at 25C
with a maximum ambient temperature of 70C. The arra~f
will be driven by 7416 TTL hex inverters from the 5.0"
TTL supply. The desired circuit and calculations are shown
in Figure 2.4.3.5-2.

An array of high efficiency red LED lamps is to be strobed

on a 10% duty factor with an npn transistor. The desired
luminous intensity is 1.5 mcd at 25C with a maximum
ambient temperature of 70C. The desired circuit and
calculations are shown in Figure 2.4.3.5-3.

When several LED lamps are used in an application, the cost

of the associated LED drive circuitry can often be reduced
by connecting the LEOs in a multiplexed array. For
example, suppose 16 LED lamps are used as status
indicators on a panel. Each LED can be driven by an
individual transistor or logic gate as shown in Section 2.4.3.
This configuration requires 16 LED drivers, 16 current
limiting resistors, and 17 address lines. The LED lamps can
also be connected as a 4x4 multiplexed array. In this
configuration, only 8 LED drivers, 4 current limiting
resistors, and 8 address lines are required. In general, P"q
LED lamps can be driven by P"q transistors or gates and
P"q current limiting resistors on a DC basis. Each LED lamp
can be considered to be an element, aij' where i = 1,2, ... P
and j = 1,2, ... q, of an x-y addressable array of P"q
individual LEDs, such as shown in Figure 2.4.4.1-1. If the
LED lamps are connected as a multiplexed x-y...array with
p"q..;elements, then only p+q transistors or gates will be
required to drive the P"q lamps and p or q current limiting
resistors will be needed. In many applications, such as with
a microprocessor,
the information
is available on a
multiplexed basis and minimal logic is required for proper
decoding.
If dc signals are available, they can be
multiplexed
with inexpensive multiplexers
or a bus
configuration. One final advantage of the multiplexed array
is that the number of wires required to connect the LED
lamps to the drive circuitry;;is reduced. The dc driven array
requires (P"q+1) wires to connect the lamps but the
multiplexed x-y array requires only p+q wires.
Figures 2.4.4.1-2 and 2.4.4.1-3 show the two basic types of
LED arrays. The array shown in Figure 2.4.4.1-2 is dc
driven because only one LED lamp is turned on at anyone
time. By selecting one x axis and one y axis, the single LED
specified by the coordinates ("j' Yi) is turned on. Multiple
LED lamps can be driven on a dc basis if they can all be
addressed by one common x or y axis. Figure 2.4.4.1-3
shows the more general type of x-y addressable array where
any combination of LED lamps can be turned on. One axis
is sequentially selected on a lip or llq duty factor. While
the proper axis is selected, any combination of address lines
of the opposite axis can be enabled, turning on the
corresponding LED lamps. If the display is refreshed at a

is for character generation, such as an alphanumeric display,

where the LEDs are mounted in close proximity to each
other. In applications of this type, besides designing the
drive circuitry to obtain the proper time averaged luminous
intensity as described in Section 2.4.2, each LED lamp used
in the array should be preselected to have less than a
specified maximum luminous intensity ratio. Luminous
intensity ratios of 2.3: 1.00 between adjacent LEDs will be
noticable by an observer. A luminous intensity ratio of
2.0: 1.0 or less is recommended for this application.

Y1

Y2
CJ

z
CD

Y3

a:
Ien

:!:
a:
0

Y4

/
/ap3
X2
(
Figure 2.4.4.1-1

/ap4
X3

/apq
X4

Xq

COLUM:;TROBI~

LED Lamps Can Be Connected as an X-V Addressable

Array of PO Elements with P Rows and 0 Columns.

The first step in the design of an x-y addressable array is the

selection of the LED lamp package that is to be used. The
designer can choose between hermetic lamps, T-l 3/4
lamps, T-l 3/4 low dome lamps, T-l lamps, subminiature
lamps, or rectangular lamps. The designer can mix package
styles or LED colors in the same array, although LEDs in a
single row for a column strobed application or single
column for a row strobed application should have similar
electrical and luminous intensity characteristics. As a guide,
standard red GaAsP substrate LED lamps; high-efficiency
red or yellow GaP substrate LED lamps; and green GaP
substrate LED lamps have substantial electrical or luminous
intensity differences. To compensate for these differences,
these three lamp categories should be driven in separate
rows of a column strobed circuit (or separate columns of a
row strobed circuit) so that a different current limiting
resistor can be used for each lamp category. For character
generation applications, such as an alphanumeric display,
the lamp package should be chosen depending on the
desired array size. Figure 2.4.4.2-1 shows the minimum
LED lamp spacing for T-13/4, T-l, subminiature, and
rectangular lamps.

X2

X3

q
Figure 2.4.4.1-2

One LED Will Be Turned On By Applying the

Proper Signal to One X Axis and One V Axis.

fast enough refresh rate (> 100Hz) then all LED lamps will
appear to be dc driven. The present nomenclature used for
x-y addressable arrays is that the y axis addresses (Yi;
i = 1, 2
p) are called rows and the x axis addresses (Xj;
j = 1,2
q) are called columns. If the x axis addresses are
sequentially selected, then the array is column strobed and
if the y axis addresses are sequentially selected, then the
array is row strobed.

Now the desired luminous intensity of the array should be

specified. The desired luminous intensity sets a limit on the
minimum duty factor that can be used to strobe the x-y
array because of the peak current and pulse width
limitations of the LED lamps. The minimum duty factor by
which the x-y array can be driven is given by Equation
2.4.2-1, reproduced below:

DUTY

FACTOR

lIpEAKl

where:

all variables
IpEAK

Two basic
applications
exist for LED arrays. One
application connects the LED lamps in an x-y addressable
array as a means of simplifying the LED drive circuitry.
However, each LED is individually mounted on a large
surface. The second application of an x-y addressable array

~ _lI_V_T_'_M_E_A_V_G_l_[_'S_P_E_C_l_[T/_<_IS_P_E_C_I]

are defined

is the maximum

[T/(IPEAKI1

in Section
tolerable

lIV

SPEC]

2.4.2

and

peak current.

Table 2.4.4.2-1 shows four representative lamps and gives

the maximum peak current and duty factor limitations to
obtain a desired luminous intensity. The duty factors given

/'
/'

!
4

X,

Figure 2.4.4.1-3

In a Muitilllexed Array Any Combination of LEOs

Can Be Turned On by Sequentially Strobing One
Axis.

X2

IpEAr

X3

DUTY
FACTOR
DC
1/2
1/4
1/8
1/16
1/32

8bit Latch
8x 1 =
7x 2 =
6x 4=
5x 8 =
4x16 =
3x32 =

Lamp Type

8
14
24
40
64
96

LEOs
LEOs
LEOs
LEOs
LEOs
LEOs

16-bit Latch

24-bit Latch

16x 1 = 16 LEOs
15x 2= 30 LEOs
14x 4= 56 LEOs
13x 8 = 104 LEOs
12x16 = 192 LEOs
11x32 = 352 LEOs

24x 1= 24 LEOs
23x 2= 46 LEOs
22x 4= 88 LEOs
21x 8 = 168 LEOs
20x16 = 320 LEOs
19x32 = 608 LEOs

Typical Light Output

Maximum
Peak
Current

32-bit Latch
32x 1 =
31x 2=
30x 4 =
29x 8 =
28x16 =
27x32 =

32
62
120
232
448
864

LEOs
LEOs
LEOs
LEOs
LEOs
LEOs

Minimum Duty Factor to Obtain

Desired Time Averaged Iv
.5 mcd 1.0 mcd
1.5 mcd
2.0 mcd

1A

1/64

1/32

----- ..

----- ...

2.0 mcd - 10 mA dc
17(60mA)/17( 10mA)=1.54

60mA

1/40

1/18

1/12

1/9

GaP Yellow (583nm)

1.8 mcd - 10 mA de
17(60mA)/17( 10mA)=1.54

60mA

1/32

1/16

1/11

1/8

GaP Green (565nm)

1.8 mcd - 20 mA dc
17(60mA)/17(20mA)=1.44

60mA

1/16

1/8

1/5

1/4

GaAsP Red (655nm)

.8 mcd - 20 mA dc
17(1A)/17(20mA)=.8

GaP HER (635nm)

in the table can be decreased only by specifying brighter

lamps or by exceeding the maximum tolerable peak current
ratings of the LED lamps. When several different types of
LED lamps are used in a single array, the lamp with the
largest minimum duty factor determines the minimum duty
factor for the entire array. The minimum duty factor as
specified above sets an upper limit on one dimension of the
x-y array.
For example, suppose that an array is
constructed of the standard red, high-efficiency red, yellow
and green lamps specified in Table 2.4.4.2-1 and the desired
time averaged luminous intensity of the display is 1.0 mcd
for standard red lamp and 1.5 mcd for the other three
lamps. The minimum duty factor of the array is 1/5, but a
1/4 duty factor is selected to simplify logic decoding. The
other dimension of the x-y addressable array is determined
by the total number of LED lamps in the array and the
number of each type of LED lamp with different electrical
or luminous intensity characteristics. Suppose that in the
previous example, the array consists of 3 red lamps, 2
high-efficiency red lamps, 6 yellow lamps, and 1 green lamp.
The x-y addressable array would be dimensioned as 4x4
even though only 12 lamps are used. Since the duty factor
and desired time averaged luminous intensity for each lamp
is known, the peak current can be determined for each type
of LED lamp. Figure 2.4.4.2-2 shows the completely
specified array described in the text.

RED:

IpEAK = 97 mA

HIGH EFFICIENCY
YELLOW:
GREEN:

(
Figure 2.4.4.2-2

RED:

IpEAK = 24 mA

IpEAK = 26 mA
IpEAK

= 50

COI:MN

mA

STROBE)

Example of X-Y Addressable Array Using 3 Red L1:0s,

2 High Efficiency
Red LEOs. 6 Yellow LEOs and
1 Green LED.

The final step in designing the LED x-y addressable array i:;
the design and specification of the associated LED driv{:
circuitry. The LED anode drivers must be able to sourc{
current to the LED array and the LED cathode drivers must
be able to sink current from the array. The anode and
cathode drivers can be implemented with NPN, PNP, FET
transistors or commercially available LED drivers. Figure
2.4.4.2-3 shows some of the commonly used LED drive
schemes implemented with NPN, PNP or FET transistors.
The transistors should be selected so that the average power
supply dissipation within the device or multiple transistOJ
array is less than the manufacturers specifications. Rl and
R2 are selected so that the transistor drivers remain in
saturation under worst case conditions of YCC' YOV YOH'
transistor HFE, and resistor tolerances at the maximum
peak currents specified for the LED array. Finally, the LED
current limiting resistors are calculated as shown in Section
2.4.3. Figures 2.4.4.24 and 2.4.4.2-5 show examples of
row and column strobed x-y addressable arrays using PNP
anode drivers and NPN cathode drivers. The calculations of
the maximum average and peak currents of each transistor
assume that all LED lamps in the array are on. Active
pullup drivers can also be used to drive the array provided
that the BYR restrictions of the LED lamps are not
exceeded. If an LED lamp is connected to a low impedance
driver with a potential greater than BYR' current can flow

through the lamp in the reverse direction. This reverse

leakage current will not harm the LED unless the maximum
power dissipation of the lamp is exceeded, however, this
reverse leakage current can flow through adjacent LED
lamps in the forward direction and thus cause unwanted
LED ghosting.

A microprocessor can be used to control an x-y addressable

array of LED lamps. The external circuitry that is required
for the array is minimal and only a small portion of
microprocessor time is used to refresh the LED array. The
technique that is used is to periodically strobe data into a
latch from the microprocessor and use the outputs of the
latch to address the array by row or by column. A specified
time later, new information for another row or column is
strobed into the latch. If the x-y addressable array is row or
coumn strobed at a refresh rate greater than 100 Hz, then
the entire array will appear to be dc driven. Refresh timing
for the array is handled by generating an interrupt after a
specific time interval. The latch is partitioned so that n bits
specify one of 2n rows (row strobe) or 2n columns (column
strobe) and the remaining bits specify the coordinates of
the LED lamps of the opposite axis which should be turned
on. The total number of LED lamps that can be addressed
is determined by the minimum duty factor of the array and
the size of the latch. Table 2.4.4.3-1 shows the maximum
number of LED lamps that can be driven with latches
ranging from 8 to 32 bits and for duty factors ranging from
1/32 to dc. Figure 2.4.4.3-1 shows an example of a 5x8
LED array that is designed to interface to an Intel 8080A
microprocessor. A low to high transition of output Q of the
monostable multivibrator requests an interrupt from the
microprocessor. The interrupt circuitry (not shown) forces
a RST7 instruction into the microprocessor. Following the
RSTI instruction, the microprocessor executes the program
shown in Figure 2.4.4.3-2. The microprocessor
then
updates output port (n) which is the 8 bit latch used by the
x-y addressable array. If the microprocessor uses a 2 MHz
clock, then the percentage of total time required to refresh
the LED array at a 100 Hz repetition rate is as follows:

REFRESH TIME =

(143C -1) R
MICROPROCESSOR CLOCK RATE
= 5.7% FOR C = 8, R = 100 Hz

where:

C is the number of columns in the display and R is

the refresh rate.

For the remaining 94% of the time, the microprocessor can

be used to update the contents of the LED array in RAM
and perform countless other tasks required by the system.

To address larger LED arrays, several 8 bit latches can be

used. Each 8 bit latch is addressed by a separate output
address. For N eight bit latches, each latch would be
addressed as output (n+i), where i = 0,1,2, ... N-l. The
program is modified by inserting INX HL, MOV A, M and
OUT (n+i) instructions into the program as shown in Figure
2.4.4.3-2 and changing the CPI (17)16 instruction to CPI
(OF+8N)16' The monostable should be triggered by the
high to low transition of the strobe input of the final 8 bit
latch. To prevent a small amount of ghosting while data is
being
strobed into the latches, the output of the
monostable can be used to disable the 1 of N decoder and
thus turn all columns off from the time that the interrupt is
requested until all the data has been strobed into the
latches. The extra time required to address several 8 bit

latches as compared to a single 8 bit latch is negligible. For

example, suppose four 8 bit latches are used to address a
29x8 LED array. If the microprocessor uses a 2 MHz clock,
the total percentage of time required to refresh the LED
array at a 100 Hz repetition rate will increase to:

REFRESH TIME =

=
where:

[C (121 + 22N) -1] R

MICROPROCESSOR CLOCK RATE
8.4% FOR C = 8, N = 4, R = 100 Hz

C is the number of columns in the display, N is

the number of eight bit latches, and R is the
refresh rate.

ANODE DRIVERS
PNP OR PNP DARLINGTON
+VCC

NPN OR NPN DARLINGTON

OPEN DRAIN P CHANNEL

+VCC

+VCC

-4d~SD

ACTIVE
LOW INPUT

ACTIVE LOW INPUT

VCC

VOHlJ
VOL

VO~'

CATHODE DRIVERS
NPN OR NPN DARLINGTON

(NOTE 21
ACTIVE
LOW INPUT
_ VR +
-R2
(NOTE 3)
VCC--,
rVOLLJ

ACTIVE
HIGH INPUT

+ VCE SAT WITH RL

(VOL +VR+VBE)
IF RL OMITTED

VOHn
VOL-.J

L
-----.J

ACTIVE
HIGH INPUT

(NOTE 2)

NOTES:

1.

RL OMITTED ON COLUMN DRIVERS WITH COLUMN STROBED CIRCUIT OR

ON ROW DRIVERS WITH ROW :;TROBED CIRCUIT.

2.

R1 CAN BE OMITTED IF TRANmSTOR

3.

R2 CAN BE OMITTED IF RL 0111 ITTED AND TRANSISTOR REMAINS ACTIVE.

LEAKAGE IS SMALL.

l~DS

Y1

I ROW:;-

R1

Y2

--

R3

;;;-

R4

IROW3

Y4

-Y

RO

'ROWpR1

R2

R3

R2

'ROW2

;;3

Rp

Rq

R4

ICOLq

ICOLq

IpEAK

lAVE IROW i)" ~ IROW i' WHERE q IS THE NUMBER

lAVE ICOL j)" ~ ICOL j' WHERE P IS THE NUMBER OF ROWS

IAVEIROW

1
q
i)" N- "~ICOL
)=1

I
PEAK IROW i)"

Figure 2.4.4.2-4

q
~
j=1

lAVE ICOL j)"

j' WHERE q IS THE NUMBER OF COLUMNS

Row Strobed XV Addressable Array. V 1 throl gh

V are Sequentially Strobed on a 1/N Duty Cycle,
N~P.

P
i~1 IROW i' WHERE P IS THE NUMBER OF ROWS

N
p

IpEAK
ICOL j

OF COLUMNS

ICOL j)"

~ I ROW i
i=l

Figure 2.4.4.2-5 Column Strobed X-V Addressable Array. X

1
through X are Sequentially Strobed on a
1/N Duty qCycle, N ;;;'P.

2.4.4.4 Analog Bar Graph Arrays

In many
applications,
analog information must be
converted into a visual display. Traditionally analog panel
meters have been used for low cost applications requiring
only moderate accuracy and resolution. LED linear displays
can often be substituted for these applications. LED lin~ar
displays consist of a linear or circular array of LED lamps
that are driven by a device that decodes the varying anal og
or digital signal into a bar graph or position indica":or
display. The input signal can be decoded so that all LEDs
with thresholds below the input are turned on (bar grarh)
or that only the LED with its threshold closest to the input
is turned on (position indicator). Some examples of thEse
type of displays are shown in Figure 2.4.4.4-1. LED linear
arrays have many advantages over analog panel mete rs.
These advantages include higher reliability, higher resistan:::e
to mechanical shock and higher visibility in low and
moderate ambient lighting. Since the LED linear array is a
light emitting device, it is more effective at getting the
viewer's attention than a panel meter. Red, yellow ar.d
green LEDs can be used to quickly identify the propl~r
limits
of instrument
operation.
Distinct switching
thresholds can be selected to allow the linear array 10

simplifY the machine operator's decision. Finally, in

consumer
oriented equipment,
the LED linear array
provides a new and distinctive selling feature for the
product. Depending on the number of LEDs in the array, if
the input signal is in analog form, the linear array decoder
can be one or more operational amplifiers, a monolithic
analog decoder, or a low cost analog to digital converter
followed by a simple digital decoder. If binary or BCD
information is available, such as from an analog to digital
converter or a microprocessor, then only a simple digital
decoder is required. Examples of several of these circuits
will be shown.
When analog information is available, operational amplifiers
or voltage comparators can be used to interface to the LED
linear array. Since the LED lamps operate at low currents
and low voltages, most operational amplifiers can drive
LEDs without output buffering. Figure 2.4.4.4-2 shows
examples of a bar graph display and a position indicator
display. In both circuits, a five resistor voltage divider
determines the switching thresholds of the LED lamps. The
resistor values are not critical but the resistance ratio

RCA
CD4508

TO
DATA

BUS

PROPER ADDRESS CODE (n)

OUTPUT INSTRUCTION
Qngn

+VDD

/\'-,...u\

__-~~-~p:D:~~

'~QnlVl

~TA"lle

...-.....

I
NATIONAL

058863

r-----------...,

I
I

REQUEST
FORCES
RST(7)
INSTRUCTION

8 NPN DARLINGTON
COMMON EMITTER

I
I

1
2
A

6
7
8
9

I
I
L

I
.JI

ADDRESS

OPCODE

(003S116

PUSH PSW
PUSH HL
LHLD
AL

~'dv

CLOCK
CYCLES
11
11
16

A. M
OUT
(nl
INX HL
MOVA. M
OUT
(n+1)

7
10

INX HL
MOVA. M
OUT
(n+N- I)
MDVA. L
CPI
(17)16
JNZ
LOOPL
LOOPH
MVI A
(10)16
STA
AL

5
7
10

O(RED)

0000000000000

LESS THAN

HL = HL + 1
A = (HLI
STORES SECOND BYTE OF INFORMATION

7
10

o
A.

NULL METER THAT COMPARES

INPUT VARIABLE TO REFERENCE

POP PS W
EI
RET
INX HL
SHLD
AL

:Sp HL
POP PSW
EI
RET

6C1'F

10

A=L
COMPARES A TO (17116
(OF+SN)16 FOR N EIGHT BIT LATCHES
A' (17)16 JUMP TO LOOP

13

POINTERL

10
10
4
10
5
16

ENABLE INTERRUPT FLAG

RETURN
HL= HL+ 1
POINTER = POINTER + 1

5
7

A= (10116
= (10)16

10
10
4
10

IN GENERAL,
FOR C COLUMNS, N EIGHT BIT LATCHES, THE TOTAL
NUMBER OF CLOCK CYCLES REOUIRED TO REFRESH
THE ENTIRE DISPLAY ONCE EOUALS:
CLOCK CYCLES = C (121 + 22N) - 1

CONTENTS (01'00)
AH AL
AH AL

+ 1

Tx-x-10l16"
(XX 11)16
(X X 12)16
(XX 13)16
(XX 14)16
(XX 15116
(XX16)16
(XX17)16

Figure 2.4.4.3-2

POINTERL
POINTERH
-OO-OTY Y
001:
010:
011:
100'
101'

S 4Y3Y2V,

110:
1 1 1:
I

5 BIT
ROW
INFORMATI ON

POINTER = NEXT COLUMN

TO BE DISPLAYED
COLUMN
COLUMN
COLUMN
COLUMN
COLUMN
COLUMN
COLUMN
COLUMN

1
2
3
4
5
6
7
8

Intel 8080A Microprocessor Program used to Strobe

the XY Addressable Array Shown in Figure
2.4.4.3-1.

between resistors will specify the threshold, V I to V4' a~

shown in Figure 2.4.4.4-2. As shown, the linear displays can
detect an analog signal greater than zero. In circuit
2.4.4.4-2a (bar graph), if VIN < VI' where VIis
the
switching threshold of the first lamp in the array, then all

7C1'F

Figure 2.4.4.4-1

.1 2
C.

SC1'F

EMPTY

Typical Applications
Bar Graph Arrays.

1~

DIGITAL METER THAT OFFERS

5% RESOLUTION

00000.00000

:Sp HL

POINTER-'

(GREEN)

EQUAL TO

ADDED
FOR N
EIGHT BIT
LATCHES

LOOP

(RED)

GREATER THAN

14

000
~

14

FULL

for Position Indicator and

lamps in the array will be off. As VIN increases above each

switching threshold, VIto V4' the corresponding output of
each operational amplifier conducts current to ground
which turns the LED on. R is selected to limit the current
through the LED to the desired value. Since each LED lamp
has its own current limiting resistor, standard red, high
efficiency red, yellow, and green lamps can be mixed in the
same array. Voltage comparators with open collector
outputs can also be used in the circuit. In circuit 2.4.4.4-2b
(position indicator) only one LED lamp in the array will be
on. If VIN < VI' the outputs of all operational amplifiers
will be pulled up to V cc. U 1 will source current to LED1,
turning it on but LED2 through LEDS will be turned off
with approximately zero bias. As VIN increases above VI'
the output of UI will conduct current to ground, turning
LEDI off but turning LED2 on. Each LED lamp will turn
on sequentially as VIN increases above the specified
thresholds. Either circuit can be expanded in size to
accomodate any desired bar graph display. The circuits
shown in Figure 2.4.4.4-2 can be operated from a single
polarity power supply providing that only positive signals
are to be detected. Both positive and negative signals can be
detected if a dual polarity power supply is used. A dual
operational amplifier with three LEDs can be connected as
shown in circuit 2.4.4.4-2b to make a solid state null
detector circuit, that compares VIN to a specified VREF
determined by the resistor divider network. The LEOs
would
indicate
V INV REFt.),
(VREFt.)<VIN<
(V REF+t.), and VIN>(V REF+t,), or negative, zero or
positive if VREF is equal to zero and t. is small. Such a
device could replace a conventional analog null meter.
F or longer LED linear arrays, IC decoders are available that
decode the analog input signal into a bar graph or a position
indicator display. Figure 2.4.4.43 shows an example of an
LED bar graph display that uses the Siemens UAA-180
decoder and a position indicator display that uses the
Siemens UAA-170 decoder. For both devices, a voltage
divider network consisting of R3, R4, and RS determine

MIN

eeeo

MAX

MIN

MAX

ooeoo
NATIONAL
LM324

NATIONAL
LM324

VIN
V,

< V,=

LED, "ON"

< VIN < V2:

LE02 "ON"

V2 < V1N < V3: LE03 "ON"

V3 < V1N < V4: LE04 "ON"
VIN> V4: LEOS "ON"

Figure 2.4.4.4-2

Operational Amplif ers or Voltage Comparators

Used to Decode an ~nalog Signal into a Bar Graph
or Position Indicator Display.

the switching range of the linear array. Using the UAA180

decoder, if VIN is less than VMIN, all LED lamps will be
off. If V
is greater than VMIN, the input voltage will be
IN
decoded as a bar graph array with all LEDs on when V [N is
greater than VMAX For the UAA170 decoder when VIN
is less than V
, then LED1 will be on. As VIN increases
MIN
above V
then LED2 through LED16 will turr on
MIN
sequentially. When VIN is greater than VMAX' LEDl6 will
be on. Since the UAA-170 and UAA180 have constant
current drivers for the LEDs, no current limiting resi~tors
are Jequired. However, the constant current drivers are
programmed externally by RI and R2- If longer li:1ear
arrays are desired, additional UAA-170s or UAA-180s can
be cascaded in series_

Digital information can also be decoded as an LED bar

graph or a position indicator display. If an analog signal is
available, it may be desirable and cost effective to use a low
cost analog to digital converter and then decode the digital
outputs as a linear array. The position indicator display can
be decoded with a one of n decoder as shown in Figure
2.4.4.4-4. One of n decoders can be cascaded to form any
size of LED position indicator display desired. Since only
one LED in the position indicator display will be on at a
time, an x-y addressable array can be used to simplify
decoding. Figure 2.4.4.4-5 shows how two one of eight
decoders and 17 external components can be used to
address 64 LED lamps in a position indicator display. Using
commercially available 1 of 4, 1 of 8, 1 of 10, or 1 of 16

10
18
SIEMENS
UAA180

LEDn

4
S
6

SIEMENS
UAA170

LED-11
LED,O
LEDU

7
LEDS

R4

LED7

LED6

VM1N
16
R3
VIN

LEDS

17
R2
2
R1

Figure 2.4.4.4-3 Use of Siemens UAA 170 and UAA 180 Position
Indicator and Ear Graph Decoders.

74156

7
6"
c1 5"
Cz 4
G1
3"
Gz 2"
1
0

A
B

74156
A

C1

z 3"

G1

6"
7

E=0

1 OF 64 LED LAMPS WI LL LIGHT

E= 1

ALL LED LAMPS OFF

Figure 2.4.4.4-5
Figure 2.4.4.4-4

One of Eight Decoder

Decoder/Driver.

Used as Position Indicator

decoders, almost any size LED position indicator array can

be decoded. Using the technique shown in Figure 2.4.4.4-4,
four 1 of 16 decoders, four inverters (4/6 package) and 64
external components would be used to address the same 64
LED lamps.
Digital logic can also be used to decode an LED bar graph
display. Figure 2.4.4.4-6 shows how a one of eight decoder
and an And gate network can be used to decode an eight
LED bar graph display. The display can be expanded to any
length with the proper decoder and an extended And gate
network. For large displays, a strobed x-y addressable
approach is simpler. Figure 2.4.4.4-7 shows an example of
such an approach. The lower three bits of the input are
connected to an eight output bar graph decoder. The upper
three bits of the input are continuously compared to the
outputs of a divide by eight counter. The output of the
counter is decoded by a one of eight decoder which row
strobes the 64 LED matrix on a 1/8 duty factor. If the
value of the counter is less than the upper three bits of the
input, then all LEDs in that row are turned on. When the
value of the counter is equal to the upper three bits of the
input, the lower three bits of the input are decoded by the
bar graph decoder. For values of the counter that are
greater than the input, all LED lamps for that row are
turned off. This technique can also be used to decode BCD
inputs by substituting a decade counter, one of ten

Two One of Eight Decoders Can Be Used to Decode

a 64 LED Position Indicator Display.

decoder, and ten line bar graph decoder for the devices
specified. The maximum size of such an array is limited
only by the requirement of a certain minimum duty factor
to obtain the desired time averaged luminous intensity.
Microprocessors can be interfaced directly to any of the
digital decoder circuits previously shown by adding a latch
to hold the input information. The microprocessor would
need to update the latch only when the information is to be
changed. The techniques described in section 2.4.4.3 can
also be used to implement a linear array. Unless the
application is microprocessor time limited, these techniques
are more cost effective than the bar graph decoder circuit
described in Figure 2.4.4.4-7. The microprocessor would be
required to decode the input information in software and
continuously refresh the linear array. The circuit described
in Figure 2.4.4.3-1 can be used to implement either a bar
graph or a position indicator display.

Information can be more readily assimilated from a panel

by an observer if only those symbols or characters relevant
to a particular condition are visible. Ideally, the panel has a
"dead front" appearance for all symbols except those
required to be seen, and those are made visible with light

projected through them from a source in back of the pa el;

hence the term "backlighting", illustrated in Figure 2.4.5-1.
Red LEDs
light
to
technology
yellow and

of earlier technology did not produce enough

make backlighting
practical. More recmt
not only makes red backlighting practical, but
green as well.

Front lighting differs from backlighting in that it is med

mainly as a substitute for or a supplement to ambient
illumination. Because of scattered and stray light, it is

difficult to selectively display symbols by a front lighting

technique. LEDs can also be used in front lighting since the
supplementary light is usually needed only when ambient
light is subdued. Front lighting is, therefore, a simple design
and does not require as much design attention
as
backlighting.

There are only four parameters to consider. In the order of

usual importance, they are:

7404

74138
2

21

22

,-

,-

7409

2"
3

2"
E

G2

3"

G2

3"

4"
5

4"

5
6

6"
7

6"

7"
S

7"
ACTIVE HIGH OUTPUTS
(TOTEM POLE)

ACTIVE LOW OUTPUTS

(OPEN COLLECTOR)

8"

LEOS

SIGNETICS
8243
I

0
0
0
0
0
0
0
0

0
0
0
0
1
1
1

0
0
0
0
0
0
0
0

NOTE,

1 234

0
1
0
1
0
1
0

0
0
1
1
0
0
1

0 0 000
0
1 000
000
11000000
1 1 1 0 000
1 1 1 1 0 000
111
1 1 000
1 1 , , , '00
, , , , , , ,

(i

, , ,

,
NOTE2

X
X

X
X

5 6 7 8

X
X

0
0
0

ACTIVE
ACTIVE
ACTIVE
ACTIVE

""""

Figure 2.4.4.4-6

CONFIGURATION
CONnGURATION
CONFIGURATION
CONnGURATION

C<0 0 0 0 0 0 0

HIGH
LOW
HIGH
LOW

S,
S2

7
6
5
4
3

X = DON'T CARE
FOR
FOR
FOR
FOR

So

1=0
I=X
E= X
E= 0

One of Eight Decoder Used as Bar Graph

Decoder/Driver.

0
ACTIVE HIGH OUTPUTS
(OPEN COLLECTOR)

24
25

20

22
7485

BO
B1 A<B
B2
B3 A>B
AO
A1
A2
A3
A<8

74156
A

0-

,-

C1

C2
G1
G2

4
5

6
7

Figure 2.4.4.4-7

Strobed Bar Graph Decoder that Uses a Multiplexed

X-V Addressable Array to Simplify Decoding.

Contrast
Uniformity
ON Sterance*
Radiation Pattern
*Because this section deals only in photometric
(no
radiometric) quantities, the prefix luminous and subscript v
may be dropped from the terms and symbols:
1.
2.
3.
4.
5.

Luminous
Luminous
Luminous
Luminous
Luminous

flux, tPv
incidance, Ev
exitance, Mv
intensity, Iv
sterance, Iv

The radiation pattern of the backlighted symbol is ideally

lambertian;
but if the symbol complexity makes it
unrecognizable anyway at large off-axis angles, the angular
variation of intensity (or sterance) is of little importance.

sterance variation (2: I) is usually acceptable, and this is the

basis of the developments described later.
Contrast, the most important consideration,
terms of three contrast ratios:

is evaluated in

ON/OFFCR
ON/BACKGROUND CR
OFF /BACKGROUND CR
ON/OFF CR is the ratio of ON sterance to OFF sterance of
the symbol. The OFF sterance is the product of surface
reflectance and ambient light, so ON/OFF CR varies
inversely
as the ambient
light level.
Similarly,
ON/BACKGROUND CR is the ratio of the ON sterance of
the symbol to the BACKGROUND sterance, which also is
the product of surface reflectance and ambient light.
Neither ON/OFF CR nor ON/BACKGROUND CR should
always be "as high as possible". If ON/BACKGROUND CR
is too high, eye fatigure may result. This is because the ON
sterance results from radiated, not reflected, flux. Both
ON/OFF CR and ON/BACKGROUND CR should, of
course, be adequate and ideally they are equal.
If ON/OFF CR = ON/BACKGROUND CR, it follows that
OFF/BACKGROUND CR is 1.00. When tradeoffs are being
made, because of high ambient conditions and marginal
flux availability,
the usual
optimization
favors
ON/BACKGROUND CR, but OFF/BACKGROUND CR
must not be ignored.
Ade:(luate
ON/BACKGROUND
CR requirements
are
subjective and BACKGROUND reflectances vary widely,
but an approximation to the ON sterance requirement can
be drawn from Figure 2.4.5-2. The curve is drawn from:

Figure 2.4.5-1

Backlighting Used to Selectively Cause Special

Symbols to Appear.

ON sterance describes the intensity per unit area of th,~

symbol when backlighting is applied. For a particular
background and symbol reflectance, the ON steranCI~
determines the ambient light level where the backlighted
symbol can be adequately distinguished. "Adequate"
distinction of the symbol is not precisely definable becaus(~
it bears a subjective relationship to the detail which must be
resolved in the symbol.
Uniformity of sterance over the illuminated area is alsco
somewhat subjective. In some applications, it has cosmetic
value only, but in others, non-uniformity can cause loss of
detail which may cause misrecognition of the symbol.
Uniformity does, however, have an objective criteria fOJ
evaluation. Because eye response is logarithmic, a 3dE

Assume lambertian
MBG = 1T LBG

BG

(cd) =
m2

[1 (cd)]

[1 ( cd)]

1T

1T
=

1m
1m

3.18~

m2

radiation

[E

of flux reflected,

(Ix)]

R
(%)
[GB
]
100%

so

Require ON/BACKGROUND CR, (LON/LBG) to

be within the limits 2.5 < (LON/LBG) < 10; choose
(LON/LBG) = 5 and calculate:

0.1

EvA -AMBI
1.0

ENT LUMINOUS INCIDANCE -FOOTCANDLES

10
100
1000

10k

10 k
N

0:

ul
~.
ul
:, 1000

LON = LBG (LON/LBG)

'.

~i

= 3.18 cd (5) = 15.92 cd/m2

(J)

ul

I-

~!

0::
w

~
~
l>

m2

For the tolerable limits required in step 5, the curve in

Figure 2.4.5.2 shows that an ambient ranging from 250 Ix
to 1000 Ix can be accomodated. Less narrow requirements
would, of course, broaden the range.

lD

::;;

100

<t

~i

~;

10

<'

~
I

~.

::
:,

w
u
1.0 z

:::!
ul

<t

~~

0::
w

Ci ~
ul
~.

I(J)

U,

There are two requirements for the ON sterance. It must be

adequately high and uniform. To obtain uniform sterance
requires a diffuser, as shown in Figure 2.4.5-3. The
diffuser/LED
combination,
properly designed, gives a
background of uniform sterance over which the "legend
plate", bearing the symbol opening, is placed. Without
proper diffusion, sterance would not be uniform over all
portions of the symbol. A viewer would see the LED
through only those portions of the symbol opening that lie
between the viewer's eye and the LED.

(J)

u,

0.1

:,

~;
:i:

~
::>

::

-'I

>

0.01-'

10
(DARK)
(DIM)
EvA -AMBIENT

(cd/m2) = D.O. = 0.70 (cd/m 2 )

Ey IN (Ix)
n
Ix

1000
10k
lOOk
(NORMAL)
(BRIGHT)
(SUNLIGHT)
LUMINOUS INCIDANCE- LUX
(ILLUMINANCE)

Ey(x)
-=

COS (e) =

Ey(O)

Ey(x)
Ey (0) =

(I

--x-2

(Iy(e))

i;(Ol

)2

I+(;r)
3

cos

e
DIFFUSER

(2.4.5.3)

For example, one type of diffusing mm is described as

having a transmission of 55% and a "gain" of 400%. For
this fIlm, the D.Q. is 0.55 x 4.00 = 2.2, and from equation
(2.4.5-3) we can calculate:
LOUT
Y

100

Approximate "ON" Luminous Sterance Required

Under Given Ambient Light.

A diffuser functions as diagrammed in Figure 2.4.5-3.

Normal (perpendicular) incidance, Ev' on the back surface
causes flux to be emitted at the front side, being scattered
by the diffusant such that each increment of area on the
front radiates in all directions. In an ideal (but not
realizable) diffuser, each lumen impinging at the back
would emerge from the front; that is, nothing would be
reflected and nothing lost in the diffuser. Upon emerging,
this lumen would be scattered; ideally, the scattering would
be lambertian, so the ratio of total flux emitted to normal
(perpendicular) intensity is n. Thus, in an ideal diffuser, an
incidance of one lumen per square meter would cause
emission of one lumen per square meter and normal
intensity would be l/n cande1as per square meter. A
"Diffusance Quotient" (similar to Intelligence Quotient)
can be defined as the ratio:

having a normal value of 1.00. D.Q. can be lowered by

absorption and reflection loss, and can be raised by
anisotropic (e.g., light pipe) effects in the diffuser.

Qi ffusonce guotient
D.O.

=" S:. (Cd/m2)

Ey

Figure 2.4.5-3

(2.4.5-4)

5
z

Figure 2.4.5-2

D.O. = n LVOUT/EYIN

t=
o

Ix

Diffuser Effect on Flux Direction as Related to

Radiation Pattern.

Industry standards for characterization of diffusers havt: not

yet appeared, although diffusers are available such as "light
Diffusing Film" from 3M Visual Products Division, and
"Chromafuse"
from Panelgraphic. Tradeoffs in diffuser
selection relate to D.Q. -- a high D.Q. (>1) means the
radiation pattern is narrower than lambertian because
transmission loss cannot be zero. It is unlikely that the
radiation pattern would be broader than lambertian, ;0 a
low D.Q. 1)
means a lossy diffuser, but a narrow
radiation pattern can compensate. D.Q. is therefore the one
single number that comes closest to relating diff lser
specifications to backlighting performance.

Ev (x)

--

cas

Ev (0)

4
8 FOR LAMBERTIAN RADIATION
PATTERN

A curve of equation (2.4.5-10) is given in Figure 2.4.5-4

along with curves for several non-lambertian LEDs. The
curves for the non-1ambertian LEDs were obtained by
applying equation (2.4.5-9) to the radiation patterns in
their data sheets.
Here is how to use these curves and equations for a single
LED backlighting a legend, assuming no reflectors:

Adequately high ON sterance may require a tradeoff with

uniformity or the addition of extra LEDs. This can be ~,een
by analyzing, in Figure 2.4.5-3, the incidance on the
diffuser at points displaced a distance, x, from the axis. The
analysis uses the inverse square law*:
Ev

ON Sterance, Ly

I v /d2

20 cd/m2

*accuracy of the inverse square law is within 5%

for distance, d, as short as two LED diameters.

Refer to Figure 2.4.5-3 and determine EvC0) from

equation (2.4.5-3):
E (0) =
v

1T

x 20 cd/m

= 28.56 Ix

2.2

but the NORMAL incidance component of the incidance

VECTOR is the vector magnitude multiplied by cos (), so:
x = *Y'<lOmm)2

The non-uniformity of incidance can be seen as the ratio

Ey(x)/EvC0), found by taking the ratio of equation
(2.4.5-8) to (2.4.5-6):

+ (20mm)2)

= 11.l8mm

Choose an LED and either construct a curve

according to equation (2.4.5-9) or use a curve from
Figure 2.4.5-4. Assume now a type 5082-4650/55.
Find 8 at which Ev/EvC0) = 0.5:

[Iv (())/x2 + d2)] cas ()

Iv (0)/d2
=

Iv (())

) cas 8

x2+d2

Iv (0)
__Iv (8)

d2

= (

cas

Iv (0)

In equation (2.4.5-9), the ratio [!Y(())/I/O)] is recogni:~ed

as the relative intensity vs. 8, which is the radiation pattern
as given in most LED data sheets. Thus, if the LED
radiation pattern is lambertian, cos 8 can be substituted for
[ly(8)/IvC0)] to give:

Compute LED intensity required, using equation

(2.4.5-6) and results of step 1, step 4:
IlO)

= 28.56lx

(23.98mm)2

16.42 mcd

This is about five times the 3.0 mcd minimum specified for
the 5082-4655, so a different LED is tried. Applying steps
3,4,5 to a 5082-4657/58:

Iv(el
Iv(O)

cas

FOR LAMBERTIAN

Ev (x)

w 1.0
u

1/8)

~
...J
<t

/'

<t
0

;';'

"e

""

=t:!>~ --- __

.8

11V(O)

-d

c::

T
--l

Ev(O)

z
w

.6

>

<t
...J
W

c::

.4

~>

....

.2

e -

12
OFF-AXIS

Figure 2.4.5-4

16
20
ANGLE-DEGRE:ES

24

Normalized Incidance on Diffuser Obtained from

Radiation Patterns.

incidance. A little more "cut-and-try" on paper is much less

costly.
however,
than "scrap-and-try-again"
on the
workbench.
5.

liO) =2 8.56 Ix

(32.47mm)2 = 30.11 mcd

This is marginal for the 5082-4657, but the 5082-4658 will

do, with IF ~ 16 mA.
Note that the exercise in trying the 5082-4350/55 could
have been avoided by first computing the intensity of the
required legend area:
Iv OUT = Lv OUT x LEGEND AREA

Another POSSIBLE tradeoff would be to allow a slightly

greater non-uniformity, i.e., Eix)/Ev(0)<0.5.
This would
permit a larger e, and hence a smaller d and Iv(O) would be
possible. Remember that liO) requirement varies jointly as
the SQUARE of d.

(2.4.5-11)

= 20 cd/m2 (10 mm x 20 mm)

= 4 mcd

A single LED would hardly be likely to product a legend

area intensity greater than the basic LED intensity.
Note also that the solution with the 5082-4658 requires a
distance behind the diffuser of 32.47 mm, which may be
inconveniently large. By using more than one LED, the ON
sterance and uniformity requirements over the entire legend
area can be met with much less distance required behind
the panel. The design procedure is a bit more complicated
because it is necessary to take account of the overlapping
incidance from adjacent LED(s). The 10mm x 20mm box
should not be evenly subdivided (e.g. 2 LEDs lOmm apart)
because
the corners do not receive the overlapping

Figure 2.4.5-5

Rectangular LED Backlighting

Legend Areas.

Small or Tall

2.4.5.3 Backlighting
Legend Areas

Construction

for

Small-to-Medium

For illuminating small symbols, the simplest approach i~;the

use of a rectangular LED, designed for the purpose and
having built-in diffusant that meets most uniformity
requirements. As shown in Figure 2.4.5-5, it can also be
clustered to illuminate fairly large areas. At their specified
input current levels, their ON sterance easily exceed~: 40
cd/m2. All that is needed with these is the symbol-bearing
legend plate and perhaps a contrast enhancement fIlter. If it
is desirable to have separately-illuminated
areas in the
cluster, thin black tape separators are adequate for crosstalk
control.
The bodies of these rectangular LEDs are
translucent, so for good OFF/BACKGROUND CR they
should be shielded from stray light behind the panel -- more
black tape. Painting is not recommended.
Another simple approach uses LEDs made with undiffu led
encapsulant as in Figure 2.4.5-6. This "direct-illumination"
method uses the basic design consideration of SecLon
2.4.5.2 to select the LED and spacing, d, for proper ON
sterance and uniformity. This approach is typically used for
fairly small legend areas lOmm
diameter); the large
(lOmm x 20mm) area was prescribed in the earlier
examples for purposes of illustrating design tradeoffs.
The shortest distance behind the panel is achieved by using
an "egg crate" reflector in which each LED is near the apex
of a conically- or pyramidally-shaped reflector cavity. As
seen in Figure 2.4.5-7, the reflector can be made to either
combine more than one LED to backlight a large legend, or
provide separation with a "web" to isolate separatdy
energized areas. Egg-crate reflectors offer the additional
benefit of re-directing side-emitted flux from the LED so
that efficiency is raised and uniformity is improved.
Because much of the flux reaching the diffuser comes
indirectly from the reflector rather than directly from LED,
this arrangement is described as "indirect" illumination.
Figure 2.4.5-7 also shows the recommended placement ::>f
the diffuser,
1e gend plate, and (if used) contrast
enhancement filter. Placing the legend plate behind the
diffuser would "break" the symbol edges and lower the
OFF/BACKGROUND CR; so unless fuzzy symbol edges are
desired, the legend plate should be located as ShOW1.
Contrast enhancement fIltering is always out in front.
Another method of indirect illumination uses a plastic
"light pipe" as in Figure 2.4.5-8. Except for the hole 10
accept the LED, the plastic is solid and relies cn
greater-than-critical-angle
reflections for its efficiency.
Scratches and contact with materials other than air came
light losses at the surfaces, so these surfaces should be
protected. Over a length of more than four diameter:;,
multiple reflections cause enough diffusion to perfil t

elimination of the separate diffuser; the front end of the

"pipe" may be roughened if the length is marginal.
2.4.5.4 Backlighting Construction
Areas

for Very Large Legend

For very large symbols to be viewed from large distances

requirement is sharply reduced.
This permits the LEDs to be viewed directly through the
symbol with NO DIFFUSER. To obtain highest possible
efficiency (or ON sterance), each LED in the arrays of
Figure 2.4.5-9 should be seated in a cell of an "egg
crate" -type reflector. The average ON sterance is then
found by dividing the single LED intensity by the area of a
single cell. In Figure 3.4.5-9, notice that both arrays have
the same number of LEDs, but different aspect ratios.
Except for this, the "honeycomb"
advantage over the
"square"
is rather small (a factor of only 2/.../3).
Overlapping incidance is likely to leave smaller "cold" spots
in the "honeycomb" than in the "square".

(> 10m) the uniformity

The terms "communications and signalling" are intended to

categorize all those applications in which the primary
function of the LED is to provide flux for detection by
means other than human vision (but may include some of
these as well). Such applications include:

Card/tape reader (esp.low hole/no-hole ratio)

Tape loop stabilizer (max and min loop sensors)
End-of-tape sensor (reflective or transmissive)
Optical tachometer (motor speed control)
Assembly line monitor (parts counting/orientation)
Bar code scanner (POS machine UPC)
Opto-mechanical synchronizer (ignition timing)
Safety interlock (with phase-lock loop in high
ambient)
Carriage travel sensor (beam break or reflect)
Shaft position encoder (using arrayed devices)
High voltage isolator (air gap or fiber optic)
Smoke
detector
(both
scatteringand
obscuration-type)
Densitometer (chemical analysis)
Liquid level monitor (clear as well as opaque)

Their
shorter
wavelengths
and higher modulation
bandwidths give LEDs performance superior to IREDs in
many of these applications, despite the higher quantum
efficiencies of IREDs. Now new devices are available with
wavelengths
short enough to benefit from spectral
considerations, but with quantum efficiencies so high (up
to 1.5%) that they rival that of amphoteric IREDs. These
new devices, emitting at 670 nm and 700 nm, are
sufficiently visible that optical alignment can be done
visually without the spectral viewing equipment required
with IREDs.

~ > 4 ELIMINATES

DIFFUSER

d~

Figure 2.4.5-6

Direct Illumination; Undiffused

Edges when "d" is Small.

Figure 2.4.5-8

Plastic Light Pipe; Reflects Side Emission and

Diffuses Forward.

Figure 2.4.5-9

LED Clusters - Honeycomb or Square Patterns

for Very Large Legend Area.

LED Makes Sharp

LEGEND PLATE
MATTE SURFACE

GAP -

LARGE ENOUGH TO

REDUCE "COLD" SPOT

ACCEPTABLY

Figure 2.4.5-7

Indirect Illumination;
Improves Uniformity.

Egg-Crate Reflector

2.5.1

Device Characterization
Signalling

for Communications

and

Thus the flux into a particular N.A. is the product of the

specified axial intensity, multiplied by the normalized
integral:

Specifications of luminous intensity or luminous sten.nce

are not directly applicable in most signalling applicati'Jns.
However, they can be converted to radiometric quantitil~s if
the luminous efficacy, 1]v' is known. If 1]vis not given in the
data sheet it can be found by the method described in
Section 7.1. Equation 2.5.1-1 gives the relationship te be
used in making the conversion:

(N.A. ) = I e X ( (8)
I

I _ 3.5 X 10-3 cd
e
1471m/W

sin-1(N.A.l

S - NORMALIZED
SPECTRUM OF THE SOURCE
M - SPECTRAL TRANSMITTANCE
OF THE MEDIUM
RO - RESPONSIVITY OF THE DETECTOR

Conversely, if a device used in signalling is used also for

visual effects, if 1]v is known, the photometric properties
can be obtained from given radiometric quantities.

To obtain the amount of flux that is radiated into a cone )f

half-angle e, or included-angle 2e, it is necessary to evaluate
the integral:

Wavelength compatibility can be of critical importance in

dealing with mediums having sharply varying spectral
transmittance or reflectance, or with detectors such as
photoconductors, since they have relatively narrow spectral
response. The effective flux coupling is found by spectral
integration of the product of all spectrums involved in the
system, as shown in Figure 2.5.2-1. In addition to those
shown in the figure, there may also be other spectral
effects, and these should be included in the product to be
integrated.

The most common conversion requirement is for intensity.

For example, if an LED has a luminous intens.ty
Iv = 3.5 mcd and a luminous efficacy 1]v = l47lm/W, t~en,
since 1 cd = 1 lm/sr:

Another important parameter in signalling is the radiation

pattern. This is usually given in the form of I.(e), the
relative intensity normalized at e = 0 so Ir(O) = 1.0. The
intensity at any angle is then the product of the axial
intensity multiplied by Ir(e).

~
I
W
..J

.8

>
>-

<>: :;

bl ~
W

.6

~ !l5

<>: W
..Ja:
Wa:
a: 0
1>-

.4

:;; &l
(I'J~ ~

o
I
o

.2

a:

e
(8)

J le1r(e)[27rsinel de

o
Here Ie is the axial intensity. The integral can be done
mathematically if lie) is a reasonably describable functi01.
If it is not, a summation is done by the method of Sectien
7.3.4.
When used with an optical system for which the numeric~1
aperture (N.A.) is given, the amount of flux the LED
radiates into the system is found from equation 2.5.1-:1,
since e = sin-1(N.A.). To aid designers in quickly evaluating
this flux, the data sheets for the new 670 nm and 700 mn
devices give the results of equation 2.5.1-3 in normalizd
form:

J Ir(8)[21Tsinel de

Figure 2.5.2-1

Integration to Derive Coupling of Spectral Source,

Medium, and Detector.

Stability of the LED flux is sometimes also a matter of

concern. It is unwise to depend on any randomly chosen
LED or IRED to be a source of stable flux, even if the
forward current is precisely regulated. LEDs used as optical
standards would be no exception except that they are
elaborately heat-sunk and are operated intermittently at a
forward current that is far below rated maximum; even
with these precautions, the ambient temperature must be
noted and appropriate correction applied.
For applications requiring stable flux from potentially
unstable LEDs (e.g. for photometer or radiometer transfer
standard), the best technique is the use of a beam splitter

D100%~

:..-

---5%

----

~+ ~

501

90%

---- ---- ---- MAIN BEAM

PLANE GLASS
BEAM SPLITTER

~.

2.5.3-1 and the appropriate flux integration method can be

selected from Section 7.3.4.
Although
N.A. is usually used in characterizing a
finitely-focussed optical system, a handy relationship using
the fino relates the incidance at a target, Ep to the
sterance of a source:

4(f/no)2(1

d
.2')2
dS

In equation 2.5.3-2 the transmittance, T, depends only on

reflection losses and is usually greater than 0.7. The
source-to-lens distance, dS' and lens-to-target distance, d ,
T
are, respectively, the object distance, do and image
distance, di relating to focal length, f, and magnification, m.

Figure 2.5.2-2

Servo Stabilization; Main Beam Flux Linearly

Related to Input Voltage.

d
m=....!.
do

and a PIN photo diode in a servo system as shown in Figure

2.5.2-2. This arrangement is especially good at wavelengths
below 800 nm where PIN photodiodes have a temperature
coefficient of zero. If the LED used in this arrangement has
a very narrow beam angle, the relative position of the
photodiode should be adjusted to give the same numerical
aperture (N.A.) of the optical system into which the main
beam is directed. Obviously, mechanical movement of the
components is not tolerable.

A fundamental principle of optics is that sterance is a

constant. That is, as the image passes from one image
position to another through successive lenses, the intensity
per unit area of each image is reduced only by reflection
loss (Fresnel loss) at the lens surface. The same is true for
the image-to-object sterance ratio if the object is a
lambertian emitter. It is also very nearly true when the
object is an LED, as long as the cone angle for flux
acceptance of the optical system is not so large that there is
great deviation from a lambertian pattern. This deviation
can be derived from the relative intensity, Ir(8) of the
radiation pattern, and expressed as:

Regardless of how non-lambertian the radiation pattern is.

the quantity of flux coupled into an optical system of given
numerical aperture (N.A.) can be calculated from the LEI:
radiation pattern. "Numerical aperture" is defined in Figure

(e) THE SAME FORMULAS AS FOR THIN LENSES APPLY TO

CURVED MIRRORS. LOWER I-NUMBERS ARE MORE EASILY
OBTAINED WITH CURVED MIRRORS THAN WITH LENSES.

Figure 2.5.3-1

Simple Thin Lens Formulas and Definition of

"Numerical Aperture".

The relationships of equations

2.5.3-2, -3, -4 are not
adequate for designing precision optics but are good enough
for most signalling applications of LEDs.
One application where a lens (or lenses) enhance LED
performance is in aperture or edge sensing as in FigUle
2.5.3-2. The edge moves down to cut the beam between
source and sensor (target). With no lens, the beam diameter
is as large as the LED and the "sharpness" of the cutting
depends on how small the LED or the sensor can be.
Moreover, the source-to-sensor coupling is at the mercy cf
the inverse square law.
The ideal situation for edge or hole sensing is to have arl
extremely tiny beam diameter at the beam cutting edge anj
a large flux coupled through to the detector when the edge
is not breaking the beam. Thus two lenses are ideal. The
"source lens" images the LED at the plane of the
beam-breaking edge, then the "target lens" images the LED
image on the sensor.
Confocal lens adjustment is obtained whenever two optical
systems having the same N.A. are focused on the sam~
point from opposite directions. All the flux entering th~
confocal point from one lens passes through it and enters
the other lens. In Figure 2.5.3-2b, if the "target" lens had a
shorter focal length, it could still be confocal, but dT would
be shorter.
Applying equation 2.5.3-2 to the arrangement of Figun~
2.5.3-2b, with dS= dT yields the incidance to sterance ratio
(ET/L~ shown in the figure. By comparing this result to
that of the NO-LENS arrangement in Figure 2.5.3-2a, the
coupling improvement factor of 5.26 is found. Not only i;
the coupling improved, but the beam diameter at the
beam-cutting edge is reduced by the ratio dS/dSL ~3.
In reflective pattern sensing, it is possible to operat(:
without lenses, but, as in edge sensing, the signal and
resolution are improved by the use of lenses, as in Figure
2.5.3-3. The LEDs used to irradiate the bar code should no:
be imaged at the plane of the code; although this would
increase the flux coupled to the photodiode, it might caus,:
interference patterns from interaction between the code
and details of the LED image. That is the reason a cluster
should be used for LED 1; a single LED 1 could cause ar
undesirable interference pattern. That is also the reasor.
that LED 2 (if it is used instead of LED 1) should bE
imaged by LENS 2 at the plane of LENS 1, or slightly tc
the left of it. With the image of LED2 focused at the plane
of LENS 1, it will be defocussed at the bar code plane and
the possibility of pattern interference is reduced.

The line-resolution of the code is the diameter of the

photodiode multiplied by doD/diD. In accordance with
equation 2.5.3-2 the diS/doS ratio should be made as
small as possible, consistent with the imaging requirements.

In some applications adequate performance is obtained

without lenses to modify the size and radiation pattern of
the LED optical port. Two examples are discussed here:
smoke detection and tachometry.
Smoke detection by the scattering or reflective method
requires only proper baffling and placement of the LED
and photodetector, as shown in Figure 2.5.4-1. In both the
coaxial and radial arrangement, the photodetector
and
baffle are so positioned that in the absence of smoke, no
flux from the LED reaches the photo detector. Not shown
in Figure
2.5.4-1 are the supporting walls of the
smoke-sensing chamber, but these must also be designed
with regard to the possibility of stray flux - either from the
LED or outside.
If the stray flux reaching the
photodetector is steady, sensitivity is reduced only by the
square root of the stray flux amplitude. However, if the
stray flux is variable, sensitivity is linearly reduced.
For room-style smoke detectors either the coaxial or the
radial arrangement can be used, but for smoke detection in
a flue, smoke-stack, or air-duct, the radial arrangement is
preferred because there is no obstruction by the baffle to
passage of air or smoke. In very large diameter ducts or
stacks, projection lenses for both the LED and the
photodiode should be used. The optical arrangement should
maximize the volume defined by the intersection of the
beam projected by the LED with the reception beam of the
photo detector. This intersection volume must, of course,
exclude the walls of the duct or flue.
In tachometry, a beam-breaking edge can be used, but
without
lensing the optical ports of the LED and
photodetector
might be too large to give adequate
resolution. For example, if a 25mm diameter disc is to
resolve a shaft rotation in one-degree increments, the line
spacing at the perimeter woJId be:
x =

b.e"1T" 0
0

360

0.22 mm

By placing next to the rota ing disc a stationary "shutter"

disc haVing the same line s~acing, when the rotating disc
moves there are alternate per,0ds of flux transmittance and
no transmittance at intervals as small as whatever line
spacing can be achieved. Thi allows the optical ports of the
LED and the photodetectori(on
opposite sides of the disc
pair) to have arbitrarily-sized optical ports. If each disc has

_,_+_,_
dS

dSL

FOR dS = dT AND dSL = dST

m = dSL

dS

1TL

ET=--S
4

4(f/no)2

'--."
~
CONFOCAL SPACING

LENS
INCIDANCE RA
NO LENS
ASSUMING
Figure 2.5.3-2

.8, (f/no)

no

4(f/no)2
0.65, d

x (-!L)2
DS

~ 5.26

'0 mm, DS = 3 mm

Edge Sensing; Lenses Increase Coupling and

Sharpen Precisio 1.

n lines per revolution, the photodetector output will rise

and fall n times per revolution. This gives shaft speed
information, but no directional information.
If directional information is desired, a stationary disc with
(n+!) lines next to a moving disc with n lines per revolution
will produce a Moire pattern rotating n times per shaft
revolution in a direction opposite that of the shaft. With
equal alternating light/dark lines, the Moire pattern will be
dark on one side of the shaft and have 50% transmittance
diametrically opposite. If the stationary disc has (n+2)

lines, the Moire pattern has two cycles per revolution and
rotates n/2 times per shaft revolution in opposite direction.
With (n+3) lines, the Moire has three cycles per revolution
and rotates n/3 times per revolution, again in opposite
direction. To obtain Moire rotation in the same direction as
the shaft, the stationary disc line spacing should be (n-l),
(n-2), etc. Obviously, the stationary disc pattern need not
0
go all the way around the shaft; it needs to cover only 180
of the Moire pattern cycle to allow LED/photodetector
0
pairs to be spaced at 90 of the Moire cycle.

SMOKEDIFFUSED
()

~~/~"';-~_7

BEAM

~
- ... ~.~

/V'----~

...

"//.A~~

S(~-.~~~~-----

MAGNIFIED
BAR CODE
IMAGED AT
PLANE OF
PHOTODIODE

~~~~BAFFLE
~--

------~

CF

-"~~~DETECTOR

LED 2'

'OPTIONAL

INSTEAD OF LED 1

LED 1 SHOULD BE A CLUSTER

OF TWO OR MORE LEDs, EQUALLY
SPACED AROUND THE EDGE
OF LENS 1

~
EXTENSION OF CHAMBER AT
EDGE OF APERTURE IN
BAFFLE

W.I
~

A-A
Figure 2.5.3-3

Reflective Pattern Sensing with Coaxial Lens

Arrangement.

The electrical circuit to be used in detecting the directi:m

and speed of the shaft rotation is very simple, as seen in
Figure 2.5.4-2. The dual J-K flip-flop in Figure 2.5.4-2(a) is
the basic direction detecting element giving one pulse per
Moire cycle at QI for clockwise (CW) rotation and one
pulse per Moire cycle at Q2 for anticlockwise (ACW)
rotation. While QI pulsates, Q2 is stea.dy ~ow, ~d vice
versa, so direction and velocity information IS available. If
Ql and Q2 are connected to the UP and DO~. in~u ts,
respectively, of an up/down counter, the count WIllmdicate
shaft position providing the~ are no extraneous pulses.
That is the reason for the Q-to-J feedback. After a CW
clocking of Ql (CKl), Ql cannot return to zero until t le
0
Moire pattern has moved 90 , either forward (NORMAL
CR ) or backward (BACKWARD CRI). It is therefore
esselntial that BACKWARD CRl be accompanied by cloGk
(CK2) of Q2' This will allow a DOWN count to cance~ 'ill
UP count if the shaft is vibrating through more than 90 )f
the Moire cycle.
To make sure there is no race at the CK and CR inputs,
hysteresis
should
be used in the photodetector
amplifier/comparator,
as in Figure 2.5.4-2(b). The feedback
diodes, D3 and D4 are mainly for high spe~d operation
restrict the input and output voltage excursIOns. Hysterem

.:?

ReflectiveType
Arrangements.

Smoke Detector Optical

can (and should) be used if they are omitted, but the values
will be different. Ip LH' the photocurrent needed to cause
an L-to-H transition 'at the output will be the same:

I
P,LH

cc
=-R
3

but the photo current at which an H-to-L transition occurs

will be:

cc

V
Ip,HL

R;-

(---)

R,

R2

+ R2

(2.5.4-3

Also, VOH will rise to within a few millivolts of VCC, but

VOL ::::::0.
The line patterns forming the Moire patterns need not be
integrally related. Although the Moire pattern will always
0
shift through 360 for a one-cycle shift of the moving
pattern, the physical spread of the Moire pattern can be
made whatever is necessary using the relationships in Figure
2.5.4-3(a) for radially aligned bars. Linear motion, of
course, can be interpreted by differentially spaced parallel
bars, using the same relationships.

A~
B~

BACKW,Ll,RD
CR1

CK
1

'''t"*-rq;: V~~~MAl

...-----_._-----

CK2

r'-~-'

NORMAL
CR2 -----..' ~
______
74LS73
(DUAL J-K)
(a)

-~BACKWARD
CR2

n
~

'A

r-~...,

i.L

-J

FOR Q1 AND C! WAVEFORMS, SOLID LINES REPRESENT

CLOCKWISE R,ftATlON,
DASHED LINES ANTICLOCKWISE

BASIC DIRECTIONAL
FOR ANTICLOCKWISE

SENSING. FOR CLOCKWISE

ROTATION, Q2 PULSATES.

ROTATION,

Q1 PULSATES;

VCC
+5V

}VD
Vo
(TO A, B INPUTS OF J-K)
Ip,LH
Ip HL
,

1
VD
1---

R1

VCC R2

(b)

D1
D2
D3
D4
U1

R1
10k

COMPARATOR WITH HYSTERESIS TO PREVENT TROUBLE

J-K INPUTS WHEN BACKWARD CLEAR OCCURS.

Figure 2.5.4-2

Quadrature Phase Detector for Tachometry

Direction Sen ;;ng.

with

HP-HEMT-3300
HP - 5082-4220
1N914
HP - 5082-4487
CA3130

WITH RACES AT

--\X2j-~

1
+~---MOIRE INTERVAL

P = x, x2
x,-x2

OVERLAP PATTERN
(MOIRE)

x, I--

----1-1
MOIRE "VELOCITY"

ll.L
L'lx2

=~_

x,-x2

x,/x2
(b) DIAMOND MOIRE FROM ANGULAR
POSITIONING OF ARBITRARILY
SPACED LINES

=,

.1!

'!L_ ~

2 2

tan 8

x2/2
sin e /2

Figure 2.5.4-3

=cose

GENERAL
TAN-'

Moire Pattern; Used to Obtain Direction Information

in Tachometr'l.

sine
)
x,/x2- cose

x, x2
Jx,2

+ xl-

2 x, x2 cose

Moire patterns for quadrature direction sensing can also be

produced using an angularly positioned stationary pattern
of arbitrary line spacing, as in Figure 2.5.4-3 (b). The
geometrical derivation of the Moire cycle interval, P, and
angle, p, from the line spacing Xl and x2 at angle e is seen
from the parallelogram. The parallelogram has one side
(x2/sine) and a diagonal (Xl/sine) with included angle e, so
the other side (x3) of the parallelogram is found from the
Law of Cosines:

Solving equations 2.5.4-4,5 for P gives the general formula

in Figure 2.5.4-3(b). Notice that if e = 0, this formula
becomes the same as that for the differentially spaced bars
in Figure 2.5.4-3(a). The pattern angle, p, is derived from
the triangle formed in the right half of the parallelogram by
applying the Law of Sines:
X21 sine
sinp

Xl

I sine

Sin(1T-e-p)

which, solved for p, gives the general formula in Figure

2.5 .4-3(b ).
For either parallel bar or angular Moire pattern, the
recommended detector "window" width W is less than 90
of the Moire cycle, that is P/4. Making it larger does not
give appreciably higher photo current and requires overlap
of 90 spaced photodetectors. Taking the photocurrent for
W = P/4 as the normal:

W/P

0.25
0.10
0.05

RELATIVE
"LIGHT"
CURRENT

1.000
.434
.223

"LIGHT
DARK"
RATIO

7.01
19.00
39.00

A straight-line, angularly positioned stationary pattern can

also be used with a radial moving pattern, yielding curved
Moire patterns. However, if the radius of curvature is large
enough, relative to the line spacing, quadrature spaced
sensors with simple geometry can still be used.

Straightforward signal transmission as well as reflective

pattern sensing at large distances (>lm) can be done with
LEDs, but auxiliary optics are required.

The relationship between the sterance of the LED source

and the incidance at the target was given in equation
2.5.3-2 and is repeated in Figure 2.5.5-I(a). By applying
equation
2.5.3-3
the (f/no) drops out and the
incidance/sterance ratio is:

Notice that the term in parentheses is just the area of the

lens. It is as if an enlarged source, having an area equal to
the lens and having the sterance of the LED were radiating
according to the inverse square law 1/(dT)2 toward the
receiving detector. If another lens of diameter DR' and
focal length fR, is placed in front of the receiver and
focused on the apparent source, then again applying
equation (2.5.3-2) the net result gives the incidance E at
R
the receiver:

With a source imaged on a diffusely reflecting target, the

image becomes a source which can then be imaged on a
receiver. The expression for the incidance to sterance ratio
in such a situation is given in Figure 2.5.5-I(b). Except for
the reduction (~/1T)
due to target reflectance, the
expression is very nearly the same as equation 2.5.5-2; that
is, the incidance/sterance ratio varies as the product oflens
areas and inversely as the square of the product of focal
length times separation distance. Applying this principle
yields the result in Figure 2.5 .5-1(c) for coaxial mirror
optics.
The flat secondary mirror is not the best
arrangement, but serves to illustrate the principle. In
practice, the secondary mirror (the back of the source
mirror) would be slightly convex to increase the focal
distance to the detector so it would not require as broad a
reception angle.
Both the systems of Figure 2.5.5-I(b and c) can be used for
such applications as reflective scanning, remote obstruction
sensing, etc. Focussed on an empty space, with low
background reflectance, they can be used for smoke
detection.
For straightforward signal transmission, lenses are limited
by the inverse square law applied to the separation distance.
Also they require a rigid mechanical structure. For these
reasons, fiber optics are much more convenient and,
literally, more flexible.
The principle of light transmission by a fiber optic is shown
in Figure 2.5.5-2.
Snell's law states:

(al

LED IMAGED ON A DISTANT TARGET

PRODUCES INCIDANCE VARYING
INVERSELY AS THE SQUARE OF
THE DISTANCE AND JOINTLY AS
THE LENS AREA

RT - LAMBERTIAN REFLECTANCE.
LT =

___

1RT

1T

ET

d
T

ER

LS
INCIDANCE FROM DIFFUSE TARGET
IMAGED AT RECEIVER VARIES
INVERSELY AS THE SQUARE OF
THE DISTANCE

COAXIAL MIRRORS FOCUSSED

NEAR INFINITY LOWER THE fIno
AND REDUCE FRESNEL LOSS

......
...............
r

rS

R ......
.

'~
r2R __

Figure 2.5.5-1

~1

Long-Rallge Direct and Reflective Signalling

Using Lellses and Mirrors.

1'--

REFLECTOR""
1.5 mm
(.06 in.) DIAMETER

REMOVE
MATERIAL

"

""'\

PLASTIC FIBER-OPTIC
"" 1.0 mm (.04 in.) DIAMETER

/
/

EPOXY:

FIRST APPLICATION
OPTICAL COUPLING

FOR

SECOND APPLICATION
FOR
MECHANICAL
STRENGTH
(a)

EMITTER

ASSEMBLY

1. REMOVE I\~ATERIAL

AS FAR AS PO:;SIBLE WITHOUT

BREAKING

BOND WIRE.

2. WET ROUGH SURFACE WITH FAST-CURING CLEAR EPOXY; ADJUST

POSITION OF FIBER IN "WET" EPOXY UNTIL MAXIMUM FLUX IS
OBTAINED FROM FREE END -HOLD UNTIL CURED.
3. ADD EPOXY (2ND APPL'N)

FOR MECHANICAL

CERAMIC SPACER
INSULATES DIODE
FROM CASE

STRENGTH.

~NODE BOND WIRE

--.

PIN CHIP 1.25 mm (.05") SQUARE

1.00 mm (.04") ROUND

&:TOOIODE

PLASTIC FIBER-OPTIC
1.0 mm (.04") DIAMETER
~~

EPOXY
FI RST APPLICATION - CLEAR
FOR OPTICAL COUPLING

CATHODE
BOND WIRE

1. CAREFULLY

SECOND APPLICATION - OPAQUE

FOR OPTICAL SHIELDING, ALSO
FOR MECHANICAL
STRENGTH.

BREAK WINDOW AI\ID RINSE OUT SHARDS.

2. WET PHOTODIODE SURFACE WITH FAST-SETTING CLEAR

EPOXY; ADJUST FIBER FOR MAXIMUM PHOTOCURRENTHOLD UNTIL CURED.
3. ADD OPAQUE EPOXY TO MINIMI;~E

Figure 2.5.5-3

STRAY AMBIENT

Fiber Optic Assembly Recommendation

Coupling LEt' to PIN Photodiode.

FLUX.

for

For a ray passing from a medium of refractive index n1 in10

a medium n2, and incident at an angle (J 1 with the surface
vector of the medium boundary, (J2 is the refraction angle.
For n2<n1 as (J J is increased, there is some angle, (J c' at
which (J 2 = 90 , and if (J l>fJ c the ray will be totally
reflected. Rays entering at angles less than (J 0 cm
propagate; those entering at larger angles are lost in the
cladding of the fiber. There is, therefore, a numerical
aperture for fiber optics and the relationship of N.A. to the
indices of refraction of the core (n 1) and cladding (n2) is
derived in Figure 2.5.5-2.
In coupling LEDs to fiber optics, there are two principles to
bear in mind:

2.

make the source diameter less than the fiber

diameter and the fiber diameter less than fle
detector diameter.

_
SinO

J..,2 - "l

N.A.=---

"0

sin

",

=~cos8
0"0

Bel

30

",~
"0

= -"';1 - sin'"" 8"

c

", r-il22

=_V,_(~)L

"0
Figure 2.5.5-2

When joining LEDs and photodiodes to lower-loss fiber

bundles with lower transmission loss, it is usually more
effective to first attach a stub (~100mm) to the device,
then use a connector, such as those developed by AMP
Incorporated to join the stub to the bundle. To attach a
bundle directly to the LED requires first potting the end of
the bundle in a binder to keep the fibers from spreading
during attachment to the LED. Such spreading would raise
insertion loss even further.

"2

00 = "1 sin {~-

sinO

Although the plastic fiber has a large N.A., (low insertion

loss) transmission loss (in dB/km) tends to be rather high.
As seen in Figure 2.5.5-4, the fibers with lower transmission
loss have higher insertion loss; furthermore, such fibers are
usually more costly. Also shown in Figure 2.5.5-4 is the
inverse-square-law coupling loss of a lens system.

=C

"0

These principles are applied in the fiber optic assembly

illustrated in Figure 2.5.5-3. The reason for selecting a
plastic fiber optic is mostly cost, but also simplicity of
assembly. Large N .A. plastic fibers of the kind shown in the
figure are available from DuPont and from International
Fiber Optics.

Fiber Optic Meridional Reflection

Numerical Aperture.

.'"
J,

::::i
ll.
::l

"1

Derivation of

Lenses are not much help because, as seen in equation

2.5.3-2 the flux density cannot be made larger than it is it
the surface of the source. The end of the fiber shouB,
therefore, be as close as possible to the LED chip (die).

0
..J
C1
Z

CI)

\
40

\
\

\
\

50

\
\

,,
"

....

...

... .....
............
..

....

If the fiber diameter is less than the emitting area of the

LED, flux will be lost. Even if the fiber diameter is larg~r
than the LED emitting area, some flux is lost due to the
N.A. limit. That is, fibers with large N.A. couple with low~r
insertion loss.
Again, at the receiving end, flux exiting from the fiber aLd
missing the detector is lost, so a suitably large area detector
should be used.

Figure 2.5.5-4

/30

dB + 20 109'0 (L)
1m

---

Insertion and Transmission

Compared with Lenses.

Loss of Fiber Optics

In exposing photographic fl1m with radiation from LEDs, it

mayor may not be necessary to take careful account of the
spectral response of the emulsion. If there is doubt, the film
data should be checked. The use of colored filters may be
regarded as modifying the spectral response of the film.
Emulsion having a flat response over the spectrum of the
LED requires only proper use of the typical exposure
equation:

The units of E and t depend on how the film exposure

requirement is given. If H is in "meter candle seconds"
(mcs), then E is in lux (Ix) and t is in seconds. If H is in
"ergs per square centimer" (erg/cm2) a conversion can be
made to relate this to radiant incidance, Ee in p.w/cm2, and
exposure time:
=
=

10-7 joule/em2
10-7 w'see/em2

= 10-1 (p.w/em2hec

Inserting appropriate units in equation

exposure relationships:

2.5.6-1 gives the

In equation 2.5.6-9, ~ is in lux seconds (or meter candle

seconds) and can be used directly in equation 2.5.6-3
because spectral response is not considered. Combining
equation 2.5.6-3, -7, -9 and assuming T = 0.8 yields the
result:
(f/no)2
t(see)

Ee (p.w/em2) t (see)
10

The value of either Ev or Ee depend on how the LED is

coupled to the fIlm. With no lens, and the LED a distance,
d, from the fIlm:
I (med)
y
E (Ix) = 1000- --y
d2 (mm2).

If a lens is used, i.e., a camera, the relationship in Figure

2.5.5-l(a) applies, with dT/dS being negligibly small, so
that:
Ey(lx) =

71'

4 (f/no)2
E (p.w/em2) =
e

71'

4(f/no)2

Where spectral response can be ignored, the ASA fl1m speed

is related to minimum exposure requirement, Hm by:

S
ASA

for exposure to a density 0.1 above base-plus-fog density.

In most applications, however, this is a marginal exposure.
For a satisfactory film record of an LED or LED display,
the exposure should be 10 to 100 times the value obtained
from equation 2.5.6-10, either by lowering the fino or
increasing the time, t, the sterance Lv, or the fIlm s1eed
used in recording. The exposure requirement in erg/em for
use in equation 2.5.6-4 is obtained from the film
characteristic curve (D-H curve) for wavelength ranges in
which the film response is nearly constant over the LED
spectrum.
To account for spectral effects, the incidance and exposure
time requirements must be found by integrating:

H (mes) = Ey (Ix) t(see)

2
10'
H (erg/em) =

(ed/m2).

= 71' L

2
E-e(p.w/em ) t (see)

________

10

J 'Pr (A) dA
= -"'-----J S(A)''Pr(A)dA

(2.5.6-11)
(

/
2)
erg em

The spectral sensitivity, SeA) is the spectral inverse of the

exposure requirement for a particular density, as given by
the fIlm D-H curves and spectral response curves. SeA) has
the units (erg/cm2)-1. 'PiA) is the relative spectral
output of the LED and has no units. If a spectral filter is
used, its spectral transmittance should be included as a
factor in the integrand of the integral in the denominator of
equation 2.5.6-11.

An opto isolator consists of a photon emitting device whose

flux is coupled through optically transparent insulation to
some sort of photo detector. The photon emitting device
may be an incandescent or neon lamp, or an LED. The
trans parent insulation may be air, glass, plastic, or
fiber-optic. The photodetector may be a photoconductor,
photo diode, phototransistor,
photoFET, or an integrated
combination photodiodejamplifier.
Various combinations
of these elements result in a wide variety of input
characteristics,
output
characteristics,
and coupled
characteristics. This discussion will be limited to the sort of
optoisolator having an LED input, with a thin layer of
transparent
insulation separating it from a solid-state
photo detector. The construction of such an optoisolator is
shown in Figure 3.3 .1-1, and the schematic representations
are shown in Figure 3.3.1-2.

In the design of a photoemitter for an optoisolator, the

main concern is optimization of the coupling to the
photo detector. The parameters to be optimized are gain
bandwidth, optical port, and electrical characteristics.
A low series resistance is desirable, so a GaAs-basec
photoemitter is the best choice. A low forward voltage h~
also desirable, but not as important as optimization of thf
gain and bandwidth.
For GaAs-based photoemitters, with GaAs1_x Px epitaxy,
adjustment of x affects the wavelength, efficiency (gain),
and speed
of response (bandwidth).
Gain, G, and
bandwidth, B, normalized relative to their values at x = 0,
are shown in Figure 3.1.1-1, along with a curve of th(:
gain-bandwidth
product,
GB. It is clear that th(:
gain-bandwidth product is optimized for x ~ 30%. For thi;
reason, all HP optoisolators use photoemitters in which the
GaAsl_x Px epitaxy is produced with x ~ 30%, giving a
wavelength of A ~ 700 nm.
With x ~ 30%, the forward voltage is very nearly that of ;l
"standard" red LED, which is a convenience in some digital
applications (Section 3.6.1).
Optical port considerations for the photoemitter of all
optoisolator differ considerably from those of an LED.
LEDs are made with an annular emitting region around 1
centered
bonding
pad to give a large ratio of
apparent-to-actual emitting area. For an optoisolator th~
emitting area is as small as possible, consistent with current
density consideration, and the bonding pad is offset. Th~
offset bonding pad allows minimal obscuration (shadowing)

Figure 3.1.1-1

Gain (G)' Bandwidth (B) Optimization

Direct Bandgap Photoemission.

of the emitting area and allows

coupled. The small size reduces
and minimizes the variability of
in the spacing and alignment of
region.

in

the detector to be closely

the flux loss at the edges
coupling due to variability
the photodector's sensitive

In selection of the optical medium, insulation is a vital

consideration if the emitter-to-detector
distance is very
small (e.g., "sandwich" construction). If the distance is very
large, such as through fiber-optics, lenses, or other medium
(reflective or transmissive), the insulation is less important
but spectral transmittance becomes significant, especially
where
plastics are involved. The effect of spectral
transmittance
as related to spectral properties of the
emitter and the detector is discussed in Section 2.5.2.

Most optoisolators use a transparent junction coating tJ

reduce Fresnel loss at the surfaces of the emitter and the
detector.
Since junction
coating
compounds are
non-conductors,
this provides insulation as well. 11
addition, HP optoisolators have a layer of FEP film
(transparent Teflon) betweeen emitter and detector to
insure good insulation (see Section 3.3).
Fresnel losses result from reflection when flux passes from
one material to another having a different index of
refraction. The fraction reflected is given by:

R __("2 - "1) 2
"2 +"1

(3.1.2-1)

The fraction transmitted is then derived as:

4
7 =

2+

+ ~

"1

(3.1.2-2)

OUT
Cf'IN

;7;7172;

{ 0.470 for n ; 1
0.698 for n; 1.5 ~
_
0.99990 for n ;.J (3.6)(3.5)

T(n; 1.5) ; 1.49

T(n; 1.0)
Figure 3.1.2-1

package. This can

(a)

Hybrid - costly, but permits separate optimization

of the photodiode and its amplifier;

(b)

Integrated - reduces assembly cost but compromises

performance.

There are two ways to integrate the photodetector

and
amplifier. One of these is the use of a phototransistor, in
which the photodetecting
region is the collector-base
junction. The other is the use of a photodiode whose
photocurrent
is amplified by a transistor separately
integrated on the same chip. The photo transistor is less
complicated to produce but has some inherent performance
disadvantages relative to the photodiode/transistor:
poor
linearity and low speed of response.

"2

For an optoisolator having a GaAsP emitter (n = 3.6) and ,

silicon detector (n = 3.5), the importance of having <coupling medium other than air (n = 1) is shown in Figun
3.1.2-1, applying equation 3.1.2-2 to both the changes of
refractive index.

'P

the gain element within the optoisolator

be dC!nein either of two ways:

Refractive Index Effects in Optical Coupling

Medium of Optoisolators.

The basics of photodetection by photodiodes are discussed

in Section 4.1. Silicon photodiodes are very good detectors
but usually require ancillary gain elements to produce
adequate signal levels. To add gain with elements external
to the optoisolator package is an inconvenience, raises the
package count in the system, and degrades performance
(CMR, speed). For these reasons, it is desirable to include

A linearity comparison is shown in Figure 3.1.3-1. The

photo transistor's non-linearity results from the flow of
collector current in the collector-base junction, thus causing
reduction in the collector-base depletion region which in
turn reduces its responsivity. Notice that raising the
collector voltage restores the depletion region and reduces
the non-linearity.
In the photodiode/transistor,
the
collector current does not flow in the photo diode -- even if
the cathode of the photodiode is connected to the collector
in what is called the "phototransistor connection" (because
it can be operated as a two-terminal device). Best linearity,
however, is obtained by keeping fixed the voltage across the
photodiode.
The speed of response of a phototransistor is inherently
slow because the collector-base junction must be large in
order to capture photons, and therefore has a large
(~O pF) junction capacitance amplified by Miller effect
(see Figure 3.1.3-2). In the photodiode/transistor,
the
photo diode capacitance can be made lower ("'='10 pF)
because the junction thickness does not affect the gain of
the separately integrated transistor. More importantly,
however, the collector-base capacitance to which Miller
effect applies is extremely small ("'='0.5 pF). Furthermore, as
seen in Figure 3.1.3-2, much of this is contributed by
capacitance between external pins and pin connections. For
this reason, if no base connection is required, the rise/fall
time of HP 5082-4350/51 optoisolators is improved 30% to
40% by removal of the base pin. If this is impractical, or if a
base connection is required, the board layout should
minimize capacitance between the base and collector leads.
Because
the HP photodiode/transistor
isolator is an
integrated (not hybrid) device, the voltage at the emitter of
the output transistor must always be at or below the
voltage at any other point in the isolator IC. For example,
if the emitter is driving some above-ground point, such as

10

/'

E 8

:::

~"

PHOTOTRANSISTOR
CHARACTERISTICS

V/

a:
a: 6
::l
U
a:
0
Iu 4
w
...J
...J

0
U

_~ 2

/ ,/

//

.,,~

llC

"

I-

I-

+}VC

IIp+ IC

gj

'"

.-.

Nor can the emitter be permitted to float. For example, if

the photodiode alone is to be employed, the emitter should
be connected to the base. Failure to do so yields the results
shown in Figure 3.1.3-4. The collector may float but must
not be permitted to become negative with respect to the
emitter, so it also should be connected to the emitter.

2
4
6
VC COLLECTOR VOLTAGE

~
~ l-"
,IC

8
V

10

CBC

~ "-

Figure 3.1.3-1

,-,-,
I

Ip+IC

I-

u
W
...J
...J

o 4

~, ~

CIN

'+_<1-

r--

.!.-

J?
5
-U2

PHOTODIODE BIASED
FROM COLLECTOR
("PHOTO TRANSISTOR
CONNECTION")

CpD
""10 pF

""20 pF
r---

1.-

a:
a:
::l
u

PHOTODIODE BIASED
SEPARATELY

V"

./

./

V/
V

10

:::::

~
Ip

./

I-

+}VC

another transistor, and if base bypass resistors are used to

enhance the speed of response, the resistor to the isolator
transistor base cannot be grounded. Correct connection is
shown in Figure 3.1.3-3.

'"

IAv + 1) 20 pF

CIN

10 pF + lAy + 1) 0.5 pF

Ay =-3Vc/3Vb
0

2
4
6
8
VC COLLECTOR VOLTAGE V

PHOTOTRANSISTOR

PHOTODIODE/TRANSISTOR

Phototransistor Input Capacitance Compared

to Photodiode/Transistor.

Linearity of Photodiode/Transistor
to that of Phototransistor.

RESIDUAL CHARGE ON BASE OF

aX MAY MAKE V7_ < 0 AND
CAUSE EXTREMEL~ SLOW
OPERATION

Compared

Figure 3.1.~-3

Correct Connection of Base Bypass Resistor

with Emitter Above Ground.

NOTE CHANGE OF SCALE FOR VCE

>

IC

'":E
lL

-60
-80

..:
~
..:

~CURVE
~TRACER

'-0

TERMINALS

~CURVE
--~
TRACER

, ~

TERMINJ\LS

: ~OPTIONAL
REQUIREO

INCORRECT
FLOATING SUBSTRATE (PIN 5)
IMPAI RS PHOTODIODE RESPONSE
FOR VCE < B VBE ~ 5 VOLTS

Figure 3.1.3-4

MAKING V
= 0 ALLOWS NORMAL
PHOTODIO?>t OPERATION; COLLECTOR
MAY BE FLOATED AT SMALL RISt;.

Floating Substrate Impairment

Photodiode Performance.

of Integrated

Integration of a photodiode
on the same chip with
amplifying transistor(s) entails some tradeoffs. The silicon
resistivity cannot be as high as that for an optimal PIN
photo diode because it would then be impossible tel
integrate reasonable gain/speed transistors. On the other
hand, silicon resistivity low enough to make optima
transistors would not permit a reasonably good photodiode
Within the constraints of this tradeoff, (C/ A < 90 pF /mm2
for the photodiode and GB > 500 MHz for the transistor),
a variety of detector-amplifier
configurations
can b{
executed.
Analog amplifiers in HP optoisolators are of two types,
shown in Figure 3.1.4-1. One is the single transistor with
the photodiode
anode connected to the base and the
cathode separated for reverse bias connection (e.g. VCC).
The other type is called a split-darlington; the emitter of
the first transistor is connected to the base of the second
transistor in the usual darlington manner, but the collector
of the second is separated (split) from the collector of the
first transistor to allow the output collector to drop to a
lower VCE in saturation. In the usual darlington circuit, the
collectors are common, and the lowest possible output VCE
is the sum of V BE of the second transistor and VCE (SAT)
of the first. This results in a minimum output VCE of
"" 800 mY. With the split-darlington configuration, the
output VCE is not held up by the VBE of the output
transistor and may drop to less than 100 mY. This is
especially important in digital applications requiring a low
output VCE for good noise immunity. The base of the
second transistor is available for strobing and for speed
enhancement with resistive bypassing - handy features in

VBE2r
ORDINARY
Figure 3.1.4-1

DARLINGTON

AMPLIFIER

AnalogType Photocurrent

Amplifiers.

digital
applications.
The high gain makes
the
split-darlington
amplifier useful in many low power
applications and as a gain element in a closed loop where
the input to the main amplifier must be isolated from the
reference comparator (e.g. analog power supply regulation).
However, because the base of the first transistor is not
available for feedback connection, the split-darlington
amplifier is not optimal for linear applications.
For digital applications at moderate data rates, the analog
types can be used if their gain/output-current
capabilities
are adequate. For high data rates, the transistors must be
operated at lower closed-loop gain in order to achieve the
required bandwidth.
HP detector/amplifiers
for digital
applications have a high speed linear amplifier driving a
Schottky-clamped
output transistor, as shown in Figure
3.1.4-2. Bias for the photo diode is decoupled from V CC to
reduce the possibility of "chatter" (oscillatory transition

The reason for omitting hysteresis was not to permit analog

operation, but rather to permit maximum data rate. With
hysteresis, there would be a higher immunity to both
differential- and common-mode noise but the shifting
threshold would reduce the data rate capability. The effect
of hysteresis on data rate is the opposite of peaking.

BIAS& POWER
SUPPLY
REGULATOR

3.2

AE-INVERTERFORENABLE
INPUT
Ap - NON-INVERTING PHOTOCURRENT AMPLIFIER
Os - SCHOTTKY DIODES CLAMPING 0E AND 09
0E - ENABLE SWITCH (OFF, UNLESS "ENABLE'
IS LOW)
00 - OUTPUT TRANSISTOR
Figure 3.1.4-2

High Gain, High Speed Photocurrent

for Digital Applications.

Parameter Characterization

As seen in Section 3.1, a variety of input, output, and

coupled characteristics are possible, depending on the
choices of input and output devices and the mannel in
which they are optically coupled. Figure 3.2-1 illustrates a
type of optoisolator with respect to which all the important
parameters, analog as well as digital, can be visualized and
described. Listed in order of their importance:

Amplifier

from one logic state to another due to regenerative coupling

via the power supply line). The linear amplifier has a
tendency to be unstable if the high-frequency impedance of
the power supply is not low enough. For this reason, a
low-inductance bypass capacitor (0.01 ,uF ceramic disc)
should be installed adjacent to each isolator of this type.
This, and other chatter-suppression techniques are discussed
in Section 3.3.
The Schottky clamp is a metal-silicon diode in parallel with
the base-collector junction of the output transistor. A
metal-silicon (Schottky) diode has a lower turn-on voltage
than a P-N junction, so when the transistor is driven into
saturation, the Schottky diode bypasses the current which
would otherwise enter the base-collector junction. With
reduced current entering the base-collector junction, there
is a proportionate reduction in the charge to be removed
when the transistor is to be turned off, and thus the
Schottky clamp reduces the turn-off delay. The attendant
drawback is the 400 mV higher VCE (SAT) for a Schottky
clamped transistor.
The "enable" input has threshold voltage and input current
levels resembling a TTL input. However, it is not necessary
to apply a pullup resistor to insure its remaining high.
Unless the enable is connected to a strobe, it may simply be
left open. The strobe applied to the enable input may be
either open-collector or active-pullup.
With the enable high, analog operation is possible because
there is no hysteresis. However, the dynamic range is
limited. The lower limit is the threshold input current for
operation in the active region -- this threshold may be as
high as 4 mA. The upper limit is the maximum dissipation
rating on the output. Because of a touchy bias situation,
analog separation is not recommended for designs to be
mass produced.

5.

Forward
fan-out)

coupling (Current

Transfer Ratio, CTR;

The fundamental purpose of an isolator, whether optically,

electrically, or magnetically coupled, is to enhance, in the
output, the ratio of differential-mode to common-mode
signals. Optical coupling is superior to electric or magnetic
coupling because the photons that carry the differential
mode signal do not carry any charge or require a magnetic
flux to support their movement. Thus, the only means by
which the common-mode signal can appear in the output
are by:
(a)

modulating the input current and

Means (a) is not really a property of the optoisolator and

can be eliminated by impedance balancing, as seen in Figure
3.2-1, where Rpl and Rp2 can be selected or adjusted to
make aI-F/aeCM = O. Isolation characterization is therefore
examined with respect to (b) stray capacitive coupling.
Analog isolation is seen simply as the ratio of the relative
effects
of differential-mode
voltage,
eDM' and
common-mode
voltage, eCM' on output current, IC'
expressed as the Common-Mode Rejection Ratio, CMRR:
alc/aeOM)

CMRR

= ( aiC/aeOM

(eOM)
eCM

(3.2.1-1)

INTERNAL SHIELD USED

IN SOME OPTOISOLATORS\

*CCM "'" 0.07 pF MOST ISOLATORS

"'" 0.007 pF IF INTERNALLY SHIELDED

i;--~\------,
I
IF

--

1/

',,,,

I \'

~Ip

I
I

I
I
I

-~~

---

I
I
I
I
I

II

,I

I ,--I'l-:.-T-

I"

"//

CC:M I
I

C;/-<",__ _ I

Ib~

Vb +

I
..J

ANALOG:

a Ic/aeDM
CMRR t, ~-= a lc!aeCM

DIGITAL:

MAX. TOLERABLE COM. MODE FOR PROPER DIGITAL OUTPUT

CMRV ~ MAX. TOLERABLE eCM

CMTR ~ MAX. TOLERABLE decm/dt

CMH: CMRV or CMTR for Vc ~ VOH (INPUT DIODE OFF)

CML: CMRV or CMTR for Vc ~ VOL (INPUT DIODE ON)
-INSULATION:

VI_O ~ eCM ABOVE WHICH DAMAGE MAY OCCUR.

-SPEED:
ANALOG:

3 dB BANDWIDTH FOR dVC/deCM

DIGITAL:

DELAY IN PROPAGATION OF CHANGE IN LOGIC STATE, eDM" Vc

-REVERSE COUPLING: CI-O = C, + C2 + CCM

I
-FORWARD COUPLING: CURRENT TRANSFER RATIO, CTR t, ~ x '00%
= IF
Figure 3.2-1

Important Optoisc>lator Parameters;

Brief Description

Illustration

and

CMRR is often expressed in dB, in which case it is called

the Common Mode Rejection, CMR:
CMR

= 20

10910 (CMRR)

(3.2.1-2)

CMR cannot be specified without reference to the input

circuit design. This is seen by analyzing equation 3.2.1-1:
~
alb
_ alp
_ alp
aeOM = {3 aeOM - {3 aeOM - {3 al F

al F
aeOM

(3.2.1-3)

)-1

_ alp (
dVF
-{3 RS+a'F
dlF
alc

Ib
I
deCM
={3 = (3 aeCM a(CCM -dt
)
aeCM
aeCM
=

{3 CCM (2

1T

(3.2.1-4)

fCM)

Taking the ratio of equation 3.2.1-3 to equation 3.2.1-4

gives

The photocurrent-to-input
current ratio Op/IF) and
common-mode coupling capacitance, CCM' are properties
of the optoisolator,
however, CMRR can be made
arbitrarily high by making Rs and dV F/dIF small. Note also
that CMRR decreases as the frequency, fCM' of the
common mode signal rises. There is a limit to this effect.
The cutoff frequency of the amplifier limits response to
both eDM and eCM making the ratio unimportant. Because
of the non-isolator variables affecting CMR, any meaningful
description of this parameter must give conditions for IF'
Rs and fCM' Notice that CMRR is independent of the
amplifier gain, but does depend on the diode-to-diode
current transfer ratio, Ip/IF. This ratio is fairly constant:
Ip/IF ~ 0.0015 for all lIP optoisolators and is very nearly
the same also for non-lIP types in which the "photodiode"
is the base-to-collector junction of a phototransistor. Notice
also that CCM is only a small part of the total
input-to-output capacitance, which is given in data sheets as
CI_O. CCM also is fairly constant; CCM ~ 0.07 pF in most
HP isolators,
but in the internally shielded types
CCM ~ 0.007 pF.
At present, the internal shield is offered in only the very
high speed optoisolator, 5082-4361, in which the broader
bandwidth would, without the shield, allow CMRV to dip
lower as in Figure 3.2.1-1, or shift downward and leftward
the curve in Figure 3.2.1-2.
For the unshielded single-transistor analog types, CMR CaIJ
be improved by adding a neutralizing capacitor between the
collector and either of the input pins. The value of thi~
capacitor should be {3 x CCM ~ 7 pF. Neutralization car

also be used with dual isolators of the single-transistor type,

providing the neutralizing capacitor couples each collector
to its corresponding
input diode. Obviously, the
neutralizing capacitor must have a voltage rating compatible
with the application.
Digital operation requires that the output remain in proper
logic state despite interference from eCM' The extent to
which such interference is tolerable can be described in
either of two ways. For sinusoidal eCM' there is a
maximum tolerable amplitude, called the Common Mode
Rejection Voltage, CMRV, which, if exceeded, will cause
the output to change. However, a single value of CMRV
does not describe eCM tolerance. Since eCM is capacitively
coupled, CMRV varies inversely with frequency, as in
Figure 3.2.1-1, up to the amplifier cutoff frequency,
beyond which the slope is positive. For non-sinusoidal eCM'
i.e. transients, it is more convenient to describe the
maximum tolerable rate of change, aeCM/at, in terms of
Common Mode Transient Rejection, CMTR. Just as for
CMRV, a single value of CMTR does not describe aeCM/at
tolerance. The abscissa over which CMTR is examined may
be either the amplitude, eCM' or the duration, tTR, of the
transient, as in Figure 3.2.1-2 because:

Observations have shown that e' CM is very nearly

hyperbolic function of eCM according to:

(eCM - eCMO) (e'CM - e~MO) ~ CONSTANT (3.2.1-7)

The "CONSTANT" is a function of the gain and speed of
the amplifier in the isolator output. The asymptotes, eCMO
and e' CMO' describe what is intuitively cleo.r. That is, there
is some transient excursion amplitude, eCMO which is small
enough that no matter how high the rate of rise may be, the
isolator output will remain in its proper logic state.
Similarly, for a sufficiently small rate of rise e'CMO' the
excursion amplitude eCM is not limited (except by the
insulation,
Section 3.2.2). These asymptotes can be
described in terms of current and voltage increments 'lib
and 'Ivb at the amplifier input. To some extent, therefore,
there are some circuit design choices that affect the
isolator's defense against common mode transients.
Notice in Figures 3.2.1-1 and 3.2.1-2 that different curves
are shown, CMH and CMV representing common mode
voltage tolerance with the output in logic HIGH state and
LOW state, respectively. In general, reducing the value of
RL (see Figure 3.2-1) raises CMH and lowers CMV Raising
the level of input current, IF' raises CMV For sinusoidal
common mode voltage, polarity is of no concern because
the rate of rise equals the rate of fall. For non-sinusoidal
transients, it is worth noting that if the output is in a logic
low state, a positively sloped transient has no effect. Thus
in Figure 3.2.1-2, the CMH curve applies only to positive

I
I
CMTR
COMMON
MODE
TRANSIENT
REJECTION

COMMON
MODE
REJECTION
VOLTAGE

I
I
I
I
I
I
I

CML

FOR

eCM

<0

VC=VOL

I
I
I
I

---+----------------I

I
I
I

I
I
I
I

I
:
I
I

I
I

---------+--

\~

~----..J/

CMH
Figure 3.1.1-1

Sinusoidal Common Mode Voltage Rejection

Property. C MRV.

transients. Conversely, in similar fashion, the CML curve

applies only to negative transients. Therefore, if observation
or other evidence suggests that transients predominantly
have a higher rate of change in one direction than in th(~
other (e.g. a sawtooth), RL and IF can be adjusted, a',
above, to selectively improve CMH and CML. Such unequal
rates of change are common in circuits having diodes 0:other nonlinear elements.
The curves in Figures 3.2.1-1 and 3.2.1-2 characterize onl)'
the static immunity to common mode interference
Common mode sinusoids and transients can also affect the
dynamic performance adversely. For example, a transien1
occuring during a differential-mode logic transition can
affect the rate of rise or fall in the output circuit, thus
preventing a short bit from causing a proper logic transition
at the output. Also, high frequency common mode
interference can cause more than one output transition to
occur during a single transition of the differential-mode
signal.
(See Section 3.6.2 and 3.6.6 for defensive
techniques.)
Characterization
of dynamic
immunity
to any
common-mode interference can be done quite simply with
an up/down decade counter as in Figure 3.2.1-3. With a
reference data stream at the "up" input and the data

Figure 3.2.1-2

Transient Common Mode Voltage Rejection

Property. CMTR.

transmitted through the isolator at the "down" input, the

counter should only toggle between two adjacent states -either 4, 5, 4, 5 or 5, 6, 5,6. Preset at 5 re-sets both R-S
latches, causing LEDs 2 and 4 to glow. Should the isolator
fail occasionally to make an output transition, the count
will advance until a "carry" output sets the "carry" latch.
Should extraneous transitions occur, the counter will
decrement until a "borrow" output sets the "borrow"
latch. Perfect operation is indicated as long as LEDs 1 and 3
remain off in the presence of eCM.
CAUTION: getting a clean stream of reference data to the
"up" input in the presence of eCM may pose enough
difficulty to require extreme patience and heroic measures,
especially if very high data rates are involved.

The scheme shown in Figure 3.2.1-4 has the best chance of

yielding valid results. Since the differential count is all that
matters, the use of two 7 x counters allows the reference

UP/DOWN COUNTER

+5V
74LS192
CARRY
16 V OUTPUT
10
CC

lk

12

15 C
A
5

UP

11
4

---.L

2
DOWN

8 GND
BORROW
OUTPUT

*FOR AUTOMATIC
TEST, USE 100 Hz
OSCILLATOR
FOR
"PRESET"

CONDITION

FAil

1 B
9 D

PRESET*

PASS

Oe
LOAD

U
H

13

INDICATOR
2

OFF
ON
OFF
ON
ON
FLICKER
OFF
FLICKER

ON
ON
ON
ON
OFF
FLICKER
ON
FLICKER

Figure 3.2.1-3

NUMBER
3

ILEDI

OFF
OFF
ON
ON
ON
OFF
FLICKER
FLICKER

4
ON
ON
ON
ON
OFF
ON
FLICKER
FLICKER

Up/Down Counter, Differential

Sense Transmissioll Error.

DESCRIPTION

OF CONDITION

PERFECT OPERATION
TEST INPUT MISSING SOME COUNTS
EXTRANEOUS COUNTS AT TEST INPUT
HIGH-RATE MIXTURE, CONDITIONS B&C
PROBABL Y OK - CHECK PRESET
OCCASIONAL CONDITION B
OCCASIONAL CONDITION C
LOWRATE MIXTURE, CONDITIONS B&C
Count Detector

HI-CMR. LOW-SPEED
REFERENCE
ISOLATOR CIRCUIT

G-~

UP/DOWN
COUNTER
PRESET
AT"S"

HI-SPEED
ISOLATOR CIRCUIT
UNDER TEST FOR CMR

"Heroic Measures" to Assure Clean Reference

Count in Dynamic CMR Test.

isolator to operate at a lower data rate so speed can be

traded for better CMR. The isolator circuit under test must
operate at the full data rate in the presence of the same
interference. If the test is to be done with eDM being a
randomly coded data stream, the two -:-x counters should
be pre-set at the same count (e.g. 0000) with eDM at a
fIxed logic state. For most consistent results, inverters
should be used, as needed, so that at "preset" all counter
inputs are "zero" if they are negative-edge triggered or
"one" if they are positive-edge triggered.

A common, but usually erroneous, assumption is that

insulation can be operated at any voltage up to that at
which it breaks down. Many designers are unaware of
corona and its effect on insulation. Of those who are aware
of corona, some believe it occurs only at exposed terminals.
Corona (also known as "partial discharge") can occur
within insulation materials, particularly those having an
abundance of "microvoids". Due to inhomogenous electric
fields
within the material, the local field across a
"microvoid" can rise to a level at which there is a local
breakdown resulting in a partial discharge. This partial
discharge at one microvoid shifts the fIeld so it builds up
across some other microvoid. Each partial discharge causes
ions that locally degrade the insulation. The cumulative
effect over long periods of time (weeks or months) is to
lower the terminal-to-terminal breakdown voltage.
There is a voltage, called Corona Inception Voltage, CIV,
above which corona can be observed using specialized
equipment capable of responding to each partial discharge.
As the test voltage is raised above CN, there is an increase

in the rate of occurrence of these partial dischargf s.

Correspondingly, there is an increase in the rate at whieh
the insulation is degraded.
To maintain a high quality of insulation, lIP optoisolators
are given a special treatment called "backfIlling". After
molding, they are placed in a silicone oil bath in a chamber
which is evacuated until the gases from the microvoids ha ,e
escaped. Then the chamber is pressurized, forcing silicone
oil to fill the microvoids. With this treatment, CN is
believed to be well above the rated VI_O' but lOC%
production testing for corona is not practical; with the
methods and equipment now available, it is too time
consuming.
However, to insure rejection of devices with initially
defective insulation, there is 100% production testing with
3000 Vdc applied for 5 seconds at 45% relative humidity,
rejecting for leakage greater than 1.0 p.A.
The production test assures that each part meets the
Underwriters
Laboratories requirements for 220 Vzc,
50/60 Hz operation. The UL formula for prescribing the
test voltage:
line voltage = 220 Vrms
times 2 = 440 Vrms
plus 1000 Vrms = 1440 Vrms 50/60 Hz for 1 minute
or plus 20% = 1728 Vrms 50/60 Hz for I second
or times y2 = 2444 Vdc for 1 second
Except
for those
in hermetic
packaging, all HP
optoisolators easily meet UL requirements for status as a
"Recognized Component" under File No. E55361.
Another important property of insulation is its leaka:~e
resistance. In HP optoisolators RI_O is typically 10 12 ohrr s.
This appears to be an unreasonable claim when compawd
with the 1.0 p.A leakage allowed in the 3000 Vdc test. Here
again, it is a matter of practicality. With 3000 Vdc applit:d
to 1012 ohms, the resulting current is only 3 nA. Preci,e
observation of so small a current in a 5 second test is
impractical, so the RI_O = 1012 ohms in given as typical,
with periodic verification done on randomly selectl:d
samples. Both kinds of testing (RI_O and VI_O) are
performed with all input leads connected to one side of the
test voltage and all output leads to the other side -- a
two-terminal test.

For both analog

response is highly
characteristics of
digital modes, the
use of feedback

and digital operation, the speed of

dependent on the circuit as well as on
the optoisolator. In both analog ar.d
speed of response can be enhanced t y
and peaking. These techniques a~e

discussed in Section 3.5.6 for analog operation, and in

Sec tion 3.6.3 for digital operation. In this section,
discussion centers on isolator properties affecting speed of
response, and how the speed is characterized.
Analog operation requires the isolator to operate in its
"active region", i.e., with the output collector neither cut
off nor saturated over the required excursion range. In the
circuit of Figure 3.2-1, eDM' Rs' RpI, and R2 should be
selected for IF = (IFdc LlIF) such that 0 ~ VC ~ VCC
between the excursion limits of IF. Speed of response is
then characterized either in terms of 100/0-90%rise time if
LlIF is a step function, or in terms of 3 dB bandwidth if LlIF
is a sinusoid.
In the isolator itself, the principal bandwidth-limiting
elements are the photo emitter , the amplifying transistor,
and the capacitance of the photodiode. In the photodiode,
the photoelectrons are created within a few picoseconds
after photons enter. The resulting photocurrent flows to
the base with a rise time constant as seen at the base. The
base time constant depends not only on the transistor in
the isolator but also on the circuit used with the transistor.
If the transistor is operated common-emitter as in Figure
3.2-1, then the base time constant:
TB = RB (CpO + CBC) + f3 RL CBC

(3.2.3-1}

where:
CpO ~ 10 pF is the photodiode capacitance
Base-to-collector capacitance = 0.5 pF
plus stray external capacitance between
collector and base connections
RL

Load resistance

RB

Dynamic resistance to ground at the base

If no external resistance is added to bypass the base, then

RB is just the dynamic resistance at the base:
(3.2.3-2)

25 mV
Ib

= f3

(25 mV.'=
IC)

f3(

25 mV
CTR x I
100"10
Fdc

where CTR = Current Transfer Ratio (see Section 3.2.5).

Then in equation 3.2.3-1

TB= f3

Substituting
yields

CTR
(
100%

25 mV

x IFdc

typical values, CTR

20% and IF dc = 15 mA

Equation 3.2.3-4 shows the importance of a low value of

CBC. It also shows that the base time constant is limited by
R
only if R > 175[2.
Compare
this with a
L
L
~hototransistor for which CBC (",,::20 pF) is larger by 40
tlIlles; also, CpD = 0 and the base time constant is limited
by R > 8.33[2.
L
Although this is not recommended for high speed operation
with lIP optoisolators,
the transistor can be operated
"common
collector",
as in Figure 3.2.3-1. This,
incidentally, IS the recommended circuit for optimizing the
speed of a phototransistor.
Here the base-to-collector
capacitance is not Miller-effect multiplied so the base time
constant is:

Assuming the same values for CTR and IF in the expression

of equation 3.2.3-2 for RB, equation 3.2.3-5 becomes:

RB
TB =

((3+ 1) (CpO + CBC) [R L + (3+ 1 ]

((3 + 1) (C

((3+ 1) (CpO + CBC) [RL + 8.25[2)

PO

+C

BC

) [R

(3.2.3-6)

+ 8.33[2
1 + 1/(3

Notice now that the base time constant is limited by

RL> 8.25[2 ~egardless. of the relative values of CpD and
C
Companng equatIOn 3.2.3-4 and equation 3.2.3-6, it
BC
is clear that if CBC ~ CpD as in lIP isolators, the
common-emitter
circuit
is preferred
for superior
bandwidth,
whereas
for phototransistor
types the
common-collector circuit is better, particularly if the gain
requirement makes RL ~ 8.25[2.
Although in lIP optoisolators the photoemitter is not
usually the limiting element in speed of response, it should
be mentioned. When it is current-source driven, the step
response (100/0-90% rise/fall time) is approximately 20 ns.
This corresponds to a 3 dB bandwidth greater than 15 MHz.
For a given junction area and optical port, the remaining
chip design parameters that raise the efficiency also lower
the speed. This is because the photons are produced by
electron-hole recombination. The photon emission rate is,
therefore, proportional to the recombination rate, which is
proportional to minority carrier density, which in turn is
proportional to junction charge, Qj" Thus:
</>

= radiant

flux

ex: OJ

(3.2.3-7)

Since Q is the product of forward current IF' and minority

j
carrier lifetime, Tj:

1--------I
I

I
I

~Ip

J=
I
I
I
I

-.J

Figure 3.2.3-1

Common-Collector
Optoisolator.

Operation of Single-Transistor

It is clear that, for a fixed forward current, the speed

(inverse of lifetime) can be increased only at the expense of
efficiency.
If the photoemitter is the speed limitation in a particular
circuit, some improvement can be obtained by peaking. In
analog operation, peaking the photoemitter drive can also
compensate for speed deficiency in the output circuit.
Digital operation requires the isolator to switch from one
logic state to another. For the low state IF (Figure 3.2-1)
must be large enough to bring V C well below some defined
threshold, and for the high state IF must be low enough to
allow VC to rise well above that threshold. Speed of
response is therefore defmed with respect to the time
required for VC to reach the threshold in response to a
change of logic state at the input, i.e., switching IF from
one state to the other. Characterization of speed is either in
terms of propagation delay or data rate.
Propagation delay is the time required for a change of logic
state to propagate through the isolator and cause a change
of logic state in its load. tpHL is the propagation delay in
causing the isolator output to drop from the high state to a
specified threshold. tpLH is the propagation delay in
causing the output to rise from the low state to the
threshold. tpHL and tpLH are shown in Figure 3.2.3-2.
~e
isolator characteristics (CTR, (3, CPD, CBC, etc.)
mfluence tpHL and tpLH, propagation
delay is also
infl~enced by the circuit. Raising IF reduces tpHL by
causmg a large collector current, but raises tpLH by causing
the output to be more deeply saturated and thus increasing
the "storage" time. Raising the value of RL also reduces
tpHL by reducin~ the cur~ent it sources to the VC node but
un~ortunately this also raIses tpLH by reducing the current
available to pull VC up again when the input logic state is

HP 8z
HP8007
PULSE
GEN'

Zo =50n
5 ns

tr =

10% DUTY CYCLE

l/f < 100 /is

_I

RBX

IF MONITOR

~l

-16.F

OPTIONAL

Figure 3.2.3-2

Propagatioll Delays, tpHl and tplH; Definition

and Measurement.

changed. A resistor, RBX (Figure 3.2.3-2) in parallel with

the base-to-emitter junction aids in turning off the
transistor to reduce tpLH; adding RBX will require
additional input current to keep the output current as hign
-- and tpHL as low -- as it was before RBX was added. The
input current increment, IFX' is required to make up fer
the current bypassed by RBX' so its value depends on what
type of isolator is used. As a general rule:

the fast clock (tl), the result is the classical "eye" picture
in Figure 3.2.3-3c. Data are recoverable as long as the "eye"
is "open" above and below the threshold for a period of
time greater than the "set-up" time required before
clocking. Good practice would be to design for an "open
eye" duration about twice the "set-up" time. Applying this
rule gives the maximum data rate:
1
fNRZ(MAX)

Vbe)( IF)

'FX

-BX

(mA) """ R

tSET-UP + tp(MAX)

where tp(MAX) is either tpHL or tpLH -- whichever is the

greater.
The technique
applies
also to system
characterization;
that is, for the entire system (drivers,
lines, wiring, etc.) a tpHL(SYS) and tpLH(SYS) can be
observed and applied in equation 3.2.3-11.

700(V)
;'FX

'p

(kn)
BX

(3.2.3-10)
7(V)
(kn)
BX

'FX (mA) """ R

Making IF very large, then adding RBX reduces both tpHI,

and tpLH'
Data rate characterization of speed can be directly related
to tpHL and tpLH, as shown in Figure 3.2.3-3. Data ratei::
dermed for NRZ (non return to zero) data as the maximum
rate for error-free transmission of a random data pattern.
Random data implies that following an arbitrarily lonf;
string of consecutive ones or zeroes, there may be a strinf;
of ones and zeroes alternating at the maximum rate. ThE
minimum duration of the zero or one level is the reciprocal
of the maximum data rate in bits per second (b/s) wherea~
the maximum duration is arbitrary.

With RZ (return to zero) data, or other self-clocking

patterns, the concept of random data does not apply. Each
bit interval includes a period of time at the high state and a
period of time at the low state, as in Figure 3.2.34. With
such a pattern, the longest time spent in either logic state
cannot exceed one bit interval and the shortest is never less
than half a bit interval.
Characterization of the RZ data rate in terms of tpHL or
tpLH is valid, indeed equivalent, for systems with linear
transient
response and balanced threshold. However,
optoisolators
are usually non-linear. Furthermore,
a
transmission line long enough to affect the rise time will
exhibit non-linear response. The characterization of RZ
data rate should therefore be done with RZ data. A safe
rule in judging the data rate that can be expected is:
1
fRZ(MAX)

The pattern in Figure 3.2.3-3a produces the response in

Figure 3.2.3-3b when observed with a 'scope triggered from
the slow clock (t2). If the scope is triggered at every step of

<

2 tp(MAX)

where tp(MAX) is either tpLH or tpHL, whichever is the

greater.

RANDOM PATTERN
74~S;96
FAST
CLOCK

CLK Q

1"

.,01

.,02
74LS10
TRIPLE
3-INPUT
NAND

-.j

74 LS73
-;-2
1 CLK Q
or Q

SLOW
CLOCK

ISOLATOR CIRCUIT
UNDER TEST

1/2

~~

=~TO'SCOPE
VERT. INPUT

CLR

'SCOPE TRIGGERS

13

1--.,01 PERIOD

t1 (NEGATIVE SLOPE)
t2 (POSITIVE SLOPE)

I...

~)2PERIOD---

T----u1mul

(a) SIMPLE SCHEME FOR PRODUCING PSEUDO-RANDOM PATTERN

WITH INDEPENDENT INTERVAL ADJUSTMENTS, 'P1 AND .,02'

>

:>

l-

t-

l-

1-

::>

:J 4
~

IL-

tpHl

:J

< tpLH

tpLH

::>

C)

a:

IE: 3

I-

< tpHL

(::>

l-

.:( 2

<
..J

.oJ
(::>

CI)

en 1

:>

>

5
4
6
3
2
PERIODS OF FAST CLOCK

(b) WAVEFORMS OBT~NED

3
5
6
2
4
PERIODS OF FAST CLOCK

WITH TRIGGER FROM t2

>

:>

l-

I-

::>

:J
~ 4
l-

IL-

1-

:J

0 3

<::>

a:

IE:

<::>

I-

<
..J

::>

1-

c~

3
2

.oJ

<::>
(I)

en 1

:>

>

1
PERIODS OF FAST CLOCK

Figure 3.2.3-3

Pseudo-Rando n Code Generator

of NRZ Data Flate.

1
PERIODS OF FAST CLOCK

and Observation

BITINTERVAL~

~
[CLOCK.SNYCHRONIZED
1

DATA

ZERO~
CLOCK

)D-0IJTPUT

ONE~

~LOoP

EXAMPLE
DATA .
\--...;

....-

'--

__

--l

L-I

r-t~1

WORSTCASE RZ CODE GENERATORAL TERNATE STRINGS OF ONES AND ZEROES.

Figure 3.2.3-4

RZ Data Waveforms and Worst-Case Code Pattl!rn

Generator

An important consideration
in many applications of
isolators is the degree to which they prevent the flo'W of
ground loop current. Indeed, this is often the basis for use
of optically coupled isolators in preference to other kinds.
Ground loop current can be troublesome not only in the
module generating the signal to be transmitted, but also in
the module receiving the signal. As seen in Figure 3.210,
voltage drops in connecting wires can cause annoying
offsets.
Optoisolators
virtually
eliminate dc ground loops
(RI_O ~ 1012 ohms), but they do allow some ac ground
loop current. Internal capacitance between the input and
output circuits permit ac ground loop current according to
deCM
iGL = CI_O

(3.2.4-1)
dt

Note that the capacitance here is not merely the 0.07 pF of

CCM in Figure 3.2-1. The internally shielded optoisolators
have a lower CCM' but CI_O is the same in both kir ds.
There are no circuit tricks that can be done to reduce C[_O
or its effects.
In applications
where
the CI_O ~ 1 pF of most
optoisolators is intolerably high, the only recourse is the
selection of a type that has a larger physical separat ion
between the input and output circuits. Such types are
usually more costly because they require lenses or fiJer
optics to obtain adequate optical coupling between widely
separated circuits.

.-/
'"\. eCM

Figure 3.2.4-1

Wire-Resistance Offset from Ground

loop Via Input-Output Capacitance.

With CI_O = 1 pF, the ground loop current that flows from
a 60 Hz eCM is only 533 pA per volt rms. Even with 220 V
rms, the ground loop current is less than 120 nA. If several
isolators are used, however, or if eCM has a high frequency
or voltage, the resulting ground loop current should be
considered.

With respect to analog types of optoisolators (Figure 3.2-1),

the gain is simply dermed as the ratio of the output current
to the input current. This property is called the Current
Transfer Ratio (CTR) and is usually given in percent. If an
analog optoisolator is to be used in a digital application, it
is worth noting that the CTR is given for a very low
collector voltage. This assures a designer that for the
specified level of input current, the collector voltage will be
adequately low if the current available from the load (Ic in
Figure 3.2-1) does not exceed IF x CTR 7100%.
For optoisolators intended mainly for digital applications,
the CTR may be given differently. Rather than being
described as ratio of output/input
current, it may be
described as the "fan out" capability relative to some
particular logic family. For example, specification of a "fan
out" of 8 TTL gates means that the output of the
optoisolator can "sink" a current of 8 x 1.6 mA while
maintaining an adequately low output voltage. Where an
external pullup resistor is used, the current available from
the pullup resistor must be considered in the design.
A more common way of specifying CTR for a digital
application is to give the maximum value of the output
voltage for given levels of input and output currents. This
voltage, VOL' is the output voltage at logic low. For
example,
if VOL = OAV max at IF = 1.6 mA and
10 = 4.8 mA, this means that with V = OAV, the
o
minimum CTR is 300% and as long as the available load
current does not exceed 4.8 mA, the output voltage will be
less than OAV with an input current of 1.6 mA or more.
CTR is not constant for all levels of input current. This is
due partly to the supralinearity of the photoemitter, but a
larger portion of this variation is due to the change in the
gain, hfe, of the output amplifier, especially if it has more
than one transistor. This variation is described by giving
either the output current or the CTR as a function of input
current.

FOUR PIN EMITTER

LEAD FRAME

Figure 3.3 .1-1 shows the mechanical construction

of
Hewlett-Packard's 8 pin optically coupled isolators. The
GaAsP emitter and silicon detector are die attached and
wire bonded to separate four-pin lead frames. They are then
covered with an inert silicone junction coating. An
insulating fIlm is sandwiched between the emitter and
detector lead frames. Teflon FEP fIlm is used as the
insulating medium between the lead frames. This insulating
film assures the excellent low leakage, high voltage
insulation between input and output. Finally, the entire
assembly is encapsulated in epoxy to insure package
integrity.
The same 8 pin package can also be used for dual channel
isolators. Mechanical construction of a dual is similar to the
construction of a single channel isolator. Due to pin
limitations on the dual, VCC and GND of each detector are
normally connected together and brought out to two of the
four output pins. The remaining two output pins are used
for the output of each detector. This configuration retains
the high speed performance
obtained by biasing the
photodiode at a constant reverse voltage. Since the base of
the output transistor is unconnected, the base to collector
capacitance is lowered and the dual isolator has a somewhat
shorter propagation delay than the corresponding single
channel isolator. However, circuits that require feedback
applied to the base or an external resistor from base to
emitter to improve speed, cannot use an 8 pin dual channel
isolator. The pinouts of Hewlett-Packard's 8 pin single and
dual channel isolators are shown in Figure 3.3.1-2.

Some applications require higher reliability than can be

obtained with a plastic encapsulated opto isolator. In
general, if an application requires integrated circuits in
ceramic packages and metal can transistors for higher
reliability in severe environments, then a hermetically
packaged isolator should be used. Presently, hermetic
isolators are either packaged in a metal can or in a 16-pin
ceramic package. Figure 3.3.2-1 shows the mechanical
construction of Hewlett-Packard's 16-pin hermetic optically
coupled isolator. Silicon detectors are die attached and wire
bonded inside a 16 pin ceramic package. GaAsP emitters are
die attached and wire bonded to a separate ceramic insert.
Solder preforms are applied between the ceramic package
and
the
insert
is then soldered in place. Next,
the assembly is potted with transparent insulating material
to improve optical coupling and electrical insulation
between the emitter and detector. Finally, a metal lid is
attached to the package to insure a hermetic seal. The
fmished optically coupled isolator can withstand storage

GaAsP LED EMITTER

(UNDERNEATH
LEAD FRAME)

SILICON PHOTO
DETECTOR AND
AMPLIFYING
CIRCUITRY
FOUR PIN DETECTOR
LEAD FRAME

Figure 3.3.1-1

Mechanical Construction of Hewlett Packard's

Plastic Encapsulated Opto Isolators.

temperatures from _65C to + 150C, operating temperature

o
extremes from _55C to +12S C and 98% relative humidity
at 65C without failure.
These high reliability isolators are also available with
additional high reliablity screening for military applications.
Hewlett
Packard has established two standard high
reliability test programs, patterned after MIL-M-38S1 0, Class
B. These programs are known as the TX and TXB programs.

In many respects, six and eight pin optically coupled

isolators have direct pin compatibility.
Figure 3.3.3-1
illustrates this compatibility for six pin phototransistor or
photodarlington
isolators
and for eight
pin
photodiode-transistor
or photodiode-split
darlington
isolators. The eight pin package allows a separate VCC bias
supply that provides high speed operation and further
circuit versatility while still maintaining direct mechanical
compatibility.

---j

r.370

IA

(9.401

,""""'" I AI

rFiJ

TYPE

NUMBER

----.j
I

DATE CODE

~.120(3.06)

MAX.

I
~~

t!

.290 (7.37)
.310 (7.87)

--

.150 (3.811MIN.

1.170

---.i.-

(4.32)
MAX.

-.J ~

007 (0.18)
.013 (0.33)

.020 (0.511
MAX

.007 (0.18)
.013 (0.33)

,240 (6.10)
) .260 (6.561
290 (7.36)U

5 TYP.

:310~r

T
50824365
TX4365
TXB-4365
16

Vcc

8
VCC

15

'-...
'-...

GND
5082 - 4360

/
/

11

GND

GND

5082-4364
./
./

Outline Drawings and Pinouts of Hewlett Packard's

Opto Isolator!;.

10

GOLD PLATED
METAL CAP

3.3.4

GaAsP
LED EMITTER
(UNDERNEATH
CERAMIC INSERT)

SILICON
DETECTOR

16 PIN
CERAMIC PACKAGE

GOLD PLATED
LEADS

POTTING BETWEEN
CERAMIC INSERT AND
SILICON DETECTOR TO
IMPROVE OPTICAL
COUPLING AND
ELECTRICAL INSULATION

METAL CAP
GaAsP LED EMITTER

CERMIC PACKAGE
A

t
Figure 3.3.2-1

Mechanical Construction of Hermetic Opto

Isolator Manufactured By Hewlett Packard.

The only extra cost of this interchangeability

is an
allowance for two additional holes on the printed circuit.
These holes are used for pins 1 and 8 of the eight pin
isolator and left open for a six pin isolator. To provide
proper bias for the photodiode, pin 8 should be connected
to V cc. Since six pin isolators have an overall length of
8.89 mm (.350 inches) and eight pin isolators have an
overall length of 9.91 mm (.390 inches), both isolators
require 10.16 mm (.400 inch) spacing on standard 2.54
mm. (.100 inch) printed circuit pad spacing. Using dual
channel eight pin isolators, four isolator channels can be
realized in the same space as one 16 pin integrated circuit.
This direct
psn compatibility
holds
for all of
Hewlett-Packard's eight pin single channel optically coupled
isolators. The 50824350 family and the 50824370 family
offer switching speed advantages over photo transistor and
photodarlington
isolators.
The split
darlington
configuration of the 5082-4370 family allows the output
transistor to saturate as low as .1V for improved noise
margin with TTL circuitry. For a VCC of 5 volts, the
5082-4360 family is also directly pin compatible. The
5082-4360 family has extremely fast switching speeds, low
input current requirements and a TTL compatible enable
input.

Layout
Isolators

Considerations

for

Optically

Coupled

Optically coupled isolators have excellent common mode

rejection characteristics. However, for optimum common
mode rejection, special considerations should be given to
the circuit layout. Stray capacitance between the input
circuit and the output circuit should be minimized. This
can be done by physically separating all input and output
circuitry on the circuitboard. For line receiver applications,
the shielded cable should also be dressed properly to
minimize stray capacitance. Since the base of the output
transistor
is especially susceptible to common mode
transients, a special ground trace under the isolator can
serve as a shield. To further minimize capacitive coupling
into the base, pin 7 can be clipped off at the side of the
package. A simpler approach in an application requiring
more than one isolator is to use dual packages, because the
base is unconnected.
Circuit layout can also effect propagation delays through
the optically coupled isolators. Since most opto isolators
have an open collector output, excessive shunt capacitance
on pin 6 can limit rise and fall times of the isolator. For this
reason, the logic gate being driven by the isolator should be
as close to the isolator as possible and the pullup resistor
mounted in close proximity to the isolator. An external
resistor can be connected between pin 5 and pin 7 to
reduce tpLH of the 5082-4370 isolator. If this technique is
used, capacitance between pin 5 and pin 6 should be
minimized because of the "Miller" effect. Figure 3.3.4-1
illustrates some of these circuit layout considerations.

Some isolators such as the 5082-4360 series family require

an external bypass capacitor to prevent internal oscillations.
An isolator that is not functioning properly due to internal
oscillations exhibits the following symptoms: the isolator
has an extremely low current transfer ratio which requires
an excessively high input LED current to cause the output
collector to saturate. If noise is coupled into the isolator
from its load, the isolator may have multiple transitions for
a single input pulse. To prevent these internal oscillations, a
low inductance .01 p.F ceramic capacitor should be placed
between pins 5 and 8 as close to the device as possible. For
optimum results, the bypass capacitor should be connected
in such a manner as to minimize the coupling of noise
generated by the isolator load into the VCC and ground
connections of the isolator. This can be accomplished by
using separate VCC and ground lines for the isolator or
isolators than those used by the digital logic. A second
technique to minimize noise coupling is to layout
the
printed circuit with a topology that connects the VCC and
ground for the digital logic between the power supply and
the VCC and ground of the opto isolator. These techniques

+VCC

+VCC

50824350/1/2
B

+VCC
50824370/1

SIX AND EIGHT PIN OPTICALLY COUPLED ISOLATORS ARE DIRECTLY

PIN COMPATIBLE WHEN TWO EXTRA HOLES AND SHORT VCC TRACE
ARE ADDED TO PRINTED CIRCUIT BOARD.

MOUNTED ON .1 INCH CENTERS, SIX !C)INOPTO ISOLATORS

THE SAME SPACE AS EIGHT PIN OPTO ISOLATORS.
Figure 3.3.3-1

Pinout Comp"tibility
and 8 Pin 1501 Jtors.

and End Stackability

REQUIRE

of 6

Figure 3.3.4-1

Good Mechanical Layout for Opto Isolators in a

Line Receiver Application.

VCC
LOAD OF

.OlIlF

C.

ISOLATOR

GND

NOT RECOMMENDED
- CURRENT SPIKES
GENERATED BY SWITCHING OF LOAD
CAUSE NOISE INTO ISOLATOR BECAUSE
OF RESISTANCE AND INDUCTANCE
OF
VCCANDGROUNDTRACE&
VCC

VCC

VCC

VCC

VCC

ISOLATOR

VCC

BEST-ISOLATOR
IS CONNECTED BY
SEPARATE CONNECTIONS THAN
THOSE USED FOR COMMON VCC
AND GROUND TRACES.

LOAD OF

LOAD OF

ISOLATOR

GND

B.

VCC

VCC

+
.OlIlF

ISOLATOR

ISOLATOR
GND

GND

D.

.OlIlF

LOAD OF
ISOLATOR

ISOLATOR

VCC
GND

GOOD- LOAD BEING DRIVEN BY

ISOLATOR IS BETWEEN POWER
SUPPLY AND ISOLATOR.
Figure 3.3.5-1

Recommended Placement
for Optimum Results.

of Bypass Capacitor

GND

~
NOT RECOMMENDED
AS CONFIGURATION

- SAME REASON
C.

are illustrated in Figure 3.3.5-1. When several isolators are

used on a single printed circuit board, a separate byp,ss
capacitor should be connected across pins 5 and 8 of ea.;h
isolator. These same bypassing techniques also apply to the
5082-4365 isolator with the exception that VCC and
ground are on pins 15 and 10. When several isolators are
end stacked, a very clean mechanical layout can he
accomplished by connecting the pullup resistor to the V CC
of an adjacent isolator.
In noisy environments, bypassing may also be required br
the 5082-4350 series and for the 50824370 series isolatcrs
to prevent V C transients from being coupled into the
photodiode. Tc:ri.evoltage transients can be large enough to
cause multiple transitions of the output while the collectJr
is switching from one state to another. When extremely
high common mode transients are present, ~fficient energy
might be coupled into the photodiode and overstress Jr
cause a catastrophic failure of the detector. A resistJr
connected between VCC and pin 8 of the isolator wi1llirr jt
these current surges to safe levels but will not affect the
normal operation of the opto isolator. This technique is
illustrated in Figure 3.6.2-4.

It is an established fact that the total photon flux emittt:d

by an optoelectronic device diminishes slightly over the
operating lifetime of the device. Barring catastrophic
failures or overstressing of the optoelectronic device, this
change of photon emission is almost imperceptible for
many tens of thousands of hours in visual applications.
However, this change in light output can be measured Wdl
a sensitive photodetector. Figure 304.1-1 shows the average
change in light output vs. time for a 655 nm GaAsP lamp
manufactured
by Hewlett-Packard.
These lamps were
stressed at 50 mA DC for 40,000 hours. At lower stress
currents, the change of light output vs. time is reduced.
Since the total flux emitted by the LED normal y
diminishes over time, this change is often referred to as a
degradation of light output, although in some instances, the
light output of an LED has actually increased over time.
An optically
coupled isolator is an optoelectronic
emitter-detector pair. Any degradation of light output of
the emitter will cause a change in the apparent gain of the
entire device. Current gain of an isolator is normally
specified as a ratio of output current to input current
expressed as a percent for a specified input current and
called the current transfer ratio (CTR). The change in gain
of the isolator can be expressed as a change in current
transfer ratio over time and is commonly called CT:~
degradation.

CTR degradation is important because an excessive amount

of degradation or a marginally designed system can cause a
reduction in performance and eventual system failure unless
an allowance is made for it. Figure 304.1-2 shows a simple
optically isolated logic interface that can be used to
illustrate this problem. The available output current, 1 is
0
equal to the input current, IF' times the CTR of the
isolator. Keeping in mind the 10L capabilities of the logic
family driving the isolator LED, IF is picked so that the
resulting 10 will be large enough to sink the IlL of the logic
gate being driven by the isolator and the current flowing
through the pull-up resistor.

where: m is the number of logic inputs being driven by the

isolator, IlL is the maximum input current required by the
logic gate, VCC2 is the power supply voltage, VOL is the
maximum output voltage for the logic gate being driven by
the isolator, and RL is the pull-up resistor.
For example, if V r.C2= 5V, R::L= 5.6K, IlL = -1.6 mA,
VOL = o4V and m = 1, then 10 should be greater than or
equal to 2.4 mA. This 10 corresponds to an isolator driven
at 16 mA IF with a CTR> 15%. A reduction ofCTR over
time that lowers the CTR to less than 15% will reduce 1
0
below 204 mA and the output voltage of the isolator will
increase above o4V. The noise margin of the circuit will
then degrade until the output voltage rises higher than the
switching threshold of the gate (about 2 Vbe for TTL
families) and the circuit ceases to function. Clearly, the
designer must anticipate this CTR degradation and either
start with a CTR higher than 15% or design the circuit such
that a lower CTR will still allow the circuit to function.
Varia tions in VCC and resistor tolerances can also
contribute to system failures. A well designed circuit using
optically coupled isolators should allow some margin for
CTR degradation as well as consider the worst cased effects
of temperature, component tolerances, and power supply
variations. Also when using isolators in analog circuits,
degradation may cause a change in the gain or offset of the
overall system.

An optically coupled isolator can be modeled by the system

block diagram shown in Figure 3.4.2-1. Potential causes of
CTR degradation are a reduction in efficiency (T}) of the
emitter, a decrease in the transmission of the optical path
(K), a reduction in responsivity (R) of the photodetector,
or a change in gain (m of the output amplifier. It is
generally
accepted that CTR degradation is caused
primarily by the reduction in total efficiency (T}) of the

optoelectronic emitter. Also, since the gain of the output

amplifier
({3) is related to its input current, CTR
degradation may be compounded by the change in {3due to
a decrease in photo current (Ip) caused by a reduction in 1/.
The current transfer relationship for the isolator shown in
Figure 3.4.2-1 is:

where: 10 is the collector current, IF is the input current

1/ is the quantum efficiency of the LED expressed as
photons emitted per electron of input current, K is the
total transmission of the optical path, R is the responsivity
of the photodetector in terms of electrons of photocurrent
per photon, Ip is the photocurrent, and {3is the gain of the
output amplifier.
If the gain of the output amplifier {3is constant, than the
CTR degradation of the isolator is equal to the degradation
of light output of the optoelectronic emitter. In general,
the gain of the output amplifier is a function of its input
current so the rate of CTR degradation may be different
than the rate of LED degradation. Assuming that K, R, and
IF remain constant, and that (3 varies as a function of
photo current, then the change in CTR over time can be
related to the change in emitter efficiency by equation
3.4.2-2.
ACTR

1/
_Alo = -A1/ [ 1+(1+-)

CTR
1p(final)

10
= (1

1/

A1/

~J

+ A 1/ ) 1P (initial)
1/

where: /:, is defined as the final value minus the initial

value - Le., f::,{3 = (3(Ip(final)) -- (3(Ip(initial)); CTR, 10, 1/,
{3are the initial values; and /:,1//1/is the normalized change
in light output for a specified operating time.
Typical values of Ip (IF) and (3(Ip) for the 5082-4350 and
5082-4370 series of optically coupled isolators are shown in
Figures
3.4.2-2
and 3.4.2-3. The amount of CTR
degradation can then be calculated from Equation 3.4.2-2
by knowing the amount of LED degradation and the initi
value of IF' For example, suppose a 5082-4350 isolator i~
operated at 16 mA IF and the LED degrades by 15% sc
that /:,1//1/ = -.15.' Referring to Figure 3.4.2-2, Ip is
typically 24 IlA so that {3is typically about 145. SinCE
/:,1//1/= -.15, then Ip (final) is 20llA which corresponds to
a {3of 150. Thus f::,{3/{3 = 5/145 so that /:,CTR/CTR = -.12.
In general, I/:,CTR/CTRI < 1/:'1//1/1if f::,{3/{3 > 0 ane
I/:,CTR/CTRI > 1/:'1//1/if f::,{3/{3 < O. However, unless f::,{3/{3 is
large, the rate of CTR degradation will be approximately
equal to the rate of emitter degradation. The effect of thl
extra stage of gain in the 5082-4370 series of isolators is to

increase the sensitivity of the detector to much smaller

values of IF' This allows the 5082-4370 type of isolator to
be used at much lower input currents, thus reducing the
rate of emitter degradation. In addition, f::,{3/{3 > 0 in this
operating region which tends to further minimize the
effects of emitter degradation.
Another type of isolator contains an output amplifier that
switches rather abruptly at a certain threshold of Ip. An
example of this type of isolator is the 5082-4360 series
isolator. Typical Ip vs. 10 characteristics for the 5082-4360
series isolator are shown in Figure 3.4.2-4. Above this
threshold, the output of the isolator has more than
adequate gain to handle most types of digital circuits. The
exact value of the switching threshold, Ipth varies with
1/(IF)K R and somewhat on the processing of the
photodetector IC. The effect of emitter degradation for this
type of isolator is an increase in the amount of input
current that corresponds to IPth. The design objective for
this type of isolator is to insure that there is sufficient input
current, IF' to generate a value of Ip greater than IPth even
with some emitter degradation.
Characterization of an/at will then allow the effects of
emitter degradation on overall CTR degradation to be
determined for the various types of optically coupled
isolators.
Hewlett Packard has tested a considerable number of 700
nm LED emitters packaged in glass topped TO-18 style
hermetic packages for d1//dt at 20 mA IF' These emitters
are used in all of the opto isolators manufactured by
Hewlett Packard. Data was taken for both 25C and 125C
so a direct comparison can be made. Figure 3.4.2-5 shows
the mean, 10 and 90 percentiles of normalized efficiency
vs. time for LED's from several different wafer and
diffusion lots. These curves indicate that a decrease in
intensity of 5% at 1000 hours and 8% in 5000 hours is to
be expected when operating at 20 mA IF and 25C.
Figure 3.4.2-5 shows that operation of 125C does not
significantly alter the amount of emitter degradation over
time. Temperature is, therefore, only a minor contributor
to CTR degradation. Longer term degradation effects can
be gleaned from the curve presented earlier for 40,000
hours of operation for the very similar 655nm GaAsP
devices.
It is known that the rate of LED light output degradation is
influenced by the materials and processing parameters used
to manufacture
the GaAsP LED, and the junction
temperature of the LED in addition to the current density
through the LED. The current flowing through the LED
can be considered to be the sum of a "radiative" diffusion
current
and several components
of "non radiative"
currents. Light is generated only by the "radiative"

- ~ --

--- ---

>
....

--

IT A - 25"C

t-

..........

(i5

'

w
....

...

.8

(/)

~
1"

TA = 60C

::J

0
Z

::J
...J
c

.6

N
...J

<l:

~
cc

.4

1,000
STRESS TIME (HOURS)
Figure 3.4.1-1

Normalized Luminous Intensity vs. Operating Life

For Hermetic Ei55 nm Lamp (5082-4420) Stressed
at 50 mADC.

10

VOL
I

- _(C_T_R)_(I ..
F)
100

0-

IF 10 ~ IR - IlL' THEN OUTPUT

Typical Digital La lic Interface

Coupled Isolator.

Using An Optically

TRANSISTOR
SATURATION

WILL BE IN

Ie

INP UT

IF

TOTAL LED
EFFICIENCY

CURREN T

Ie

TRANSMISSION
OF OPTICAL
INTERFACE

1/ (IF,tl

Ie

RESPONSIVITY
OF PHOTODETECTOR
R

1/

/
1/

I
/

Photocurrent vs. Input Diode Forward Current for

Hewlett Packard Plastic Encapsulated Opto
Isolators.

Ip ~

GAIN OF
OUTPUT
AMPLIFIER

10

OUTPUT

C URRENT

/l(lp)

diffusion current. If the "non radiative" current should

increase, then less light will be emitted by the LED. The
ratio of "radiative" to "non radiative" currents changes as a
function of total current flowing through the LED. This
accounts for the supralinearity of light emission vs. input
current. Another important consideration is the magnitude
of the LED "stress" current as compared to the current at
which
the eTR
is measured.
The proportion
of
"non-radiative"
diode current to "radiative" diffusion
current tends to increase at lower measurement currents.
For example, given a stress current of 10 mA and
measurements of eTR at 1 mA and 10 mA before and after
stress, the percentage of eTR change will be greater for the
1 mA measurement than for the 10 mA measurement.
For this reason, excessive peak transient LED currents
should be avoided. Since the devices in Figure 3.4.2-5 were
stressed at 20 mA De and tested at 10 mA DC, the results
shown reflect a somewhat worst case rate of degradation.
At this time, the cause for this increase in "non radiative"
current over time is not completely understood. Figure
3.4.2-6 shows the change in eTR vs. time for 50824370
type couplers. Note the Significant difference between the
20 mA stress and 1.6 mA stess current curves -- both had
eTR measurements made at IF = 1.6 mA.
Typical
results
of degradation
at 2K hours plus
extrapolated 10K hour data is shown in Table 3.4.2-1.
While no completely valid model exists for extrapolation,
results to date suggest that a continuous exponential decay
model is probably valid. The 10K extrapolated data is based
on this assumption. Results of the 40,000 hour operating
life test shown in Figure 3.4.1-1 on the very similar 655 nm
GaAsP LEDs tend to substantiate this "exponential decay
model".

~ ..

/'

I
w

..~

200

/
V

(I)

It)
M

"t
N
a:

,\

100

""-

l..,...oo'

!.~

...

(I)

.,~ .".

.............

i-""

\
\

",

"t
N
co

It)

'"

"t\.V
1\,

...

.......
~
1
Ip . PHOTOC:URRENT -IlA

r--

~CE=4V

\.

5082-4370 V

a:
0

......

(I)

[\

100-

20,000
~E~5.0V

--...

......\

.., ..~ ~

...
~ V ....

50824350 \.

0
u.

\
\

co

-"-

VCE = .4V

It)

"r\

(I)

a:

,,-

I
VCE = 5.0V

10

a:

"' ....
~

--

......

0
u.
10,000

~
""-

r--- ..

'-

DC Current Gain for the Output Amplifier of the

5082-4350 Seril!S and 5082-4370 Series Optolsolators.
Isolators.

DEVICE TYPE
50824350
50824360
5082-4365
5082-4370

IF STRESS
TA =MAX
5mA
20mA
10mA
20mA
1.6mA

2K HOURS
-7%t.CTR
-10%t.CTR
-10% t.IF
-10% t.IF
-5%t.CTR

10K HOURS
-12%t.CTR
-15%t.CTR
-15%t.11=
-15% t.11=
-10%t.CTR

TABLE 3.4.2-1 Typical CTR Degradation vs. Time for

Hewlett Packard Opto Isolators.

In a critical application, the user may want to multiply

these "typical" figures by a factor of 2 or 3. The spread
between mean, 10 and 90 percentile in Figure 3.4.2-5
shows this to be a reasonable assumption.
When using this CTR degradation information, three things
should be kept in mind. This data is for continuous
operation of the LED at maximum LED current. In most
applications, the input LED is operated in a pulsed mode or
in a mode in which it is on for only a fraction of the total
system operating time. Since CTR degradation occurs only
during the time that the LED is on, the required operating
life of the isolator may only be a small fraction of the
projected system operating lifetime. Thus the effects of
CTR degradation can be minimized by phasing digital logic
such that the isolator LED is normally off. Lower stress
currents will also decrease the rate of degradation. Finally,
this data only applies to the 700 nm emitters manufactured

by Hewlett Packard. Isolator emitters manufactured at

different wavelengths by different GaAsP or GaAs processes
may have significantly different rates of degradation.
3.4.3

Worst Case Circuit

Degration

Design to

Allow

for

CTR

Proper circuit design can alleviate most of the potential

problems of CTR degradation. For the circuit of Figure
3.4.1-2, the minimum value of RL was selected by equation
3.4.1-1. This equation can be modified to allow a
guardband for CTR degradation and to worst case VCC l'
VCC2' RIN, and RL variations. Every circuit design should
account for these variations plus the effects of ambient
temperature extremes on device parameters. The minimum
value of RL is determined by the minimum value of 1
0
resulting from a minimum CTR device at a given IF' In this
state, the isolator must guarantee a logical "0" to the input
of the logic circuitry. The maximum value of R is
L
determined by the maximum amount of leakage, 10H' into
the collector of the isolator at the maximum ambient
temperature of the circuit. In this state, the isolator must
guarantee a logical "I" to the input of the logic circuitry.
RL should be selected to satisfy these constraints:

Rl(MIN)~

VCC2 (MAX) - VOL

_.------------IF (MIN) CTR (MIN) (1-x/100)
100

-m Illll

(MAX) ~

VCC2(MIN)

Within these limits for RV a tradeoff exists between RL

and propagation
delay. tpLH is determined by the
saturation delay of the isolator plus an RC constant
determined by RL and the total shunt capacitance, Cv at
the collector. tpHL is inversely proportional to IF and
generally independent of RV Figure 3.4.3-1a shows a worst
case TTL design that uses equations 3.4.3-1 and 3.4.3-2
to allow for CTR degradation. In this example, several
design choices are available to the designer and one or a
combination of them an optimize this design for cost and
effectiveness.

- V_OH

IOH(MAX) +

IIH

where: VOV VOH' IlL> IIH are guaranteed minimums and

maximums for the particular logic family, m is the total
number oflogic inputs connected to the isolator. CTR, IOH
are guaranteed isolator parameters. VCC2 and IF are circuit
parameters based on worst cased values for VCC 1 and R .
IN
x is the maximum % degradation allowed over the system
operating life.

1--

.-

"

E 10
I

IZ

a:
a:
::>

a:

W
...J
...J

u
I

,/

I)

1.0
Ip - PHOTOCURRENT -IJ.A
Figure 3.4.2-4

Output Curn nt vs. Photocurrent

Series Opto I iolators.

for 5082-4360

1.0

~-I!f

-- I-...

I-

:>
<>.

I-

:>
0

I-

- - ,.
I"-

.9

.. t--

I"",

90 PERCENTILE

----- ._-- ~-~..-- ""-

1-_-

:I:

...

-- --

(:l

I-

~
MEAN

--- -

--

--

10 PERCENTILE

1-

C
LIJ

:i .8

:!:

-----

a:

TA = 25C
I
T A = 125"C

.7

Normali2ed Luminous Intensity vs. Operational

Life for ~/OO nm LED Emitters Used for HewlettPackard Optically Coupled Isolators.

o
~
a:

1.1

~;;:

1.0

3.

a:u.

A TTL buffer gate can be substituted

to increase IF slightly.

for the 7405

VlE

:i:~
~;:
..:

...

.9

ZC

Ww
a: a:
a:::>
::>Vl

.8

IF STRESS = 20 mA,
10 60 mA. V

u..:
@~
N:;

.7

~
a:

.6

cc = 5.0V

o
z

Figure 3.4.26

I.

Normalized Current Transfer Ratio vs. Operating

Life for 50824370 Series Opto Isolators.

Vee I' Vee2 tolerances can be tightened (possible

addition of zener diode).

A simple and cost effective design technique that reduces

the effects of eTR degradation and improves the overall
performance of the system is to replace the gate being
driven by the isolator with a gate having a lower IlL
requirement. For example, Figure 3.4.3-1 b shows a TTL to
LSTTL worst case design. This circuit is identical to Figure
3.4.3-la with the exception that the 7404 gate is replaced
with
a 74LS04 gate. Since TTL and LSTTL have
approximately the same propagation delay, the circuit of
Figure 3.4.3-lb will have a significantly shorter tpLH than
the circuit of Figure 3.4.3-la because R is smaller. Figure
L
3.4.3-lb can be optimized to allow more than one load, a
larger amount of eTR degradation, or a lower eTR isolator
(5082-4350) at lower cost.

*7404 - FIGURE 3.4.3-1a

VCC1' VCC2 = 5.0V:!S%

*74LS04 - FIGURE 3.4.3-1b

RIN, R ARE STANDARD

L
5% TOLERANCE RESISTORS

TTL FAMILY

IlL

74S

-2mA

74H

-2mA

74

-1.6 mA

74LS

-.36 mA

74L

-.18 mA

IIH

VIH

IOL

VOL

IOH

.8V

50J.LA

2V

20mA

50J.LA

2V

20mA

.5V
.4V

-1000 J.LA

.8V

500J.LA

2.7V
2.4V

.8V

4OJ.LA

2V

16mA

.4V

400J.LA

2.4V

.8V

20J.LA

2V

8mA

400 J.LA

.7V

10 J.LA

2V

3.6mA

.5V
.4V

200J.LA

2.7V
2.4V

IL

OH

5082-4351

FIGURE 3.4.3-1a (TTL TO TTL INTERFACE)

VCC1 - VF - VOL

RIN = ---------

IF

= 194Q

= 1.5V

IF = 16 mA

SELECT 180n:!S%

RIN SELECTED AS 180n:!S%

:. IF (MIN) = 13.8 mA

R ;;;'30.41<,EON. 3.4.3-1
L
R ~58.0k, EON. 3.4.3-2
L
ASSUME THAT AN ISOLATOR WITH CTR ;;;'19%

R ;;;'3.41<,EON 3.4.3-1
L
R

;;;'7.7k, EON. 3.4.3-1

~58.Ok. EON 3.4.3-2

Figure 3.4.3-1

(MAX) = 19.6 mA

I F x ~ 15%, CTR ;;;'1 5% AT 16 mA IF

IF x~15%

IOH ~.5J.LA AT VCC = 5.5V, 25C

:.I

VCC(MAX)-VF(MIN)-VOL
(MAX) = ----------R
(MIN)
IN
= 19.6 mA

CAN BE SPECIFIED.

CTR;;;' 15% AT 16 mA IF' Vo

_ VCC (MIN) - VF (MAX) - VOL

.. IF (MIN) - ----------R
(MAX)
1N
= 13.8 mA
.
.. I

VF = 1.5V (1.75V MAX) AT 20 mA IF

~ 1Oak, EON 3.4.3-2

;;;'4.01<,EON 3.4.3-1

R ~ 100k, EON 3.4.3-1

L

Worst CaseD3signof TTL to TTL/LSTTL

Interface Using 5082-4351 Isolator that Allows
for CTR Degladation.

.4V

Some optically coupled isolators such as the 5082-4360

family have a highly non linear current transfer functiorr
which is shown in Figure 3.4.3-2. This transfer function carr
be described more accurately by a minimum threshold of
IFth that is required to turn the isolator on rather than a
current transfer ratio. As the light output of the LED
degrades, the net effect is a slight increase in the value of
IFth. Allowance for a change in IFth due to LED
degradation can best be accomplished by driving the
isolator with a somewhat higher IF than the minimum IF
required for a guaranteed VOL (at a specified maximum
IOL)' The calculation of the required IF to allow LED
degradation of X% is shown below:
Ip

0::

(IF)n

Vcc
TA

>

5.0V

= 25"C

~
>..J

o
>
>::J

~
::J
o
I

>::J

:f'

(3.4.3-3)

IFH
X
[1-

'iOO

1/n

where: IFH is the guaranteed minimum isolator inpl,t

current for a specified IOL and VOL' n is the slope of Ip v;.
IF on logarithmic coordinates (Figure 3.4.2-2), and x is the
maximum % degradation allowed over the system operating
life.
Figure 3.4.3-3 shows a worst case LSTTL to TTL desigrr
using equation 3.4.3-5 to allow for CTR degradation. Since
IOL of this type of optically coupled isolator does net
decrease with operating life, the equations for RL can be
simplified to:

Vcc
TA

5.0V

= 25"C

>
I

'>--'

<t
..J

>::J

~
::J

VCC2 (MAX) - VOL

o
>

RL ~ IOL (MIN) - m IIILI

4
IF -INPUT

10

DIODE FORWARD CURRENT - mA

VCC2 (MIN) - VOH

,;;;;

IOH (MAX) + mliH

where: VOL' VOH' VIL' IIH are guaranteed minimums and

maximums for the particular logic family, m is the total
number oflogic inputs connected to the isolator. IOUMI};)
and IOH(MAX) are guaranteed isolator parameters. V CC2
is the power supply voltage.
Since it is very important to insure that a guardband exists
for IF and some applications, i.e., line receiver applications
have a wide range of input currents, it may be necessary tel
"regulate" the isolator input current. One circuit for doing
this is shown in Figure 3.6.1-6 and Figure 3.6.1-7.

11
GND

Figure 3.4.3-2

10.

Input vs. Output Characteristics for 5082-4360

Series Opto Isolators.

+VCC1 = 5V

.01
/.If

VCC ' VCC = 5.00V 5%

1
R1N' RL A~ STANDARD 5%
TOLERANCE RESISTORS
5082-4360

(LSTTL TO TTL INTERFACE)

VF = 1.5V (1.75V MAX) AT 10 mA IF

I FH ~ 5 mA TO GUARANTEE
R

~ VCC1 (min) - VF(max) - VOL

IN

VOL ~ .6V AT IOL = 13 mA

IF
IOH ~ 250J,.lA
~ 375n, SELECT 360n 5%

x~ 25%

DUE TO SUPRALINEARITY OF LED FOR WORST CASE VALUES

OF RIN' VCC1' and VF; IF n = 1.2, THEN x =29%
. I (
) = VCC(max) - VF(min) - VOL
.. F max
RL(min)

RL ~ 11 k,m= 1
:. RL SELECTED AS 510n 5%,m = 1

Figure 3.4.3.3

Worst Case Design of LSTTL to TTL Interface

Using 5082-43liO Isolator that Allows for CTR
Degradation.

The minimum IF should be set approximately 30% above

the minimum guaranteed IFH for the isolator (5 mA for tte
5082-4360, 5082-4364 and 10 mA for the 5082-4365) or
as calculated in equation 3.4.3-5. The regulator will insure
that any excess current is shunted through the externll
transistor. This has the added benefit of protecting the LE D
against harmful transients.
Isolators such as the 5082-4360 family require a bypa';s
capacitor across V CC and Gnd of each isolator (see section
3.3.5). If this bypass capacitor is mistakenly omitted or is
located too far away from the isolator, the isolator
amplifier can oscillate internally. The symptoms of the~,e
internal oscillations are the same as those of LED
degradation of this type of isolator -- an increase in tt e
switching threshold. However, this type of "degradation"
can be cured by proper bypassing.

performance of Figures 3.4.3-1a, 3.4.3-1b, or 3.4.3-3,

although an external resistor connected across pins 5 and 7
of the isolator will improve the switching performance of
Figures 3.4.4-1 a and 3.4.4-1 b (see section 3.6.3 and Figure
3.6.3-5).
The analog amplifier explained in section 3.5.4 is an
example of "degradation-proof'
circuit that uses optical
feedback to reduce the effects of CTR degradation. The
circuit uses two isolators connected so that one isolator is
forced to track the input current of a second isolator by
servo action. The transfer function of this circuit is given by
equation 3.5.4-1 and is reproduced below in simplified

fO'~2. [njR K
C

/n

1 ,] (lF1) n1 2

(3.4.4-1)

R3K2

where: each isolator can be described by equation 3.4.4-2:

3.4.4

CTR "Degradation
Techniques

- Proof" Isolator Circuit Design

Some optically coupled isolator applications require overall

system performance for several tens of thousands of hour;.
In designs of this type, a worst case circuit design such HS
shown in section 3.4.3 may not be adequate. No one knows
exactly how much an LED will degrade in 10,000 hours, so
how much CTR degradation should the design allow? The
circuits illustrated in Figures 3.4.3-1 and 3.4.3-2 can not be
designed for x = 80% because the isolator specified has
insufficient initial CTR. Clearly an isolator that has
degraded by 100% is a catastrophic failure but an isolatcr
that has degraded by 99% is still marginally functional. For
the purposes of this discussion, a "degradation-proof'
design is one that will allow the isolator or isolators t,)
degrade by 80% and still guarantee proper logic or analo~
signals to the output circuit. "Degradation-proof'
circuits
use either a very high CTR device or optical feedback that
reduces the effects of CTR degradation.
One example of the first type of "degradation-proof'
circuit is the use of an isolator with a high CTR in a circuit
that was worst case designed to use a low CTR device. Theil
the high gain isolator can degrade until it reaches the
minimum allowed CTR of the design. An example of thi;
technique is shown in Figure 3.4.4-1a. In this example, a
high gain isolator is simply retrofitted into a design for a
low gain isolator. A worst case design that is optimized fo:'
a high gain isolator is shown in Figure 3.4.4-1b. Figure
3.4.4-1b is a better design than Figure 3.4.4-1a because in
3.4.4-1b, the isolator is operated at a lower IF so its rate of
degradation has been decreased. Especially in designs of thi~,
type, the designer should insure that the leakage current,
IOH' of the isolator be considerably lower than the
"desired" output current, IOV if a low current application
is being considered. Neither circuit has the switching

10

K[2-J

IF'

where: 10 is the collector current of the output transistor

when biased in the active region, IF is the input LED
current, IF' is the input LED current where K is measured,
K is the output current when IF = IF" and n is the slope of
10 vs. IF on logarithmic coordinates. Then C in equation
3.4.4-1 is equal to
~F2'J

1 [OF1') n1/n2

(3.4.4-3)

This model applies particularly to the 5082-4350 type

isolator, but can also be extended to the 5082-4370 series
of isolators. Section 3.5.2 covers this model in more detail.
At some value of IF l' ~ and R3 can be adjusted so that
IF2 = IF1 Then as IF1 is varied, IF2 will remain
approximately equal to IF1 if n1 ::>;: n2' As the isolators
degrade, K is reduced. If the isolators degrade at the same
rate, then IF2 will remain approximately equal to IF1.
Since nl and n2 are greater than one, the change in IF1/IF2
will be less than the change in K1/K2. For instance, if
isolator 1 degrades by 80% and isolator2 does not degrade at
all, (R2K1)/(R3K2) =.2 but ifn2 = 1.8, then IFI/IF2 =.41
and the apparent effects of CTR degradation will be
reduced. This servo feedback technique can also be applied
to digital circuits.
3.5

Analog Applications of Optically Coupled

Isolators

Optically coupled isolators are useful for applications where

analog or DC signals need to be transferred between two
isolated systems in the presence of a large potential

*74LS04 - Figure 3.4.4-1 a

*7404 - Figure 3.4.4]b
VCC1' VCC2 = 5.0V 15%
RIN, RL ARE STANDARD 5% TOLERANCE RESISTORS
5082-4370
V

= 1.4V (1.7V MAX) at 1.6 mA IF

CTR;;'

FIGURE 3.4.41b (TTL TO TTL INTERFACE)

300% at 1.6 mA IF' V 0 = AV

IOH ";;250 1JAat V

CC

I :.

= 7V, 70C

R ";;S.1k, EON. 3.4.3-2

L
RL SELECTED AS 6.Sk 5%

(TTL to LSTTL INTERFACE)

FROM EON. 3.4.3-1:

x";;100

100 m IIILI
1-------IF(MIN) CTR (MIN)
~

100 (VCC2 (MAX) - VOL)

]
------------RL (MIN) IF (MIN) CTR (MINI

VCC2(MAX)
- VOL
lOUT (MIN);;' [ .
R (MIN)
L

+ n IIILI

~ /[

x
]
11- )
100

;;. 11.S mA, x = SO%

ASSUM E CTR ;;. 300%, 1.6 mA ,,;;IF";; 5.0 mA

:. x ";;26% FOR DESIGN SHOWN IN FIG. 3.4.3-1b

(50S2-4351:

IN

= 1S0n, R

:. x ";;96% FOR SAME DESIGN AS ABOVE

BUT REPLACING THE ISOLATOR WITH A
5082-4370, ASSUMI NG CTR

> 300% AT

:.I

= 4.3k)

13.S mA IF

(MIN);;'

11.S ;;. 3.9 mA

VCC1 (MIN) - VF (MAX) - VOL

R
(MAX) ,,;;-----------IN
IF (MIN)
";;680n
:. RIN SELECTED AS 620n5%

Two Worst CaseDesignsof TTL Interfaces that

are Optimized f Jr Maximum Operating Life.

difference or induced noise. They are an inexpensive way to

eliminate the shock hazard between an input transducer
and an output circuit. Optically coupled isolators can also
reduce the common mode noise generated between an
isolated input circuit and an output circuit. Potential
applications include those in which large transformers,
expensive instrumentation amplifiers, or complicated AID
conversion schemes have been used. Examples include
sensing circuits (thermocouples, transducers ... ), patient

monitoring
equipment, power supply feedback, high
voltage current monitoring, and audio or video amplifiers.
In many applications, the isolator can transmit the analog
signal directly. However, in applications where very high
linearity and stability are critical, the analog signal can be
converted into a digital form and then isolated. Overall
circuit parameters like linearity bandwidth, and stability
determine which approach is best.

./

For an optically coupled isolator with a photodiode

detector, or with an integrated photodiode and transistor
detector such as the 50824350 family, the output current
for a wide range of input currents can be expressed as:

"

/
/

where IF is the input LED current, IF~ is the input LED

current where K is measured, K is the output current whtm
IF = IF" and n is the slope of 10 vs. IF on logarithmic
coordinates. For a photo diode opto isolator, 1 is the
0
current flowing into the cathode of the zero or revene
biased photodiode.
For an integrated photodiode and
transistor isolator, 10 is the collector current of the output
transistor when biased in the active mode.

<t

.2

>2

w
~
::>

.1

OJ

.08

:J

.06

>-

~
::>

o .04

S'

IJ

1/

If n is equal to one, then the input to output trans[.~r

function of the opto isolator is linear. For most opticaLy
coupled isolators, n is not equal to one. For the 50824350
isolator, n varies from approximately 2 at very low inputs
currents to ~ ~r less at higher input currents. Typical 1 vs.
0
IF charactenstIcs for the 50824350 type opto isolator a:'e
shown in Figure 3.5.2-1.

,,
.001
.1

.2

Figure 3.5.2-1

There are several analog techniques that use optically

coupled isolators to isolate an analog signal. The Servo and
Differential techniques are dc coupled isolated amplifiers
that use two opto isolators for improved linearity and
temperature stability. The servo linearizer forces the input
current of one optically coupled isolator to track the input
current of a second opto isolator by servo action.[f
nl ~ n2 over the excursion range of the input current of
the isolators, then the non linearities will cancel and the
overall transfer function will be linear. The differential
linearizer causes the input current of one opto isolator to
increase in response to an input signal while the input
current of the second optically coupled isolator decreases
by an equal amount. If n} ~ n2 over the excursion range of
the input currents of the isolators, then a gain increment in
the first opto isolator will be approximately balanced by a
gain decrement in the second opto isolator and the overall
transfer
function
will be linear. For AC coupled
applications, reasonable linearity can be obtained with a
single optically coupled isolator. The opto isolator is biased
at higher levels of input LED current where the ratio cf
incremental
photodiode
current to incremental LED
current (alp/aIF)
is more nearly consfant. For the
5082-4350 series of opto isolators, this occurs at input
currents greater than 15 mA, as shown by Figure 3.5.3-1.
The input current of the optically coupled isolator is
modulated by the input signal and the modulation of the
photo diode current is detected and amplified. The linearity

.4.6.8

1
2
'F - INPUT CURRENT

4
6
- mA

8 10

20

50

Output Current vs. Input Current for the

5082-4350,4351,4352,4354,
and 4355
Optically Coupled Isolators.

of this type of circuit is determined by the amount by

which incremental CTR (alp/aIF) changes over the range
over which the input LED current is modulated. For many
applications, at least one of these three techniques can be
used.
Table
3.5 .3-} compares the advantages and
disadvantages of each of these analog techniques.

The servo isolation amplifier shown in Figure 3.5.4-1

operates on the principle that two optically coupled
isolators will track each other if their gain changes by the
same amount over some operating region. U2 compares the
outputs of each opto isolator and forces IF2 through the
second isolator to be proportional to IF} through the first
isolator. The constant current sources bias each input LED
at 3 mA quiescent current. R} has been selected so that IF}
varies from 2 mA to 4 mA as V IN varies from -5V to +5V.
R} can be selected to accomodate any desired input range.
The zero adjustment potentiometer
must have sufficient
dynamic range to compensate for a worst case spread of
opto isolator current transfer ratios (the values of K/IF' x
100%) at the input quiescent current of 3 mA. Then with
VIN at some value, R4 can be adjusted for a gain of one.
This potentiometer only requires sufficient dynamic range
to compensate for ICC} not being equal to to ICC2' If

1.0

,,- .----

.E-1.lL

'" I '"
a:

I-

.8

()

/'

..J

<t

I-

w
:iE
w
a:
()

.6

c
w

N
..J

<t
:iE

a:

""""",.--

/'

,/

.4

10
Figure 3.5.3-1

IF - INPUT CURRENT - mA

Normalized Incremental CTR vs. Input Current for 5082-4350,

20

4351, 4352 Optically Coupled Isolators.

SERVO

DIFFERENTIAL

AC COUPLED

Number of Opto Isolators per Channel:

2: 50B2-4350/1/2
OR
1: 50B2-4354/5

2: 5082-4350/1/2
OR
1: 50824354/5

1: 5082-4350/1/2

Overall Linearity:

.5%

1%

1%

Input Current Range:

2-4 mA

2-4mA

15-25 mA

Frequency Response:

DC

DC

AC

Bandwidth:

100 KHz

Temperature Stability - Offset:

.01%/C

.04%;oC

Not Applicable

Temperature Stability - Gain:

-.03%tC

-.4%;oC*

-.5%;oC*

Common Mode Rejection

>46 dl3 @ 1 KHz

>70 dB

1 MHz

10 MHz

1 KHz

>22 dB

1 MHz

*can be improved with additional thermistor

Table 3.5.3-1

Comparison of the Typical Characteristics of Servo, Differential, and ac Coupled Isolated Amplifiers.

3.33

1% linearity for 10V pop dynamic range

Unity voltage gain
25 KHz bandwidth (limited by U1, U2)
Gain drift: -.03%;0 C
Offset drift: 1 mV;oC
Common mode rejection: 46 dB at 1 KHz
500V dc insulation (3000V if 2 single isolators are used)
TABLE 3.5.4-1 Typical Performance for the Servo
linearized DC Amplifier.

VOUT

IC,

2kU GAIN ADJUST

K 'T[IF jn,
F,

3.5.5

I
1 is equal to ICC2' R4 can be replaced with a single
CC
resistor equal to Rl. The transfer function of the servo
isolation amplifier is:

ft

n1

K1R2(1CC1)
IF '2) K R (IF '1)n1
)
4L
2 3
~

1/n

(3.5.4-1)

VOUT = R

(1 +

1
\ n /n2

:I~

-ICC2]

1 CCl)

After zero adjustment, this transfer function reduces to:

(3.5.4-2)
VOUT = R41CC2 [( 1 + x)

IN
n1
1l where x = ---,
n= n:::
R1 ICC1
"-.

Using a binomial expansion, V OUT can be written as:

(3.5.4-31
n (n-1) x2
VOUT=R4ICC2

nx+

The differential isolation amplifier shown in Figure 3.5.5-1

operates on the principle that an operating region exists
where a gain increment in one optically coupled isolator
can be approximately balanced by a gain decrement in a
second opto isolator. As IFI increases in the first opto
isolator due to changes in VIN, IF2 in the second opto
isolator decreases by an equal amount. If nl "" n2' then the
gain increment caused by increases in IFI will be
approximately balanced by the gain decrement caused by
decreases in IF2. The constant current source biases each
input LED at 3 mA quiescent current. Rl and R2 are
selected so that IF varies from 2 mA to 4 mA as V IN varies
from -5V to +5V. Rl and R2 can be selected to
accomodate any desired input range. At the output, U3 and
U4 force VOUT to be proportional to the difference
between ICl and IC2'

2.

n(n-1)(n-2)
+

3._

x3+ ...

Differential Isolation Amplifier

VOUT = R5 [(R3/R4) IC1 -IC2l

(3.5.5-1)

The zero adjustment potentiomenter, R3, allows VOUT to

be set to zero when VIN is equal to zero. The gain
adjustment potentiometer, R5, allows the overall gain to be
set to one. R3 and R5 must have sufficient dynamic range
to compensate for a worst case spread of opto isolator
current transfer ratios (the values of K/IF, x 100%) at the
input quiescent current of 3 mA. The transfer function of
the differential isolation amplifier is:
(3.5.5-2)

The linearity error in the transfer function when nl =1= n:~

can be written as:
linearity error
desired signal

(1 + x)n - nx-1

VOUT"R5[(K::%:~;"f (1+ RV::cf

nx

For example, if Ixl ",;;.35, nl = 1.9 and n2 = 1.8, then the

error is 1% of the desired signal. Overall linearity can bll
improved by reducing Ixl or by matching the optically
coupled isolators for n. The current transfer ratio (K/IF, x
100%) has no effect on overall linearity .
While temperature
stability is also dependent on thll
stability of current sources and resistors, changes in current
transfer ratio (the value of K/IF" x 100%) of the optically
coupled isolators will have negligable effect on overall gain
and offset as long as the ratio of Kl to ~ remain;
constant. Typical characteristics
of the servo isolator
amplifier (Figure 3.5.4-1) are given in Table 3.5.4-1.

-K

(~\n2

(1 _ ~)n2

2IF'2)

if R =: R 1

R ICC

=: R2

After zero adjustment, this transfer function reduces to:

= R5K.l1(1 + VIN
OUT

where K' =

\n1

-(1

RICC)
K1 R3 (~\
R4

21F'1 )

V'N \

n~3].5'5-3)

RICC)
1
n = K2

(~)n2
21F'2

Using a binomial expansion, V OUT can be written as:

(3.5.5-4)
VOUT=

R5K'

(n1+n2)x+

[n1(n1-1)-n2(n2-1)]
--------

x2
+

2!

n1 (n1-1 )(n2-2)-n2(n2-1

)(n2-2)

x3

+ ...

3!
where x

VIN

RICC

If both
isolators have a square law response, i.e.,
n1 :::;n2 = 2, then all non linear terms will cancel and VOlD'
will be proportional to VIN. However, if n1 = n2' then aL
even order terms will cancel, and the total linearity error
can be reduced below 1%. The linearity error in the transfer
function for any values of n can be written as:
(3.5.5-5)
- (1 - x)n2 - (n1 + n2) x
(n1 + n2) x

Por example, if Ixl~.35, nl = 1.9 and n2 = 1.8, then the

linearity error is 1.5% of the desired signal. Overall linearity
can be improved by reducing Ixl or by matching the
optically coupled isolators for n. The current transfer ratio
has no effect on linearity.
While temperature
stability is also dependent on the
stability of current sources and resistors, changes in current
transfer ratio of the optically coupled isolators over
temperature will cause a change in gain of the circuit. This
change in gain can be compensated with a thermistor in
either the input or output circuit. Por example, if R5 is
replaced by a positive TC thermistor, or R1 and R2 are
replaced with negative TC thermistors, then the gain will
tend to increase as temperature goes up. Zero offset over
temperature will remain stable as long as the ratio of Kl to
K2 remains
constant. Typical characteristics of the
differential isolation amplifier (Pigure 3.5.5-1) are given in
Table 3.5.5-1.

3"10 linearity for 10V POpdynamic range

Unity voltage gain
25 kHz bandwidth (limited by U1, U2, U3, U4)
Gain drift: -.4%fC
Offset drift: 4 mVfC
Common mode rejection: 70 dB at 1 kHz
3000V DC insulation

Table 3.5.5-1
Typical Performance of the Differential
Amplifier

The AC coupled isolation amplifier shown in Pigure 3.5.6-1

operates on the principle that a single optically coupled
isolator can be biased in a region where incremental CTR
(alp/alp) is constant. Ql is biased by R1, R2 and R3 for a
quiescent collector current of 20 mA. R3 allows Ip to vary
from 15 mA to 25 mA for a IV peak-to-peak input signal.
Under these operating conditions, the 5082-4351 operates
in a region of almost constant incremental CTR as shown
by Pigure 3.5.3-1. The varying photon flux is detected by
the photodiode and amplified by Q2 and Q3' Q2 and Q3
form a cascade amplifier with feedback applied by R4 and
R6 13 is selected to allow Q3 to operate at its maximum
gain bandwidth product. R6 is selected as Vbe/13 and R7 is
selected to allow maximum excursions of VOUT without
clipping: R7 = (V CC - Vbe - Vce sat)/(213) R5 provides dc
bias for Q3' Closed loop gain of the output amplifier can be
adjusted with the gain adjustment potentiometer,
R4.
Linearity can be improved at the expense of signal to noise
ratio by reducing the excursions of Ip. This can be
accomplished by increasing R3 and adding a resistor from
the collector of Q1 to ground to obtain the desired
quiescent collector current of 20 mA. This circuit has no
feedback around the optically coupled isolator, so any
parameter that causes the incremental CTR (alp/alp) to
vary will cause a change in gain of the circuit. Since the
quantum
efficiency
of the input LED varies with
temperature, the alp/alp and thus the overall gain will also
vary with temperature. However, a thermistor can be used
in the output amplifier to compensate for this change in
gain. Por example, if R7 is replaced by a positive TC
thermistor, or R6 is replaced by a negative TC thermistor,
then the gain will tend to increase as temperature goes up.
The transfer function of the AC coupled isolation amplifier
is:
OUT
V
VIN

"",[a J
1p

a1F

J rl J
R

[_1
R3

4 R
R6

Typical
characteristics
of the AC coupled isolation
amplifier (Pigure 3.5.6-1) are given in Table 3.5.61:

2% linearity over 1V Popdynamic range

Unity voltage gain
10 MHz bandwidth
Gain drift: -.6%fC
Common mode rejection: 22 dB at 1 MHz
3000V DC insulation

Table 3.5.6-1
Linearized DC

Typical Performance of the Wide Bandwidth

AC Amplifier

OFFSET
ADJUST
'SkSl

___
Rl
2.7 kn

R2

1__

50 kn

Rs

2.7 kn

R7

330n
~r
2~F

The Servo, Differential, and AC coupled analog circuit:;

offer a simple and low cost means to optically isolate arl
analog signal. However, the smallest linearity error that car
be achieved with these techniques is limited to about .5% to
1%. Moreover, stability of gain and offset over temperaturE
may not be adequate
for some applications. FOl
applications that require higher linearity or better stability
than can be achieved with analog techniques, isolation can
be realized with digital techniques. A second reason for
using digital
isolation techniques is that in many
applications, the analog signal must be converted into a
digital format anyway before it can be processed by a
microprocessor or by a digital output device. Optically
coupled isolators can then provide isolation between the
analog input device and the digital output device and at the
same time provide all the advantages of digital isolation
techniques.

With digital isolation techniques, the analog signal is

converted into some type of digital code that can be
transmitted through an optically coupled isolator. At the
output, the digital code is converted back into analog form,
or used directly by digital processing circuitry. Since the
isolator is used only as a high speed switch, linearity,
stability and bandwidth are determined by the conversion
accuracy, temperature stability and conversion time of the
A to D and D to A converters. Selection of the proper opto
isolator is determined by such factors as maximum data
rate, input current limitations, and proper logic interface at
the output.

These techniques convert the analog signal into a digital

signal through an isolated interface but leave the digital
signal in a form that can be used by some type of digital
processor. The A to D converter technique and variable
pulse width monostable multivibrator technique illustrate
two ways in which this analog to digital conversion can be
accomplished. The circuits that are shown illustrate the use
of optically coupled isolators to provide isolation between
analog and digital circuitry and how high speed isolators
can reduce systems cost.
The most commonly used digital technique to isolate an
analog signal is the use of an A to D converter with
optically coupled isolators providing isolation of the digital
signals. This technique is outlined in Figure 3.5.7.1-1. The
digital information can be transmitted through the isolators
in either serial or parallel format, depending on the outputs
available on the A to D converter. Serial transmission is
especially practical for 8, 10 and 12 bit A to D converters
because it replaces one isolator per bit with one or two high

multivibrator is triggered by a negative going pulse at

TRIGGER, QO goes high and stays high for a time
proportional to the magnitude of V IN. The time during
which QO is high is measured and used to provide a digital
representation of VIN to the output circuit. QO can be used
to gate an oscillator into a counter. After QO goes low, the
contents of the counter can be displayed by an LED display
or read into a microprocessor. This concept is illustrated by
Figure 3.5.7.1-2a. The counter can also be implemented in
microprocessor software. With this approach, TRIGGER is
attached to one bit of an output port of the microprocessor
and QO is attached to one bit of an input port. For this
example, QO is attached to D7 to simplify the software
implementation.
With the program illustrated in Figure
3.5.7.l-2b,
register HL will contain a binary number
proportional to VIN after the conversion is completed. The
total
time,
t, that has elapsed will be equal to
(29)(HL)(clock
rate of microprocessor).
While this
approach uses less hardware than the first approach, the
microprocessor
must
wait until the conversion is
completed. Commercially available timers, such as the 555
timer, can be used to implement this technique at minimal
cost.

ISOLATED AID CONVERTER

PARALLEL DATA OUTPUTS
+VCC

+VCC

0,
O2

03
04
a:
w

l-

DS

06

a:

>
z

ANALOG

INPUT

07
08

3.5.7.2 Isolated Analog to Digital to Analog Techniques

ISOLATED AID CONVERTER
SERIAL DATA OUTPUTS
+VCC

SIPO

[',
['2
['3
['4

+VCC

[S
DATA
CLOCK

DATA
CLOCK

[8

a:
ANALOG
INPUT

[6
[']

w
I-

a:
w

>
z

END

o
u
c
~START

Figure 3.5.7.1-1

Opto Isolators Provide Isolation for AID Converters

High Speed Isolators can Reduce Cost by Transmitting Data in Serial Form.

speed isolators. If serial outputs are not available from the

A to D converter, parallel outputs can be converted into
serial format with a PISO shift register. The A to D
converter digital isolation technique is especially useful
where the digital information will be used as data for some
type of digital system such as a microprocessor or LED
display. A to D converters are available with straight binary,
offset binary, two's complement binary, or binary coded
decimal outputs.
A second technique that can be used to convert an analog
signal into digital form while maintaining electrical isolation
between the analog signal and the digital circuitry is the
circuit shown in Figure 3.5.7.1-2. Whenever the monostable

These techniques perform the same type of transfer

functions as the strictly analog servo, differential and AC
coupled circuits described earlier. The difference between
these techniques and the analog circuits discussed earlier is
that the optically coupled isolators are used in a digital
fashion. Thus, these circuits are more immune to CTR
degradation and some types of common mode transients.
These techniques are useful where some type of high
stability isolated instrumentation amplifier is required. The
pulse width modulator and the voltage to frequency
converter illustrate two ways in which this analog to digital
to analog conversion can be accomplished.
The circuit shown in Figure 3.5.7.2-1 shows a pulse width
modulated scheme to isolate an analog signal. The oscillator
operates at a fixed frequency, f, and provides a continuous
trigger to the monostable multivibrator. Once triggered, the
monostable
multivibrator
gives an output pulse
proportional
to the value of VIN. The output of the
monostable multivibrator is a square wave of frequency f
but with a duty factor proportional to VIN. The maximum
frequency of the oscillator is determined by the required
linearity of the circuit and the propagation delay of the
optically coupled isolator.
(tmax - tmin)(required linearity);;;' ItpLH-tpHL I
For example, if f is 1 MHz, and the monostable
multivibrator can vary the duty factor from 10% to 90%,
then (tmax-tmin) = 800 nS. If the isolator is a 5082-4360,
then ItpLH-tpHL I varies from about 10 nS to 60 nS
depending on drive currents, and temperature. Thus, the

'MICROPROCESSOR
ONE
SHOT

INTERRUPT:

SYSTEM

I
I
I

COUNTER

5082-4351
8
2
6

RIN
VIN

00

IN

CLOCK

BINARY
OR BCD
OUTPUTS
CONNECTED TO
} INPORT PART OF
MICROPROCESSOR

RESET

MONO
STABLE

RL

I
I

5082-4351
8

107

RIN

TRIG

INPUT
PORT

OUTPUT
PORT

I
I
I
I
I

Figure 3.5.7.1-2a

Analog Isolation Using Variable Pulse Width

Monostable Multivibrator.
0 Interrupts
Microprocessor when Converr.on is Completed,
I

I
I
5082-4351

IN

00

MONO
STABLE

8
6

INPUT
PORT
(m)

RL
00

5082-4351

RIN

OUTPUT
PORT

I
I

Same as 3.5.7.1-2a Except that Microprocessor

Increments a Register Until 00 Goes Low.

(n)

I
I

A-D CONVERSION
DI
LHLD, (OOOO)H
MVA, (80)H
OUT (n)
IN (m)
ORA,A
INX, H
JM, LOOP

ROUTINE

I
I

~'"flll

--L~ r--r--:---l
__ ..i
Lf--T

r----'
INTEGRATOR
(DEMODULATOR)

Figure 3.5.7.2-1 Pulse Width Modulation Techniques Can Be Used

to Isolate an Analog Signal.

I OPTIONAL I
I L. P. FILTER ~VOUT
L
..J

worst case linearity error would be <7.5%. At 100 KHz,

the worst case linearity error would be reduced to <.75%,
assuming that the monostab1e multivibrator and output
circuits had perfect linearity. At the output, the squarewave
would be integrated or demodulated such that V OUT
would
be proportional
to the duty factor of the
squarewave. An additional low pass ftlter would help to
reduce the ripple at the output frequency f.
Figure 3.5.7.2-2 shows a voltage to frequency conversion
scheme to isolate an analog signal. The voltage to frequency
converter gives an output frequency that is proportional to
VIN. The maximum frequency that can be transmitted
through the isolator is limited by the maximum data rate of
the isolator. At the output, the frequency is converted back
into a voltage. The overall circuit linearity is dependent
only on the linearity of the voltage to frequency and
frequency to voltage converters. Another modification on
this technique is frequency modulation about a carrier
frequency, fO' Here, VIN modulates fO by t:.f depending
on the amplitude of V IN' At the output, V OUT is
reconstructed with a phase locked loop or similar circuit.

VOUT
V
IN

k
1 2

Figure 3.5.7.2-2

Voltage to Frequency Conversion Techniques

Be Used to Isolate an Analog Signal.

In the "off' state, the voltage of the input diode is usually

zero or negative. It may, however, be slightly positive. It is
worth noting that at forward voltage as high as 1.2 volts,
the forward current is often neglible IOIlA). The "off'
state consideration may also have some relationship to the
threshold condition that should be considered.
For the "on"
state and threshold condition, the
consideration is illustrated in Figure 3.6-1. In analog types
of optoisolators, the output current is a relatively smooth
function of input current, the ratio being the CTR (Current
Transfer Ratio, see Section 3.2.5). As indicated in Figure
3.6-1{a) the required minimum level of forward current, IF'
for a proper "on" state is related to the output circuit. A
good practice is to design the drive circuit so that IF is at
the level for which data sheet specs are given, then adjust
the load accordingly, allowing an appropriate guardband for
CTR degradation. Note also that there are maximum limits
on IF for both DC and pulsed operation.
The threshold condition can also be defmed, with reference
to Figure 3.6-I{a), as the threshold input current, IFth, at
which the output voltage, V0' is at the threshold level,
VOth' for tlle particular logic type used in the output
circuit.
For
example,
with
TTL
logic,
V Oth "'" 2V be "'" 1.5V. Because of the influence of
temperature, CTR degradation, and unit-to-unit variability,
IFth is not precise enough to describe the threshold for a
comparator--type application unless the reference current is
much greater than IFth. Nevertheless, IFth can have some
influence on the propagation delay through the isolator,
and should therefore be considered.

Can

While there are many different digital applications of

optoisolators, there are some input design considerations
they all have in common. These considerations are:
the "off' state: input diode forward current very
small, zero, or negative (reverse biased)

100%
IF ;;. CTR
(al

100%
x IOl = CTR

(Ill +

FOR ANALOG

VCC - VOL
R
I

ISOLATORS

.01

the "on" state: input diode forward current at a

steady, adequate value.
c)

IlF

I
..2.
l

threshold: a transitional but defmable level of input

diode forward current.

For (a) and (c), the only consideration usually given is that
they are unimportant, but this is not always true. For (b),
the "worst case" limits are usually narrower than for (a)
and (c), so (b) receives more attention.

I ;;. MIN. SPECIFIED

F AT GIVEN IOl' VOL
(bl FOR DIGITAL
Figure 3.6-1

Rl;;.

VCC - VOL
IOl - III

ISOLATORS

Input and Output Considerations

Operated Optoisolators.

for Digitally

In digital types of optoisolators, the threshold and "on"

state
input
current levels are not subject to load
adjustment. The output current is not a smooth function of
input current. An example is shown in Figure 3.6-I(b). III
the 5082-4360 type of optoisolator the threshold may b ~
anywhere between the limits specified for 10H and 101:
There is, however, a specified level of IF above which th~
output transistor will be capable of sinking a specified
current.
Guardbanding the load to allow for CTR
degradation is not useful. Good practice is to design th~
drive circuit to provide a level of IF slightly higher than the
specified minimum. (This technique may also be applied to
the circuit of Figure 3.6-1(a).) This form of guardbanding
for CTR degradation narrows the spread between the
minimum IF required and the maximum data sheet ratin~;.
Techniques are given in Section 3.6.1.2 for dealing with a
narrow spread.

RS =

VL-VF
IF

(a) SINGLE SERIES RESISTOR, IL = IF

(b) SINGLE PARALLEL RESISTOR, VL = VF

NOT RECOMMENDED IF IF <t 'L

When digital data is transmitted over any appreciable lengb

of transmission line (even less than a meter), there arises a
possibility
of ground
shift, ground looping, etc.
Optoisolators can reduce the amount of ground loop
current and the effects of the resulting common mode
voltage. They are, therefore, very useful as line receiver:;.
This section discusses the fundamentals
of designing
optoisolators into line receivers.
Most line drivers are capable of sourcing a line voltage
higher than the minimum (~ 1.5V) needed to tum "on" an
optoisolator. They usually also deliver a line current greater
than the value of IF as determined with reference to Figure
3.6-1. It is possible, of course, to design a line driver thGt
will source the proper current and no more, but that is
usually not good practice, as will be described later. Good
practice is to drive the line with all the voltage and currert
available, then deal with the excess in the design of the
termination of the line.
In the design of the termination, the "off' state is usually
ignored unless pre-bias is used, and this is discussed itl
Section 3.6.3. There are usually three objectives in the
design of the termination:
(A)
(B)
(C)

proper "on" state IF

threshold level, IFth
reflections due to impedance mismatch.

As seen earlier, objective (A) is mandatory; objectives (B)

and (C) are discretionary. If discretion allows (B) and (C) to
be neglected, the termination may be as simple as in Figure
3.6.1-1 requiring only a single resistor. In most cases, the

Figure 3.6.1-1

Single-Resistor Terminations using Optoisolators

as Line Receivers.

series resistor termination of Figure 3.6.l-1(a) would be

used because it will accomodate a broader range of driver
and line resistance variables. It is slower than the single
parallel resistor termination of Figure 3.6.1-1 (b) because
the input diode is driven from a higher impedance.
Slowness can be remedied with a peaking capacitor in
parallel with Rs; if peaking capacitance is applied, the
anti-parallel diode should be used (even if the driver is
polarity non-reversing) to allow the peaking capacitor to
charge and discharge the maximum amount.
The anti-parallel LED in Figure 3.6.1-1 is recommended
also when polarity-reversing drive is used. It should be a
GaAsP/GaAs device (e.g. HP type 5082-4850), or the input
diode of another isolator of the same type to provide a
balanced output.
When, in addition to proper IF' consideration must be given
to either threshold current, IFth, or to line reflec~ions, an
additional
resistor provides one additional degree of
freedom. A two-resistor termination can accomodate the
additional objective of either setting the IFth level or of
approximate
impedance
matching
but not both.
Approximate impedance matching will be described first
because it is then easier to describe the condition for
threshold setting.
First of all, the nature of the reflection problem must be
understood. This can best be explained with reference to a
"design" having large, oscillatory reflections, illustrated in
Figure
3.6.1-2.
(This "design", incidentally is not
recommended; it is given here only as an example of what
can take place.) When the driver makes a logic transition
from "low" to "high", the source line is changed so rapidly

POLARITY
REVERSING
DRIVER
CLAMP AT
Ig = 8 mA
Vg =3V

LOW LOSS
TRANSMISSION

Zo

UNE

= 130n
I
+20
g,
LINE CURRENT

LOAD
LINE

mA
QH STEADY STATE,
IF, VF

+10

LOW- TO-HIGH
CONSTRUCTION

<.
~//

+1

."

+2

Vg,-L1NEVOLTAGE-VOLTS
HIGH-TO-LOW
CONSTRUCTION

,/
-10
QL STEADY STATE.
IF' VF
LOAD
LINE

REFLECTION CONSTRUCTION FOR LOW-TO-HIGH TRANSITION

(ASSUME QUIESCENT LOW INITIAL CONDITION, QL)
FROM QL CONSTRUCT A L.INE WITH SLOPE av/a 1= +Zo TO
WHERE IT INTERSECTS THE SOURCE LINE FOR LOGIC
HIGH AT POINT ig, vg1' THIS IS THE INITIAL STEP CURRENT
IN THE LINE.

1,

FROM
vg1' CONSTRUCT A LINE WITH SLOPE av/a I =~
TO
WHERE-IT1NTERSECTS
THE LOAD LINE FOR LOGIC HIGH AT
POINT i, vl' THIS IS THE INITIAL STEP CURRENT IN THE
LOAD.
FROM i, vl CONSTRUCT
FROM vg2 CONSTRUCT

+ZO LINE TO vg2 ON SOURCE LINE.

-Zo

LINE TO i2 ON LOAD LINE.

ETC., ETC., UNTIL CONSTRUCTION

Figure 3.6.1-2

+3

CONVERGES

AT QH'

Graphical Analysis of Reflections with Non-Linear

Line Driver & I.oad

as to be regarded as a step function. At this instant, the

load on the driver is just the transmission line with dynamic
resistance Zo and initial conditions of voltage and curreat
corresponding to point QL' Thus the initial voltage and
current in the line at the driver end are found as the
intersection of the "high" source line with a line through
Q and having a POSITIVE Zo slope. This point is labelled,
L
ig, vgl' and this is the voltage and current that will
propagate toward the load. Upon reaching the termination,
the transmission line is now regarded as a source having Zo
internal resistance with point i ,v 1 as its initial condition.
It is, therefore, represented as ~.Illie passing through i ,v'JI
with a NEGATIVE Zo slope, and its intersection witfl tJle
load line gives the first voltage and current, iQ,vQ1, at the
termination (load). This initial value, iQ,v l' will now travel
back to the driver where it presents to ~e driver a load of
dynamic resistance Zo and initial conditions iQ,vQl' This is
represented by a line through iQ,vQl with a POSITIVE Zo
slope and, intersecting the source line gives a voltage v 2'
the second terminal voltage at the driver. The dri~er
terminal voltage, v 2' now travels back to the load, again
along a NEGATIV~ Zo slope, to intersect the load line at
iQ2' the second load terminal current. Then a POSITIVE Zo
slope from iQ2 intersects the source line at v 3 and a
NEGATIVE Zo slope through vg3 intersects the lo~d line at
iQ3. The construction is repeated until it converges at QH'
Notice that the transmission line is alternately represented
as a Zo load (POSITIVE Zo slope to intersect source line)
and as a Zo source (NEGATIVE Zo slope to intersect load
line ).
This graphical method does not account for lossiness in the
line, or for transient capacitive or inductive effects at the
terminals. Neverthless, it is a powerful aid in visualizing the
effects of any load line or source line adjustments made hy
adding or changing resistances. Consider, for example, the
addition of a series resistor, Rs' at the load in Figure
3.6.1-2. This would be represented as a positive slope of R
starting from ~ 1.5 volts at zero current. This would hai'~
brought iQ2 much closer to v 2 without having changed the
current at QH by very muck Consider also the case of a
purely resistive Zo load; QL and Q would then be joined
H
by a +Zo sloped line and there would be no reflections.
Such reflectionless design is not possible with optoisolat:>r
terminations because the impedance lowers as the volta,~e
rises from below the turn-on voltage of the input diode to
above turn-on. The best that can be done is to make sure
that whatever reflections there are permit a monotonic
approach to convergence at the quiescent point (QL or
QH)' rather than the oscillatory convergence seen in Figure
3.6.1-2. Monotonic convergence means that each successive
step at the termination lies between the last step and the
quiescent point -- not beyond the quiescent point. The
problem with oscillatory convergence is that the first stt~P
may be high enough to initiate turn-on, while the next stt~P

might be below turn-on, etc. Thus, in response to a

single-edge logic transition there may be one or more
extraneous pulses, i.e., one or more extra edges resulting
from one edge.

The optimal termination lies somewhere between CASE 2

and CASE 3, with CASE 1 being a good starting point.
Graphical construction of two-resistor terminations are
shown in Figures 3.6.1-3 and 3.6.1-4. They are called the
low-threshold (LT) and high threshold (HT) circuits with
reference to the line voltage at which turn-on begins. They
have the same relative merits, respectively, as the series and
parallel one-resistor terminations of Figure 3.6.1-1.
Applying CASE 1, then doing a reflection construction by
the method outlined in Figure 3.6.1-2 suggests that CASE 1
would be an optimal design, but while this is nearly true, it
neglects the fact that turn-on of the input diode does not
happen instantly. The first load voltage point must lie on an
extension of the dynamic resistance below turn-on. This is
why CASE 1 is a good first approximation, but upward or
downward adjustment of the dynamic resistance above and
below turn-on might be necessary. Direction of adjustment
depends on the dynamic resistance of the driver in the
neighborhood of the intersection of its source line with the
Zo line. If its dynamic resistance here exceeds Zo adjust
toward CASE 3, but if less than Zo adjust toward CASE 2.
Going all the way to the CASE 2 or CASE 3 limits might be
too far because they may place the quiescent point in a
region where the dynamic resistance of the source line is
radically different from what it is at its intersection with
the Zo line.
If the source is very non-linear, a good approach is to select
as the quiescent condition that point on the source line
where the dynamic resistance is Zo (-Zo slope along source
line). Then use general construction to find the resistor
values for either the LT or the HT termination.
For line lengths greater than 50 meters, reflections are
unlikely to be a problem, so attention can be focused on
making a threshold selection.
There are THREE SITUATIONS
adjustment makes any sense:
(a)

for which

threshold

to establish the threshold for a comparator-type

application with slowly varying line drive.

I
I
<{

I /
1/

40

,-1'

....
o

a::
a::
:::> 30
u

/
/

/ I

I
I
/

..J

/
/
/

/
I
I
I

/
/

/
/
10

"

__

,
/,,,

A-::: ~-

_ .

IFth SOURCE
LINE IF

I
2.0
1.0
VL - LINE VOLTAGE - V
GENERAL CONSTRUCTION FOR ANV POINT, X, ON SOURCE LINE:
FROM POINT V AT A CURRENT IF BELOW X, DRAW LINE TO ORIGIN 0,
INTERSECTING VF LINE AT Z, THEN DRAW ZX.
aiL
1
aiL
1
1
SLOPE OF OZ: av =
SLOPE OF ZX: av - =
+R
SL
L
PL
L
PL

CASE 1 CONSTRUCTION - DRAW Zo LINE TO INTERSECT SOURCE LINE AT Xl'

V
THEN PROCEED AS FOR POINT X TO FINE RpL, RSL'
L
=Zo
IL
CASE 2 CONSTRUCTION FROM A POINT IF ABOVE 0, DRAW LINE PARALLEL
av
TO Zo INTERSECTING SOURCE LINE AT X2, THEN
_L
~ Zo
PROCEED AS FOR X TO FIND RpL, RSL'

aiL

CASE 3 CONSTRUCTION - CONSTRUCT A Z.' LOCUS OF INTERSECTIONS OBTAINED

FROM SEVERAL POINTS X' ON SOURCE LINE: THROUGH
EACH X' DRAVIIA LINE SLOPED 1/Z0 TO INTERSECT AT Z'
WITH LINE OV', WHERE V' IS AT A CURRENT I BELOW X'.
X3 IS THE POINT X' FOR WHICH THE Z' LOCU~INTERSECTS
THE VF LINE. FROM X3 PROCEED AS FOR X TO FIND
RpL, RSL'
Figure 3.6.1-3

Graphical Design of Low-Threshold

Termination.
--

Two-Resistor

E
0

30

a:
a:

:::::>

I /

I I

Co)

z
..J

20

I /
I I
I I
I I

..J

I I

II

II

,,/-/

'I / /:./
I

//
-J4.I/

/".

I .

3.0

VF
2.0
VL - LINE VOLTAGE - V
GENERAL CONSTRUCTION FOR ANY POINT X ON SOURCE LINE:
A LINE FROM X TO VF IS INTERSECTED AT Z BY A CONSTANT-CURRENT
LINE THROUGH Y AT A CURRENT IF BELOW X.
aV
SLOPE OF OZ: - L = RSH + RpH
aI

aV
SLOPE OF ZX: - L
aiL

= RSH

CASE' CONSTRUCTION - DRAW Zo LINE TO INTERSECT SOURCE LINE AT

X" THEI\ PROCEED AS FOR POINT X TO FIND
VL
RpH, RSH
=Zo
IL
CASE 2 CONSTRUCTION - CONSTRUCT Z LOCUS (OF TURN-ON VOLTAGES),
FOR SEV ERAL POINTS X, USING "GENERAL CONSTRUCTION':
FROM VF DRAW A LINE THROUGH THE INTERSECTION OF
THE Z LOCUS AND THE Zo LINE TO INTERSECT SOURCE
LINE AT )(2' THEN PROCEED AS FOR POINT X TO FIND
RpH, RSH
FROM VF DRAW A LINE SLOPED '/Zo TO INTERSECT SOURCE
LINE AT )(3, THEN PROCEED AS FOR POINT X TO FIND
RpH,

RSH

Graphical Design of High-Threshold

Termination.
--

Two-Resistor

to balance delays when a single isolator is used with

polarity non-reversing drive.
to optimize data rate with either single or dual
isolators and polarity reversing drive.
The last one (c) is easy: make the threshold as low as
possible. This is best done with an active voltage-clamp
termination, as will be seen later. In designing for (a) or (b),
it is first necessary to determine what the SOURCE LINE is
for IFth.
The most precise determination of the IFth source line is
by transient measurement of the V-I characteristics at the
termination for the instant at which threshold current is
desired. This is usually more labor than is justifiable by the
results because any solution will yield only approximate
resistor values and subsequent tweaking will be needed for
either (a) or (b) above.
The next best is to make shrewd assumptions based on
known system characteristics. Some good possibilities:
1.

2.

If the driver is back-matched (impedance = ZO)

construct the IFth SOURCE LINE parallel to the
steady-state source line -- spaced halfway between
zero and steady-state for situation (b), and spaced
at whatever is appropriate for (a).
If the line is long enough that the round trip travel
time of a pulse exceeds tpHL or tpLH, then, for (b)
construct the first two legs of the reflection diagram
(Figure 3.6.1-2) starting from VVIL = O. If the
second leg (NEGATIVE Zo slope through i ,vgl)
falls below the steady-state source line, take litis as
the IFth source line. If it falls above the steady-state
source line, forget it.

To satisfy all three of the design objectives requires a

three-resistor
termination
in either a "1T" or "T"
configuration. This is a rather complicated procedure, and
for some driver/load combinations, solutions do not exist.
The subject is discussed in EDN, in a five-part article: Feb.
5,20, Mar. 5, 20, and AprilS, 1976.

Unlike resistive terminations,

active terminations
can
respond to a variety of driver and line conditions to deliver
the proper IF to the optoisolator input, provide a low
threshold, and absorb re-reflections from a mismatched
driver. They should, therefore, be considered for the
following situations:

3.

For either (a) or (b), assume that the driver voltage

and current rise proportionately and construct the
IFth source line as the locus of points obtained by
taking points on the steady state source line and
reducing the voltage and current by the same
proportion -- by 0.5 for (b), and by whatever is
appropriate for (a).

Next, for points on the steady-state source line perform the

general construction (Figure 3.6.1-3 or 3.6.1-4) to find the
point at which the desired values of IF and IFth are
obtained. With a backmatched source and assumption No.
1, the solution can be found analytically by substituting
IFth and Eth for IF and E in the circuit equation (E and
Eth are intercepts of the steady-state and IFth source lines
at IL = 0). These substitutions yield a pair of simultaneous
equations to be solved for the resistor values. For the LT
circuit, use:

Busing or current looping where changing the

number of stations affects the current available to
each station.
Power supply fluctuation at the driver.
Design flexibility requirement; same termination to
be used with any of several types of drivers.
Accomodation of variation in line characteristics or
length, where line resistance is significant.
Temperature change on long lines (6R/R ~ O.4%tc
for copper wire).
Data rate enhancement where a long or lossy line
degrades the rise time at the termination.
Driver mismatch causing reflections.

Active terminations are simply current regulators for the

optoisolator
input
diode.
There are two classes:
current-clamp regulators allowing terminal voltage to rise,
and voltage-clamp regulators allowing terminal current to
rise. Whereas for resistive terminations, there are three
design objectives to consider, with active terminations there
is really only one: proper steady-state IF. So far as
threshold is concerned, a voltage-clamp regulator allows it
to be lower than any resistive termination can permit
without allowing IF to rise excessively. Reflections are of
no concern unless there are other stations on the line that
might
suffer
-- the active termination
absorbs the
current/voltage fluctuations caused by reflections. If the
active termination causes reflection problems for other
stations, these can be reduced by addition of series or shunt
resistance.

Current-clamp circuits are shown in Figure 3.6.1-5, and the

operating principles for polarity non-reversing drive are seen
in the V-I characteristics. Note that isolator input curreLt
does not flow until VL > (V F+V be)' and beyond tha:,
there is some time delay in turning on Q2. Lowering R2 to
speed up the Q2 turn-on might cause unsatisfactory
regulation of IF in the one-port circuit, but is no problem
with the two-port. The main advantage of the one-port ov:r
the two-port
is the simplicity of converting it to
polarity-reversing drive if IVLI > (4 Vbe +'VF) (if this wele
not so, a current-clamp regulator would be a poor choice
anyway) -- a diode bridge allows the same regulator to limit
current in both directions.
Voltage Clamp circuits are shown in Figure 3.6.1-6 for
polarity non-reversing drive, along with V-I characteristic>.
Circuits for polarity-reversing drive are shown in FigUJe
3.6.1-7. Note that turn-on begins at VF, and fully regulated
current is obtained with VL = (VF + Vbe) for either
polarity-reversing or polarity non-reversing drive. Compale
this
with the current-clamp
circuits
requiring
VL = (VF + 2Vbe)
for polarity
non-reversing and
VL = (V F + 3V be) or (V F + 4V be) for polarity-reversir g
drive.
Note also that the regulator does not become active until
Q1 turns on. That is, there is no transistor turn-on delay of
isolator input current (as for Q2 in the current-clamp
circuits). Moreover, since the turn-on of Q 1 lags the flow of
line current, there will be a momentary surge of line current
into the isolator input diode. A similar lag in turn-off of Q 1
hastens the turn-off of the optoisolator. This peaking effect
enhances the data rate especially where long or lossy lines
cause long rise-time at the termination.
The terminal voltage compatibility makes the voltage-clamp
regulator ideal for busing. Although slight differences in
VF, Vbe, or {3 might cause one regulator to operate at
higher current than others in parallel with it, the volta~;e
will be the same on all, so each isolator input diode will
have its own proper IF. That is, isolators requiring high IF
can be terminated in parallel with low-IF types, providin:~,
of course, the line driver can supply the total current
needed.
In current looping, the build-up of terminal voltages around
the loop limits the number of stations that can be operated.
Also, as stations are added or removed, there may be a
considerable change in loop current. For these reasons, tr.e
low operating voltage and broad range of regulated
operation makes the voltage-clamp regulator a good choicl:.
3.6.2

Common Mode Rejection (CMR) Enhancement

Common mode interference is so named because it appears

as a voltage which is, as referred to output ground, commcn

to both input terminals of the optoisolator. It is labelled

eCM in Figure 3.6.2-1. The desired signal appears
differentially and is therefore called the differential mode
signal, eDM.
Characterization of the basic optoisolator CMR properties
are discussed in Section 3.2.1. Attention here is given to
various circuit techniques for improving CMR over that of
the basic isolator.
eCM comes about in either of two ways: "induced",
represented by el in Figure 3.6.2-1, or "conducted",
represented by e2. In some situations, e2 is inherent to the
system; for example, if the isolator is used to control a
module floating at some large voltage with respect to driver
ground, this large voltage is represented by e2. Another
example of conducted interference is that which comes
through
the interwinding
capacitance of a power
transformer.
e1 represents the voltage between the
transmission line and ground or between the line and some
parallel conductor. Although shown as capacitively coupled
in Figure 3.6.2-1, e1 could also be magnetically induced, as
might occur in industrial machine control applications in
which control and power lines are close together over
appreciable distances.
If e1 is more tightly coupled to one side of the line than to
the other, the difference can cause a substantial voltage
difference between the two lines. Twisted pair lines are
used to balance the common mode coupling, and a shield
over the twisted pair aids in balancing induced interference
even if the shield floats; that is, no shield connection at
either end. If the shield is to be connected anywhere, it
should be to a point at the junction of two resistors
connected in series across the receiving end of the
transmission line. The only exception to this rule occurs
when the shield is used as a third conductor to pre-bias a
balanced split-phase receiver, as in Figure 3.6.3-3. A further
aid in balancing out e1 is to have the internal driver
impedance balanced to driver ground. The effect of
unbalanced
induced
interference can be somewhat
compensated by adjustment of the resistors from each side
of the line to the shield at the receiving end.
Even with e1 balanced, it is still present on both terminals
at the receiver, and, along with e2, comprises eCM' which
can be capacitively coupled to the amplifier on the output
side of the isolator, as discussed in Section 3.2.1. There are
a number of techniques for dealing with eCM:

Neutralization
Balanced differential amplification
Amplifier de-sensitization
Selective flip-flop output
Exclusive-OR flip-flop
Use of high-CMR devices

25
ONE-PORT CHARACTERISTICS
c(

20

R2SLOPES

I-

z
w
ex:
ex:

--

IL

(b) (e)

(a)

15

::l 10
(J

VLr

w
Z

...J

...J

I
00

4
2
3
V L- LINE VOLTAGE

5
-V
IL--

(b)POLARITYREVERSING,
USING
ISOLATOR INPUT DIODES AS HALF
OF BRIDGE SO V L(MAX) V CEO + V F)

(e) POLARITY

REVERSING
WITH EXTERNAL
BRIDGE RECTIFIER

25
c(

VL(MAX)~

20

I-

z
w

ex:
ex:

::l

(J

15

IC(MAX) OF Q2
R2 Slopes (e)

(d)

10

(cI)POLARITYNON-R EVERSING

u.

.::.

V
BE
R1 = I
,ALL
F
VL(MAX)
R2=

n
In

CIRCUITS
-VF -nVBE
In

IF(MAX)

(e) POLARITY-

IC(MAX)

Figure 3.6.1-5

__
F

REVERSING
SHARES R1
BUT REQUIRES
SEPARATE Q1'
Q2,AND R2
Current-Clamp Active Terminations:
la, b, c) and Tv/a-Port Id, e).

One-Port

CHARACTERISTICS
OF (b)
AND (a)

E
I

~ 40

z
w

":":>

()

z
...J

20

I
/I~

...J

IF(MAX)

VL - LINE VOLTAGE

- V

20
40
IL- LINE CURRENT - mA
IN BOTH (a) AND (b) R1 = VBE/IF
Figure 3.6.1-6

Voltage-Clamp
Non-Revening

Active Terminations
Drive.

IF-

IL(a) LOW-GAIN

for Polarity-

vcc

--

DIODES DF AND DR
ARE REQUIRED TO
PREVENT REVERSE
TURN-ON OF THE
TRANSISTORS

VLr

V{

./
./

GND
5082-4364

(b)HIGHGAIN

--

THE EMITTERBASE
JUNCTIONS OF Q2

+
-

AND Q2R SERVE

AS DF AND DR IN (a).

V
L

REVERSE BREAKDOWN
IS NO CONCERN
BECAUSE THE LINE
VOLTAGE IS CLAMPED.
Figure 3.6.1-7

Voltage-Clamp Active Terminations

Reversing Dr ive.

for Polarity-

* CAPACITORS ARE SCHEMATIC

REPRESENTATION OF HOW EMI
INTERFERENCE IS REDUCED
BY TWISTING AND SHIELDING.

,-----....,

z
o

I
I

ccw~

0>
w-z
>1--

I
I
I
I
L

_~:E
I-cncc
uww

CCI-

e2 REPRESENTS INHERENT OR CONDUCTED INTERFERENCE

e, REPRESENTS EMI (ELECTROMAGNETICALLY INDUCED) INTERFERENCE
8CM = (e, + e2)

COMMON MODE INTERFERENCE

eDM = DIFFERENTIAL MODE SIGNAL

Figure 3.6.2-1

Common Mode Interference; Schematic

Representation.

Neutralization can be used with optoisolators having a

single-transistor amplifier operated common-emitter. Also,
it is effective only while the transistor is active -- not
saturated or cut-off. However, since it is while the transistor
is active that eCM is most troublesome, it is a technique
worth considering. It consists simply of a neutralizing
capacitor, CN in Figure 3.6.2-2(a) that sources a current
CN (deCM/dt) which is opposite in polarity to the collector
current J3 CCM (deCM/dt) caused by eCM If CN = J3 CCM'
these opposing currents will be balanced and the collector
voltage will not be affected by eCM. Even if CN does not
precisely neutralize eCM' its effect is beneficial. CA UTION:
CN must have a voltage rating high enough to accomodate
the maximum voltage that eCM might attain.
Balanced differential amplification also works only if the
isolator output amplifiers are active. It is clear in Figure
3.6.2-3 that if both isolators are driven to cutoff or
saturation by an eCM transient, they cannot maintain a
differential response to eDM.
Amplifier de-sensitization is perhaps the simplest defense to
raise against eCM. The base bypass resistor, RBE in Figure
3.6.2-4 reduces the impedance in which the current flows
that is coupled via CCM by any eCM transient. Thus a larger
eCM transient can be tolerated. It also reduces the
amplifier's sensitivity to photocurrent resulting from eDM'
ann therefore, requires a higher current in the input diode

for a proper "on" state. The base bypass can be used with
either single-transistor or split-darlington types of amplifiers
and offers the additional benefit of higher data rate
capability (see Figure 3.6.3-5). RSK is recommended where
there is a risk that a body charged with static electricity
may be discharged with all or part of the surge current
passing
through CCM. Being amplified by the first
transistor, such surge current can then be destructive.
A selective flip-flop output circuit can take advantage of a
situation in which the eCM transients have a higher rate of
change in one direction than in the other, such as a
sawtooth. It can also take advantage of a circuit with
CMH > CML or vice versa (see Section 3.2.1). Since a
NAND flip-flop can tolerate having both inputs high, it
should be used where the likelihood is greater for both
isolator outputs to be high (due to eCM transient) than is
the likelihood that both outputs will be low. Conversely, a
NOR flip-flop can tolerate having both inputs low, so the
rule is reversed. These rules are summarized in Figure
3.6.2-5.
Exclusive-OR
flip-flop, whether of NOR or NAND
construction can tolerate either both inputs high or both
inputs low without either of its outputs changing state. The
outputs can change state only in response to a change in
differential input, so it has infinite common mode rejection
for a static condition of eDM in either logic state. Note in

CONNECT
EITHER
ANODE OR
CATHODE

(f3C CM )aeCM
at

CURRENTS NEUTRALIZE
IF CN

= f3CCM
~7pF

(a) BASIC SCHEME FOR NEUTRALIZATION.

CAPACITOR MUST HAVE ADEQUATE

NEUTRALIZING
VOLTAGE RATING.

DRIVER

DRIVER
2

~
5082-4354/55

N2

(b)NEUTRALIZATION
APPLIED TO OPTOISOLATOR WITH DUAL
SINGLETRANSISTOR
AMPLIFIERS.
SEPARATE NEUTRALIZING
CAPACITORS ARE NEEDED FOR EACH COLLECTOR REQUIRING
NEUTRAL,.IZATION.
INPUT DRIVERS NEED NOT HAVE A COMMON
GROUND REFERENCE.
Figure 3.6.2-2

Neutralization of Common-Mode
External Cil pacitor.

Interference

with

+5V
VCC

+5V
VCC
R1

R2

--

IF1

VFr

~7405

R3
VFr

--

5082 - 4354/55

IF2
1

IF1

IF2

OUTPUT

IL

IH

IH

IL

INPUT

IF1

INPUT

VCC-

IF2
2VF

VCCR

R1 + R3

RESULTS OBTAINED
R1

R3

180, R2

+ R3

VCC-VF

VCC - VF

VCC-VF
1

2VF

R1

R2

R2
VF

-R;

WITH
270

INPUT

IF1 (mA)

IF2 (mA)

5.6

18.5

19.4

4.6

R2

REQUIRED RATIO
OF RESISTORS

2(1 _ VF )

VCC
R1;;;'

;;;.
L

V C-VF
C
IC(MAX)

VCC - VF

----

IH

::::::470

IT IS NOT NECESSARY TO HAVE PRECISE BALANCE OF ON AND OFF-STATE INPUT CURRENTS

AS LONG AS THE RATIO OF THE LOWER HIGH'-O THE HIGHER LOW EXCEEDS THE RATIO OF
THE HIGHER CTR TO THE LOWER CTR. SUCH L NBALANCE CAN ALSO BE COMPENSATED BY
ADJUSTMENT OF RL.

Figure 3.6.2-3

Active Differen tial Drive for CMR Enhancement

and High Data I~ate.

REDUCE RBE, RL FOR:

LOWER AMPLIFIER

GAIN

HIGHER CMR
LOWER CTR

RESTRICTS SURGE; DOES NOT HELP OR HURT

SK
CMR OR CTR. RECOMMENDED VALUE:
1 VOLT
R SK-.- F
15xI

UNLESS VBE >0.6V, OUTPUT

TRANSISTOR IS OFF, AND
UNLESS V <VOth OUTPUT

LOGIC IS HIGH.

1000

i=

a:

800

a:
w

u.

en
z

a:

600

lIZ

w 400

a:
a:
~

u
I

a:

I- 200
U

1
IF - INPUT CURRENT - mA
Figure 3.6.2-4

Amplifier J)esensitization
by Lowering CTR.

for CMR Enhancement

NOR OR NAND
FLIP-FLOP

CMR
WITH RESPECT TO:

PREFER NAND
PREFER NOR

ISOLATOR
PROPERTIES

eCM
WAVEFORM

CML < CMH

-----\

CML>

Figure 3.6.2-5

CMH

r----

ISOLATOR
PROPAGATION
DELAY
tpHL

>

tpLH

tpHL

<

tpLH

Selective Flip-Flop Output for CMR Enhancement

and Edge Sharpening.

Figure 3.6.2-6 that the exclusive -OR flip-flop requires the

output of each isolator to drive two gate inputs. If the
isolator cannot handle two inputs, a buffer inverter can be
inserted at no loss of CMR.
CAUTION: although the exclusive-OR flip-flop has infinite
static immunity to eCM its dynamic immunity is not
infinite_ If a common mode transient capable of holding
both isolator outputs either high or low should persist
throughout the duration of a differential mode pulse, the
pulse will escape detection. Other than this, the worst a
common mode transient can do to the exclusive-OR
flip-flop is to advance or retard the timing of the flip-flop
output transition, i.e., cause jitter of the edge.
High CMR devices are those in which the ratio of optical
coupling to capacitive coupling has been made higher than
that of the usual "sandwich" -type optoisolators. One such
technique is the use of a conductive transparent screen over
the detector (see Section 3.2.1). Another is the use of
lenses or fiber-optics to couple the optical signal efficiently
over longer distances, thereby physically separating the
detector from the input and reducing capacitive coupling
without
a severe penalty in optical coupling. Such
arrangements are discussed further in Section 2.5.5.

In addition to the tpHL and tpLH data rate restrictions

imposed by the optoisolator there are system limitations.
Some of these can be optimized and others can be
compensated. In compensating, the usual tradeoffs are
reduction
in differential-mode
noise
margin or
common-mode rejection. This is especially true when
single-ended (rather than balanced split-phase) optoisolators
are used because the compensation reduces the amount of
differential-mode
voltage change needed to cause the
isolator output circuit to change logic state. The circuit is
therefore more vulnerable to differential-mode noise and to
common-mode noise which has been partly differentialized
by unbalanced impedance.
System optimization consists of reducing the degradation in
the rise time of step changes in the differential mode signal,
eDM' If such degradation is due to stray wiring capacitance
or inductance the remedies include reducing the impedance
of the eDM signal source, enlarging the wires or circuit
board traces, and dressing leads properly.
The potential data rate devastation of a transmission line is
often underestimated. This is because its transient response

Vcc
+5V

Vcc
W

(,J

.oJ

In

en

J:

Q,.

..,:. w
.oJ

>

./
./

a:
en c

Q,.

GND
5082-4364

Exclusive-Or F'lip-Flop for Infinite CMR Under

Steady State Conditions.

dB)

m=

aO (

_6_d_B
_
LINE LENGTH (m)

fOto = 0.164

inw
z

3
o

co."

3....:

0.1

10

g f3
-Z

E3

80
.-w

C(C/l

<l:

09

e:

" J:

t;

a:<l:

e:

"":.::

<l:

Ie:

J:
U
Z

!g~
I~

zz
2u
I-a:

:>
1E

I:

<l:a:
.01

1.0 ~ ~

<l:
I

~o
I-Z

<l:<l:

~o

/
I
/

/
I
/
/
.001

10

0.1

f - FRElUENCY
Figure 3.6.3-1

Transmission
Frequency.

- MHz

Line Attenuation

as a Function

of

is not the same as the familiar RC-type transient. A good

estimate of transmission line performance, and appropriate
design choices, can be made using the relationships
described in Figures 3.6.3-1 and 3.6.3-2.
Due to skin effect, most transmission lines have an
attenuation vs. frequency that varies as the square root of
frequency over the frequency range of the solid lines in
Figure 3.6.3-1. If a curve is not available, it is simple
enough to generate one by measuring the relative output
voltage, Viand V 2' from two lengths, Q1 and Q2' of cable
at some particular frequency, f. Then on coordinates of log
Q
vs. log f, plot the point obtained by measurement at f:
o

Through this point, draw a line sloping upward at half a

decade of Qo per decade of f. The lengths of line used in the
measurement need not be the same as required for the
system -- sample lengths will do, but the longer the samples,
the greater the precision. If the generator and load
impedances match the characteristic impedance of the line,
a single length, Ql, can be used; then substitute Q2 = 0 and
V2 = generator voltage (with matched load connected) in
equation 3.6.3-1. The frequency used should be in the
range of 1 to 10 MHz.
Response of a transmission line of length, Q, to a step input
is found as follows:
1.

[20 log (V21V,)1

(f)=-_

[Q, _ Q

Compute
dB:

(~)

o by dividing the line length, Q, into 6

6 dB
Q(m)

eltl!

E TL

eltl/
E

[
RC

1-

-.693..!.. ]

to

e~tll

= [l-A(~)J

v't7tO

= [l-erf

TLV

(.

=2 [l-P(~)J

.,Jtit;.

~)l
tlt J
o

cV2=0.674500
c = 0.476944

tl)= time required for e(t) to reach % final value, E (see Figure 3.6.3-1)

0 0.6
::>
I::::i
Q.

:e

w
>
~

..J

0.4

a:
I

~Iw

0.2

3
10
t/ta - ~ELATIVE TIME SCALE
Figure 3.6.3-2

Transmission
the Input.

I.ine Response to a Step Change at

With ao from Step 1, enter Figure 3.6.3-1 and find

fo at the intersection of ao with the curve for the
selected cable. (fo is the 6-dB frequency for the
length,
Q, and could
be obtained by the
measurement procedure described above.)

A(x)'

b:/ :

-J>"

21
-x

....[2;

A (x V 2 ) =
Step response, excluding travel time, is found fro:n
the normalized curve (solid line) in Figure 3.6.3-:~,
where to is the time required for the transient 10
change through half its asymptotic value. The log.c
delay, tTV imposed by the transmission line is
then:
tTL

to x hth/to)

(3.6.3-n

where (tth/to) is the value of (t/to) at which

e(t)

where:

Eo = initial line voltage

Eas = asymptotic voltage
Eth = voltage where switching occurs

Step 1 gives ao

dt

- Y2

t dt

~f

v'1r

_t

dt

Some math tables give the area A(x) between the limits -x
and +x; others the area P(x) for _00to +x; still others give
the area between +x and +00. Since the normal curve is
symmetrical and the total area, _00to +00, is just 1.0, any of
these tables can be used. Jahnke and Emde (4th Edition,
p.24) gives erf (x) with adequate precision. A great deal of
precision is not worthwhile because the results of the
procedure given here are only estimates to be used for
avoiding serious problems with data rate limitation.

4.

Given threshold, initial, and asymptotic voltages,

use equations 3.6.3-4 and 3.6.3-5 and Figure 3.6.3-2
to find a value of to for tolerable delay, tTL-

0.06

Step 2 gives fo = 8.5 MHz (typical curve, Figure 3.6.3-1)

Step 4 if f>E/E = 0.7, (tth/to) = 3
tTL = 19.3 ns x 3 = 57.9 ns
Note that if f>E/E had been 0.8, tTL = 19.3 ns X 7 = 135.:
ns, which is more than twice the time delay for only a small
change of threshold. The importance of threshold in dealing
with transmission line effects can hardly be given enough
emphasis.
Shown for comparison in Figure 3.6.3-1 is a linear (RC)
transient which also rises to half its fmal value at t = to' bU1
note that it attains 90% in just 3.3 X to while the
transmission
line transient requires ::::::
30 X to' The
attenuation vs. frequency of the same RC network is shown
in Figure 3.6.3-1 for comparison with the performance of
100m of twisted pair shielded cable -- scales at right side.
The normalized transmission line transient can be plotted
either from tables of erf (x) or from tables of the area
under the "normal curve of error", A(x):

Enter Figure 3.6.3-1 with fo to where it intersects

the cable curve at ao'
From equation 3.6.3-2, compute the maximum
length of cable that can be used. If that is too short,
calculate ao from equation 3.6.3-2 for the desired
length; then the intersection of this value of ao with
f from Step 3 gives a point on Figure 3.6.3-1
o
below which the cable curve must lie. A proper
cable may thus be defined.
A handy relationship to bear in mind is that to increases as
the square of the length of the cable.
The curve of Figure 3.6.3-2 can be used even if the line is
mismatched at both ends. During the transient, the
transmission line, as seen at the termination, is a source
with impedance Zo and open-circuit (Thevenin) voltage
changing in the manner described by Figure 3.6.3-2. The
voltages, Eo, Eas' Eth to be used in equation 3.6.3-5 are
found most easily by graphical construction, as shown in
Figure 3.6.3-3. Worst-case delays occur under initial

conditions as determined by the quiescent point (QL or

QH). For the transient going from QL toward QH' a line
with NEGATIVE Zo slope through QL intersects the IL = 0
line at Eo. (NOTE: Because IL = 0 at QV the NEGATIVE
Zo line has zero length in this example. For a non-zero
example, see Figure 3.6.34b.) To find Eas begin with a
POSITIVE Zo sloped line through QL to intersect the
SOURCE LINE for QH (not necessarily at QH); from this
intersection, a line with NEGATIVE Zo slope intersects the
IL = 0 line at Eas. Eth is a property of the load
characteristic upon which there is a point, Xth, defining
threshold conditions for the termination. Thus a line with
NEGATIVE Zo slope through Xth intersects the IL = 0 line
at Eth. For a transient going from QH to QV construction
is symmetrical with a balanced polarity-reversing system;
with polarity non-reversing, construction is not symmetrical
but the rules above are the same for finding Eo, Eas' and
Eth
Compensation
for either linear or transmission line
transients can be done by peaking or by threshold-lowering,
or both.
Without peaking or threshold-lowering the lowest possible
threshold voltage at the terminals of an isolator input
circui tare:

where the series resistance may be either RS (Figure

3.6.1-1) or RSL (Figure 3.6.1-3) or Rl (Figures 3-.6.1-6,7).
VUh is the voltage at the point Xth in Figure 3.6.3-3. By
prebiasing the input diode, this threshold can be lowered.
As long as the prebias voltage is below 1.2V, the forward
current in the isolator input diode will be less than 10 IlA.
The effects of such pre bias on the logic delay, where the
isolator terminates a transmission line, are shown in Figure
3.6.3-3. The technique is useful also where no transmission
line is involved, but transient delays are present.
With a single-ended system, the threshold may already be
near optimum, as seen in Figure 3.6.3-3(a). The net effect
of threshold adjustment is determined with reference to
Figure 3.6.3-2 for a transmission line transient -- in other
situations, the effects can be found with reference to the
particular kind of transient involved, e.g. RC. In general,
threshold adjustment in a single-ended circuit shortens
delay in one direction while lengthening it in the other, so
it can be used to compensate for unbalanced tpLH and
tpHL in the isolator.
With a balanced, polarity reversing system, the optimum
threshold is at VL = 0, so the objective is to make it as low
as possible. The effect on data rate can be profound. In

Figure 3.6.3-3(b), the one-volt pre bias changes the delay

from"'" 12 X to to "'"2.4 X to. If to is only 20 ns, this
changes the delay from 240 ns down to 48 ns and raises the
NRZ data rate from less than 3.5 Mb/s to approximately 10
Mb/s.
Peaking is another form of compensation.
The usual
technique
is to place a capacitor across the series
current-limiting resistor. Then when the driver changes
state, the voltage on the capacitor, for as long as it persists,
augments the source. With respect to the circuit in Figure
3.6.3-4(a), the peaking can be analyzed with or without
transmission line effect. The construction at the left shows
the "temporary" load line caused by the voltage, Vp, on
the peaking capacitor. As the source line changes from QL
to QH' turn-off begins immediately in one input diode and
requires only a 2 X V F change of source voltage for turn-on
to begin in the other. If Vp persists while the source line
changes all the way from QL to QH' there will be a surge of
forward current in the diode being turned on. As described
in Section 3.2.3, this will enhance the data rate. Caution:
Peaking must not allow ISURGE to exceed the amplitude
or duration permitted by maximum ratings. If the source or
transmission line impedance is too low, peaking within
limits permitted can be done with bypassing around only
part of RS. In general, the rule for construction of the
"temporary" load line is to represent only that portion of
the load which is NOT bypassed, and shift it by the voltage
on the peaking capacitor.
If peaking is applied to a single-ended polarity-non-reversing
termination, the construction rule is the same as above. It
should be apparent, in construction, that an anti-parallel
diode is needed across the isolator input diode to allow Cp
to charge and discharge, even though the drive polarity does
not reverse.
A safe, but slightly less effective peaking technique for
polarity-reversing
balanced drive is shown in Figure
3.6.3-4(b). With respect to steady state conditions (Q
V
QH) the series resistor limits input current. In the transient
state, the diode being turned off has its charge transferred
to the diode being turned on. Until the initially-on diode is
discharged, the voltage on it augments the line-to-line drive.
Threshold is very near zero, being offset only by the series
resistance, RD, of the initially-on diode. Discharge time is
very short ("'" 20 ns), and if the drive transient does not rise
through 6E/E within this time, the threshold shifts over to
the ~h point on the steady state load line. This form of
peaking is, therefore, effective only for short lines or very
low-loss lines, having _to < 20 ns (see Figure 3.6.3-1).
Safe peaking is also obtained with the voltage-clamp
regulators of Figure 3.6.1-6,7. Until the regulator transistor
is turned on, current through the input diode of the isolator
is limited only by rl and ZOoThe effect can be enhanced

75n CABLE.

...-"--..

7417

E 40
I
IZ

I-

a:
a:
::>
(,J
w 20
z

a:
a:
::>

40

Z
w

~ 20

""
"",~

-I

I
-I

'\~

-I

'-

-I

'4

VL - LINE VOLTAGE

-"

LOW PREBIAS - RS = 0
RC
*TAKING ACCOUN TOF
INTERNAL RESISTANC E OF GATE

RS*
L-+Ii

130n
0 + 10*

RC
RS*

100n
18 + 10*

H -+ L

L-+H

H-+L

Eo

.4

3.85

1.05

4.15

Eas

3.75

.15

4.25

1.40

Eth

2.3

2.3

2.3

2.3

~E/E

.57

.42

.39

.67

~E
E

-=

THRESHOLD ADJUSTMENTS FOR DATA RATE ENHANCEMENT OR

DELAY BALANCING WITH :;INGLE-ENDED RECEIVERS - POLARITY
NONREVERSING.

Figure 3.6.3-3a

Threshold Adjustment (PRE-BIAS) for Data Rate

Enhancement; Polarity Non-Reversing.

0, - 2N3906

02 -2N3904

NO PREBIAS - SHIELD BALAIlICED FOR OPTIMAL CMR AT THE

EXPENSE OF DATA RATE.

S
PREBIAS -- (V CC R R+ R C
S

\')BE

< 1.2V

SHIELD RETURNS CURRENT OF "ON" SIDE, ALLOWS "OFF" SIDE TO BE PREBIASED.

ALTERNATIVE USE OF SHIELD WITH SPLIT-PHASE POLARITY-REVERSING DRIVE
AND BALANCED DUAL-ISOLATOR LOA[IEITHER CMR OPTIMIZATION OR DATA
RATE ENHANCEMENT.
Figure 3.6.3-3b

Threshold Adjustment (PRE-BIAS)

Enhancement; Polarity Reversing

with a capacitor bypass across all or part of Rl. This also

gives more consistent peaking because it is less dependent
on the transistor turn-on time.
CAUTION: Both peaking and threshold-lowering tend to
reduce CMR. If common-mode interference includes large,
rapidly
changing
transients, the CMR enhancement
techniques of Section 3.6.2 should be applied. For balanced
terminations, the exclusive-OR flip-flop of Figure 3.6.2-6 is
recommended.
Single-ended terminations
may require
internally shielded optoisolators.

Output circuitry may also require attention in data rate

optimization. With very high speed optoisolators the pullup
resistor should be made as small as possible, consistent with
current sinking capability of the optoisolator output and
current sinking requirements of the logic inputs being
driven.

for Data Rate

Slower types of optoisolators can be made faster by

"swamping" the base of the output transistor. With a
resistor, RBE, connected from the base to the emitter, the
logic delay, especially tpLH, is greatly enhanced by the
discharge path that RBE provides for the base. The gain
(CTR), however, is reduced, especially for low input
currents. It is therefore necessary to raise the input current
to obtain proper operation. Lowering RBE, at any
particular level of input current, reduces tpLH but raises
tpHL' as seen in Figure 3.6.3-5. Thus at any level of input
current there is a value of RBE at which tpLH and tpHL are
approximately
balanced. The optimum value of RBE
depends also on the value of the pullup resistor. Figure
3.6.3-5 shows recommended values of RBE and the results
to be expected as a function of input current.

*EXT. INVERTER
i74128*,74S140*,

REQUIRED

illA9614,

VFH

iDM8830

{+

ETC.

ANALYSIS WITH TRANSMISSION L1NI:

60
ct
E
I

ct
E
I

40

IZ

IZ

40

a::
a::

=>
(,)

20

=>
(,)

..J

..J

I..J -20

I..J -20
-2

-1

-3

2VF
-2

VL - LINE VOLTAGE - V

-1

(a) USE OF PEAKING CAPACITOR - RS1 x Cp

> tLH

OR tHL

SAME LINE
DRIVERS
AS IN (a)

(b)

USE OF CHARGE
STORED IN "ON"
DIODE TO PEAK
THE TURN-ON OF
THE "OFF" DIODE.
SHIELD CARRIES
A CONSTANT
CURRENT.

ct
E
I
I-

80
60

a::
a:: 40

=>
(,)

z
..J
I
..J

20

Figure 3.6.3-4

VL - LINE VOLTAGE - V

Peaking Techniques with Split-Phase PolarityReversillg Line Drive.

CHANNEL A
ESTABLISHES
REFERENCE
DELAY THROUGH
TWO INVERSIONS

5'-

AOUT
tptH

+ Vcc

RL

A
RBE

tpHL

B
'OUTPUT INVERTER CAUSES
APPARENT WRONG PHASE
OPERATING
CONDITIONS

RL

IF

(kn)

(mAl

5.6

3.6-16

1.0

3.6-16

o
""
""I

0.27

3.6-16

">
Cl:

...

~ 10J's

z
o
CJ
Cl:

a:

...

1~s

:l:

.."..."
..""

1.0-5.6

3.6

0.27-5.6

8.0

:l:

0.27-'-5.6
1.0

16.0

10
100
RBE - BASE BYPASS RESISTANCE - kU

Figure 3.6.3-5

Base Bypa~; Resistor, RBE, Effect on

Propagatior Delay.

Some situations require two or more stations to operate

from a common transmission line. There may be only one
station capable of transmitting while all others on the line
are receiving; this is called a simplex system. If two stations
are each capable of transmitting to the other, it is called a
duplex system. The more general case, where anyone of
several stations is capable of transmitting to all others, as
well as receiving, is called a multiplex system.
RS-232 is a popular form of simplex bussing, documented
by EIA as RS-232-C. The EIA specifications do not require
ground separation, however, optoisolators are often used as
ex tra protection
with lines longer than specified in
RS-232C.
The EIA receiver input specifications are
completely met using the termination in Figure 3.6.1-1(a)
with Rs;;;' 3,000 ohms as per RS-232C. Since RS-232-C
allows a line voltage as low as 3 volts, the isolator input is
required to operate with IF as low as 0.5 mA. Also,
RS-232-C limits the maximum line voltage excursions to
25V, so the power into the input diode cannot exceed
252/(4X3k) = 52 mW, and the antiparallel LED, while
recommended, is not essential.

RS-422 and RS-423 are EIA documents for simplex data

transmission at data rates higher than that specified in
RS-232-C. RS-422 is for line voltage balanced with respect
to driver
ground
and RS-423
is for unbalanced
(single-ended), but both specify polarity-reversing line
drive. Here, also, ground separation is not required, but
optoisolators of adequate speed of response can be used.
The RS-422/423 receiver specifications include a zero
threshold voltage, so the specification cannot be completely
met by any optoisolator due to the VF threshold for
turn-on. However, the RS-422/423 driver specifications are
compatible with the termination in Figure 3.6.1-7 if R1 is
90 to 100 ohms.

Wire cost in long simplex runs can be reduced using the

scheme in Figure 3.6.4-1 to transmit both clock and data
along
the same line. To avoid distortion of the
reconstructed clock pulse, the data transitions at the input
should be permitted only while the clock is low. By using a
pair of high-gain optoisolators at the output, the buffer
inverters can be omitted. Omitting the buffers will invert
the data but the clock output will be the same.

DATA

INPUT

CLOCK

INPUT

vcc

1----1
I

I
I
I
I

I
L_-r....J
7433

(PARALLELED
HALVES)

~
LSTTL
BUFFERS

SEE SECTION 3.6.1 FOR

ALTERNATIVE
TERMINATIONS

RG -120n
Rp - 820n
RS - 56n
RL - 6.8k

Figure 3.6.4-1

NOTE: DATA INPUT SHOULD

BE ALLOWED TO CHANGE
ONLY WHILE CLOCK IS
HIGH. IF DATA CHJl\NGES
WHILE CLOCK IS LOW,
THE RESTORED CLOCK
PULSE WILL BE
DISTORTED.

Single-Line Transmission
Signals.

of Both Clock and Data

Duplex data transmission between two modules is usually

done along a single line, with each module normally in the
"receive" mode. If ground separation between the module~:
is required, a pair of optoisolators can be connected a~:
shown in Figure 3.6.4-2(a) to establish two-way dat::.
transmission. In the quiescent state, i.e., if there are no
externally applied gates pulling PI or P2 low, they will b~,
pulled up by pullup resistors and both isolators will be off.
Applying a zero at PI turns on U 12 causing a zero to
appear at P2; the output of N2 remains low, and U2l
remains off, so when the zero is removed from P 1 th~
circuit returns to its quiescent state. Similarly, if a zero is
applied at P2, U21 goes on and Ul2 remains off. With both
isolators "off' in the quiescent state, loss of power in either
module does not affect the other. Use of optoisolators with
ou tpu t current
sinking capability high enough tc
accomodate more than one gate loading permits droppinE,
the buffer inverters (B1, B2) and leads to the modification
in Figure 3.6.4-2(b). Operation is similar, except that both
isolators are "on" in the quiescent state so loss of power in
one module causes a zero at P in the other. By driving the
cathodes of the input diodes, rather than the anodes, NOR
gates may be used in place of OR gates.

Multiplexing is commonly done by a "wired-or" bus as in

Figure 3.6.4-3. Any station can drop a zero on the line,
causing a zero to appear at all other stations. Obviously,
such a system requires some protocol to establish
transmission priorities. "Wired-or" bussing requires the
logic family in any station to be compatible with that of
other stations. Moreover, the common ground, shared by all
stations, may allow ground loop interference and here
optoisolators
offer a solution by permitting ground
separation at any or all of the stations. Optoisolators also
allow use of any logic family in any station, with the
isolator's "line" side performing the interface.

Each station
to be ground-separated
requires one
optoisolator to receive and one to transmit. The "receive"
isolator must be capable of operating with low enough
input current that the required number of stations can be
served, while the "transmit" isolator must be capable of
sinking enough current to cause a low at all the other
stations.

A BN RL
RS
Rp

1/67405
1/6 74LS04
1/4 74LS28
- 6800
-180
- AS NEEDED FOR LOGIC AT P1 AND P2
(a) MEDIUM-SPEED

DUPLEX ISOLATOR CIRCUIT

.01
1lF*

.01
IlF*

C -1/6 7405
0-1/47432
Rp - AS NEEDED
RS -470
RL -1000

U12, U21 OPTIONS:

50824360 FOR HIGH
DATA RATE, 5082-4370
FOR LOW DATA RATE

*CAPACITOR MAYBE
OMITTED FOR 5082-4370
(b) CIRCUIT MODIFICATION
Figure 3.6.4-2

FOR USING HIGH GAIN OR HIGH SPEED OPTOISOLATORS

Duplex Data Transmission
Ground Separation.

with Optoisolators

for

T - OPEN COLLECTOR TRANSMITTING

DEVICE
R - RECEIVER - POSSIBLY MORE THAN ONE PER STATION
Wired-OR BUi Multiplexing Requiring Common
Ground for a I Stations.

In the circuit of Figure 3.6.4-4(a), the buffer inverters

match the isolator input/output characteristics to whatever
logic family is used in the module. A low applied to PT
turns on the "transmit" isolator, causing the line voltage to
drop. Whenever the line voltage drops low enough to
approach the threshold voltage of the "receive" isolator, it
will turn off and cause PR to go low. Also, a low would
appear at the PR terminal of each optoisolator receiver on
the line. The optional diodes D3 and D2 enhance the data
rate by keeping the line voltage from dropping very far
below the "receive" threshold. An additional advantage is
that D3 will glow to indicate which, of the several statiOrlS
on the bus, is the one from which the "zero" is originating.
To have in each station a single terminal at which data c,m
be received as well as transmitted requires some kind of
anti-lock logic. For example, if in Figure 3.6.4-4(a), the
terminals PR and PT were to be connected together, the
entire system would lock low. Figure 3.6.4-4(b) shows one
form of anti-lock logic; in quiescent state the line is high w
V 0 is low and PTR is high. In the receiving mode, as the
line goes from high to low, V 0 goes from low to high, and
PTR goes from high to low, but the output of the NOR
gate remains low because V0 is applied directly to one of
its inputs, and inverted to the other. In transmitting, when
a zero is applied at PTR, since V0 is already a zero, tLe
output of the NOR gate will rise and allow the "transmit"
isolator to turn on, and its output will drop the line voltage.
This will turn off the "receive" isolator, but V0 cannot ri~e
because the Open-Collector (O.C.) anti-lock inverter, B, .s
holding V0 low. With the input to inverter A held low,
PTR will go to a high when the zero ceases to be applied.
In general, the principle of anti-lock logic is to have withirl
the station a means to:

(b)

allow the "transmit" isolator to turn on when a

zero is applied at PTR

(c)

make PTR return to a high when the zero ceases to

be applied.

Requirements (a) and (b) are met with O.C. inverters A

followed by OR-type logic (NOR or OR), however, implied
in (b) is a means to hold the input of inverter A at a low
when a low is applied at PTR, and this would then also take
care of requirement (c). Inverter B does this; note that
without inverter B, applying a low at PTR would initiate a
low line which would raise V0 and cause the line to rise
again -- oscillation would result as long as PTR is held low.
The O.C. inverter B works because the isolator output is
also O.C. If it were not, i.e., if the isolator had active
pullup, the anti-lock loop would require a gate, rather than
the "wired-or" situation at the output of the isolator.
Alternatives to the use of inverter B are shown in Figure
3.6.4-4(c) and (d). In (c) the "receive" isolator's output
transistor is held "on" with base current supplied from the
cathode of the "transmit" isolator's input diode. This
reduces component count, but the data rate capability is
reduced because of the capacitance of the input diode being
added at the base. In (d) the "receive" isolator is held "on"
by running the entire line source current through the input
diode.
This reduces component
count still further
(eliminates D3) and has a higher data rate capability than
(c), but the large current forced into the transmit isolator's
input will cause a higher rate of CTR degradation,
especially if the "transmit" mode occurs frequently. Both
(c) and (d) will have a longer delay than (b) in recovering

20mA
CURRENT
SOURCE WITH
4V COMPLIANCE

PTRb

PRa
(al SEPARATE

(bl ANTI-LOCK
LOOP TO ALLOW
INPUT/OUTPUT
TERMINAL

PTa
INPUT/OUTPUT

COMMON

TERMINALS

03 - HP-5082-4l187
02,04
- 1N914
ISOLATORS 5082-4370
PTRd
(dl
Figure 3.6.4-4

Bus Multiplexing with Optoisolators

Separation of Stations.

INPUT-DRIVE
for Ground

ANTI-LOCK

LOOP

from th~ "transmit" mode to the "receive" mode, due to

the higher base current in the isolator output transistor
during t le."transmit" mode.
Other
means of changing from "receive" mode to
"trans
t" mode are discussed in Section 3 .6.6 describ ing
the use f optoisolators with microprocessors.
Current-looping is another form of data transmission wh .ch
also allows multiplex as well as simplex operation. 1he
basic difference between current-loop multiplexing and bus
multiplexing is that in a current loop each transmitt: ng
station transmits by interrupting, rather than shunting, the
line current. Thus, as seen in Figure 3.6.4-5(a) and (b), the
"transmit" isolator is turned "off', rather than "on" when
a zero is applied at PT (or PTR).
When the "transmit" isolator is "off' the terminal volta ~e,
VT, rise to a value, VTL> which depends on how many
stations are in the loop. Worst case would be with only cne
station in the loop, in which case the full supply volt, ge
would appear across the terminals of the "transm t"
isolator's output. If the loop supply voltage exceeds t e
isolator utput voltage rating, a buffer switch can be used.
By using the isolator to bypass the buffer switch b'.se
current, the voltage across the isolator output never exceeds
the base-to-emitter voltage of the buffer switch. Also, wh~n
the "transmit" isolator is "on", loop current may exce ~d
the rating of the "receive" isolator input diode. A shunt
regulator, such as that of Figure 3.6.1-6, can be used to
bypass Lne current in excess of that required by t le
isolator.
Terminal voltage for the high state, VTH limits the number
of stations the loop can accomodate. It is, therefOle,
desirable to make it as low as possible. If separate buff ~r
switch a:ld input-current
regulator are used, the least
possible terminal voltage would be the sum of (V F + Vb ~)
for the regulator and (2V be) for the buffer, and wou.d
require f ur transistors, two of which must carry full lille
current. By combining the buffer/regulator functions as n
Figure 3.6.4-5(c), two transistors can be eliminated.
VTH may be as low as (VF + Vbe + VSAT Q2) "'"2.5V bllt
may also be somewhat higher, depending on the ~ of Q2, as
seen in Figure 3.6.4-5(c). Such dependence on ~ can te
eliminated
by using a darlington in place of Q2.
making VTH "'"3.0V, but not dependent on ().
Notice tLat with the shunt-operated buffer in Figme
3.6.4-5(c), the logic relationship between the input of the
"transmit" isolator and loop is inverted from what it is i1
(a) or (b). Thus, the anti-lock loop of Figure 3.6.4-4(b) cr
(c), but not (d), can be used.

Any apparatus connected to telephone lines must meet two

requirements. The primary requirement is that it must
respond to the desired signal only. The second is that it
must not interfere with the normal function of the line.
Optoisolators are a nearly ideal solution to the second
requirement, and proper circuit design can satisfy the first.

Ring detection requires the circuit to respond to ring signals

only. These can occur over a fairly broad range of
frequencies. Simple peak detection of the ac ring signal is
unsatisfactory because there may be other high-amplitude
voltage variations on the line, such as dialing "spikes".
When a ring occurs, there are several cycles in succession at
a higher repetition rate than dialing spikes. In Figure
3.6.5-1, the input diode of the optoisolator and the
antiparallel diode allow ac current to flow in the coupling
capacitor, C 1, with each half cycle causing a peak current
of about 0.5 mA in the input diode. The resulting current
spikes are integrated by C2 in the output circuit. After a
few cycles of the ring signal, the charge on C2 has changed
far enough to turn on the output transistor. Dial spikes are
ignored by this circuit. Loading of the telephone line
cannot be worse than the 200k established by the series
resistors. At 20 Hz, it is more nearly 450k, and at dc it is
limited only by capacitor leakage.
On/off hook detection requires a means to sense current in
the line, flowing in a particular direction. With no
interference a single optoisolator will do, as in Figure
3.6.5-2(a). If transient common mode current is present, a
pair of optoisolators with OR (or NOR) logic can be used,
as in Figure 3.6.5-2(b). Common mode current in either
direction can only turn on one of the two isolators -- a
differential current is required to turn on both and get an
output. If the common mode current is not transient, but a
persistent dc level, it may prevent response of one side or
the other to differential mode current. In such a situation
the circuit of Figure 3.6.5-2(c)can be used. Here th:
~ommo~ mode current, ICM' decrements the input current
III one lsolator as it increments
the input current in the
other. The isolators should be a type that is approximately
linear and the threshold set at a level requiring the
combined collector current resulting from differential
current, IDM, to obtain an output. This circuit can respond
only to IDM as long as IICMI < IDM. In the LED/resistor
network, the purpose of the resistor is to adjust each
isolator channel to have the same ratio of collector current
to line current. The LED is there to make this ratio nearly
constant, down to very low current levels,without requiring
a large voltage drop across the resistors.

- ,11t

VTH

TOTAL OF 8 STATIONS
IN CLOSED 5 mA LOOP

< 1.7V

VTL = 15V

PTa
la) SEPARATE INPUT/OUTPUT

PTRb

TERMINALS

Ib) ANTI-LOCK LOOP TO ALLOW COMMON

INPUT/OUTPUT TERMINALS

R2 (
VF)
IF+be +(3
R1

= 2V

VOC
-1.2V

R3

Vbe

=
I

+ VF
-

VOC
ILL< -R2

R1
PTRe
Ie) LINE BUFFER/REGULATOR

e.g. FOR IL <;;;150mA, VOC = 65V, (32~ 60

CHOOSE R2 = 10k, R1 = 180Q, R3 = 56Q
and GET VTH "'" 4V
02 - 2N2896

Figure 3.6.4-5

Current-Loop with Optoisolators

Loop Voltage Drops.

ILL

03 - RCPl15

to Accommodate

< 6.5

mA

100 kn

.02 J1F

T~
100 kn

50B;~-4370171
Figure 3.6.5-1

Telephone

If IICMI > IDM, four isolators are required, and all fOllI
must be adjusted to the same ratio of collector current to
line cu"eat, as shown in Figure 3.6.5-2(d). In this schemt:,
ICM of either polarity causes current both into and out of
the collector current summing node, leaving its volta~e
changed
only by the amount of unbalance in the
collector-current-to-line-current
ratio. IDM of eithn
polarity unbalances the output causing the summing node
voltage to move up or down according to the polarity cf
~PM' :,or the on/off hook application, only RU and the
-IDM comparator are needed, but for general applicatior.,
this scheme, by using two load resistors and tW)
comparators can not only detect IDM, but also its polarit).

Ground looping in microprocessors has been blamed for

everything from giving free "games" to blowing the entir'l
circuit. Certainly there are static electricity hazards to
circuits. These can be relieved by using optoisolators to
open the ound loops, with little or no impairment of thl:
system function. The slight additional expense is minuscult:
compared with the cost of troubleshooting.
The haza 'ds usually arise where peripheral hardwan,
(memory, I/O units, etc.) are connected with long bus runs,
but can also exist in some intra-modular situations. Th(
standard
bus interface devices do offer some CMR
protection, but are themselves vulnerable to large ground
loop surges.
An arrangement whereby a peripheral unit can exchange
information with the data bus and yet be isolated is shown
symbolically in Figure 3.6.6-1 and schematic details are in
Figure 3.6.6-2. As the truth table shows, the direction of
data flow is basically controlled by the Receive Enable
(RE) and Transmit Enable (TE) functions. With RE low,
the ~ outputs remain high regardless of what Bn is; with
TE low, the Bn outputs are "open collector" so the T
.
n
mputs cannot control Bn. With both RE and TE low, the
bus side of the interface unit is entirely "open collector";
not only are the outputs "open collector", but the inputs
also present only "open collector" loading, thus making the
current sinking efforts of other bus drivers more effective.
The schematic of Figure 3.6.6-2 shows why.

Ring Detector Circuit.

When RE is high, the QRE transistors are switched on and

are capable of sourcing 2 mA to any of the four isolator
inputs facing the bus side; so whenever B is pulled low (by
.
n.
some other transrmtter on the bus) the Isolator at Bn is
turned on and ~ goes low. Thus ~ = Bn.
When TE is high, four of the 20 mA current sources at Q E
source current to the bus lines, while the fifth energizes iRe
V CC terminals of the four "transmit"
isolators; so
whenever Tn goes low, the corresponding isolator is turned
on and Bn goes low. Thus Bn = T n .
RE and TE are permitted to be simultaneously high, but if
~ is connected to Tn' the line will lock low. If it is
desirable to have Rand
T connected so as to have a
n
n
common receive/transmit terminal on the I/O side, the
lock-low situation can be prevented by using an anti-lock
loop, such as described in Section 3.6.4. A more common
way is to use the logic drivl!!g RE and TE, shown in Figures
3.6.6-1 and 3.6.6-2. When CS (Chip Select) is high, both RE
and TE are low. With CS low, either RE or TE is high,
depending on whether R/T (Receive/Transmit) is high or
low. If R/T is high, only RE can go high, and since TE is
low, a low at Tn cannot be transmitted to B so the line
n'
will not lock low with ~ connected to Tn' If R/T is low,
only TE can go high; Bn will then go low when Tn goes
low, but with RE low, QRE is off and the output transistor
of the "receive" isolator remains off, so again the line will
not lock low if ~ is connected to Tn'
The electrical ports on the bus side of the optoisolator
interface are all compatible with the usual logic levels
required at inputs of those modules not requiring isolation.
The bus-side inputs require that non-isolated (as well as
isolated) drivers sink no more than 2 mA, because that is all
the QRE sources make available, and as little as I mA will
do, depending on what logic current the R output must
sink. At the 2 mA level of input current, the ~n output can
handle TTL.
It is usually not necessary to install optoisolator interfacing
as extensively as Figure 3.6.6-3 suggests, but it can be done.
If only one or two of the modules is likely to suffer (or
cause) difficulty, it is usually necessary only to isolate those

-- --

10M

(a) 1

0M SENSOR - SIMPLE
BUT RESPONDS TO ICM

ICM

(b) NO RESPONSE TO ICM' BUT LARGE ICM CAN

IN

IBIT RESPONSE TO 10M - HIGH SPEED

10M

----

ICM

----

RL1
SUMMING NODE

(d) RESPONSE ONLY

TO IOM EVEN IF
IIOMI

< IICMI

RO

<*

TO~~
R1

<*
<~

T1

~~

%0
R2

BO

B1

DATA BUS

T2~:t
R3

B2

<~

T3~:t

B3

TRUTH TABLE
CS

R/T

RE

TE

RE

TE
CS

EFFECT

Bn

= Tn

Rn

Bn

R/T

=1

Bn:OPEN COL.

Figure 3.6.6-1

Symbolic Representation
Interface ::ircuit.

of Isolated Bus-I/O

+5V
VCC

ILlOL17-

280<;

RO

,,1800

-"
-"

100k

BO

BUFFER
INVERTER
4
5082 -4370171

,------I SAME AS ABOVE

IL _______

B1

,-------

SAME AS ABOVE

ILlOL17-

B2

L______
280<;
17

R3

1800

BUFFER
INVERTER

I/O
01, 02 - 2N3904
ORE, 0TE - MP03546 ARRAYS
BUFFERS AS APPROPRIATE

TO LOGIC FAMILY

Figure 3.6.6-2

Schematic Diagram of Optoisolators

Bus-I/O Interf;lce Circuit.

wired for

w
-0

a:O
Ou
Ow
~O

Figure 3.6.6-3

MicroprocE ssor System; Suggested Use of Isolated

Bus-I/O Inlerface Circuit.

- ..
....

VCC

7I\~~
7t\
7r\

>.1Soo

22k

270

...

VVYl

1..

1 ...
TT

.....

-r

VVV

......

4
(~ C/R

CS

...

~so

=V.

1Soo

..
TT

S20

>--{

!:

YYV

....

TT

1S00

f:J

vV"'l

1 .

R3
I

YTT

~:/
l

~;~J.
JI~

R3

----<>

YT

..

YYT

~v

1Soo

NPN - 2N~904

YTV

PNP - MPCl3546

1S0

TRUTH TABLE

..

~".,,.~ ,

C/R

CONOIl'ION

'.;;UI~

C->II

~,~".J:

C+-Fl

C,R BOTH
OPEN COlL

470

470
C

R2

---<>

CS

470

"'"
.J

..- -[:

02

TTT

G
..

TTY

......

..

[: ,~,.,,.,,

..

TT

s-~

L~
Y

L1J;~:

[~

C3

R2

IC"~3

....

TT

~....

I--

YVVI

C2

R1

J~<h

:~~~

TT

...

R1

YYT

r(.""~~

......

....

..

I---<

[. ~"."."

YVY

RO

1:11'~

2700

jC1

ORO

Pit:/

...

.......... I

(,,~,..o)

.210

TT

7K

VVYl

(,

YYVi

1 ..

RT

....

1';;Ur ;2

22k

~....

RR

J\
7 1\
J

co

TT
2700

'(

OCR

'- Y

SV
Vcc

Q",~~

0-

Locc

particular units. If isolation is used on any of the wil es

connect Ig such a particular unit, then, for best isolatie n,
all wire' to that unit should be isolated. Any un-isolated
wires allow a ground loop closure that can affect the se
which aI ~isolated.
The micoprocessor can also be isolated from the data bus
it is cor rolling, as in Figure 3.6.6-3, top right. The reas)n
the microprocessor appears to be addressing itself is H.at
directio~al control of the flow of data (to or from the bus)
is through inputs on the bus side of the optoisolalor
interface. In most cases, the CS input of the microproces::or
isolator ould be left low and its "address decoder" wot.ld
only decode direction information from appropriate lines in
the address bus. If for any reason, modules are connected
on the bus lines between the microprocessor and the
microprocessor isolator, care should be taken to make slife
their electrical requirements are compatible with the F'n'
Tn te linal characteristics. Such caution is especially
require if the realization in Figure 3.6.6-2 is used because
it is not symmetrical. That is, the Rn outputs do not hc.ve
as large, current sinking capability as the Bn outputs.

A symmetrical interface isolator circuit is shown in Figure

3.6.6-4. Achieving symmetry requires the buffer transistor
(QCn' QRn) to obtain IOmA of isolator input current from
less than ImA of current being sunk at the input (base). If
there is assurance of at least a 10mA current sinking
capability from any other drivers on the line, then QCn
and/or QRn can be omitted, and the cathode of the isolator
input diode would be connected through 270 ohms to the
collector of the output isolator. This change would still
leave an open-collector condition whenever CS is high_
The symmetrical interface isolator is especially useful when
the data bus must serve several units that do not require
isolation as well as some that do require isolation from the
microprocessor but not from each other. The alternative to
using the symmetrical interface isolator would be to use an
interface isolator, such as the one in Figure 3.6.6-2 at each
of the units requiring isolation, and this would require
many more parts than symmetrical isolation.

This section
deals with the bare fundamentals
of
photodiode design and construction and with the basic
characteristics
of PIN photodiodes.
Amplifier
configurations are described for linear and logarithmic
response to optical signals. Also given are circuits and
suggested
applications for utilizing the performance
features of PIN photodiodes.

When a photon is absorbed in a semiconductor,

an
electron-hole pair is formed. Photo current results when the
photon-generated electron-hole pairs are separated, electrons
going to the N side, holes to the P side.
Separation of a photon-generated electron-hole pair is more
likely to occur when the pair is formed in a region of the
semiconductor where there is an electric field (see Figure
4.1.1-1).
The alternative
to separation is for the
electron-hole pair to simply recombine, thereby causing no
charge
displacement
and thus no contribution
to
photocurren t.

the largest possible number of photons to be absorbed in

the depletion region. That is, the photons should not be
absorbed until they have penetrated as far as the depletion
region, and should be absorbed before penetrating beyond
the depletion region.
The depth to which a photon penetrates before it is
absorbed is a function of the photon wavelength. Short
wavelength photons are absorbed near the surface while
those of longer wavelength may penetrate the entire
thickness of the crystal. For this reason, if a phot diode is
to have a broad spectral response (respond well to a broad
spectrum of wavelengths) it should have a very thin P-Iayer
to allow penetration of short wavelength photons, as well as
a thick depletion region to maximize photocurrent from
long wavelength photons.
The thickness of the depletion region depends on the
resistivity of the region to be depleted and on the reverse
bias.
A depletion region exists even if no reverse bias is applied.
This is due to the "built-in" field produced by diffusion of
minority carriers across the junction. Reverse bias aids the
built-in field and expands the depletion region.
The extent of the depletion region at any voltage is larger in
devices made with higher resistivity at the junction; but low
resistivity is required at both surfaces for making ohmic
contact to the device. P-N photodiodes, such as solar cells,
are made with P diffusion into N-type material of low
resistivity. In P-N photodiodes, a thin P diffusio allows
good short-wavelength response, but a relatively high
reverse bias is required to extend the depletion to the depth
required for good long-wavelength response. A deep P
diffusion degrades the short-wavelength response but lowers
the bias required for good response at longer wav,~lengths.
Optimization of both short- and long-wavelength response
at low reverse bias requires a P-I-N, rather than a P-N diode
structure. A PIN diode has a thin P-type diffusi n in the
front and an N-type diffusion into the back of a wafer of
very high resistivity silicon (see Figure 4.1.1-2). The high
resistivity
material between the P-type and N-type
diffusions is called the intrinsic region, or I-layer.

- III'
Figure 4.1.1-1

+
BIAS
SUPPLY

P-N Photodiode Junction;

Effect on Detection.

Diagram of Internal Field

Electric field distribution in a semiconductor diode is not

uniform. In the regions of the P-type diffusion (front) and
N-type diffusion (back) the field is much weaker than it is
in the region between, known as the depletion region. For
best performance, a photo diode should be made to allow

In Hewlett-Packard PIN photodiodes, the I-layer has a

resistivity so high that even at zero bias, the ( epletion
region extends from the P-Iayer to approximately halfway
through the I-layer. With as little as 5 volts reverse bias,
depletion is extended all the way to the N-Iayer; this is
called the "punch-through"
voltage. Since breakdown
voltage is over 200V, it is often desirable to operate at
reverse voltages well above punch-through so as to keep the
I-layer fully depleted even at high flux levels. This insures
best linearity and speed of response.

ROUND DISC
INTERIOR
RESPONSE
REGION

AI
0.5 (20)
Si02 8 (32)

ANNULAR
CIRCLE
EDGE
RESPONSE
REGION

~
5 (2000)

ZERO BIAS
DEPLETION
REGION

1250 (50,000)----1
1,000 (40,000)

----1

j
:::J-,50 (2,000)

Figure 4.1.1-2

P-I-N Photodiode; Isometric Cutaway Distorted to

Clarify Main Features.

As reverse voltage is increased, the depletion region expar ds

sideways, extending beyond the contact rim with as
little as 20 volts applied. This permits photons incident at
the edg l outside the ring to enter and be absorbed in
the depletion region without having to penetrate the
P-layer. Since the silicon dioxide surface passivation is v( ry
transparent, even to very short (UV) wavelengths, the
quantun' efficiency of edge response can actually exceed
unity (sle Figure 4.1.2-1). Quantum efficiency exceeding
unity is possible because the shorter wavelength photc ns
have energies greater than twice the bandgap of silicon.
- - - 11- QUANTUM EFF. - ELECTRON _
PHOTON

Virtually any kind of semiconductor junction exhib its

photoresponse, but most devices are designed and packaged

Figure 4.1.2-1

R _ RESPONSIVITY _ -AMe.
WATT

Spectral Response of Interior and Edge Regions of

HP P-I-N Photodiode.

to prevent radiant flux from interfering with their intended

function. LEDs are packaged to permit radiation of flux
from their junctions, so their junctions are easily exposed
to radiation. Of course, they are not designed for optimal
performance as photodiodes; nevertheless, they perform
well enough to be useful in many applications. Their
spectral response peak is at a wavelength much shorter than
a silicon photodiode spectral peak, so, in the absence of
fIltering, they give a closer approximation to photopic
(human visual) response.

A figure of merit for photodiode response can be described

by the quantum efficiency. Ideally, each photon (or
quantum of energy) should cause a contribution of one
electron
to the stream of photocurrent.
Quantum
efficiency,1)q' is therefore dimensioned as "electrons per
photon".
For most engineering, a more familiar performance
parameter is the flux responsivity, Rck, which takes account
of the photon energy. It is the ratio of photocurrent to spot
flux:

where:

R<j>
= flux responsivity in amps per watt
1)

quantum efficiency in electrons per photon

A = photon wavelength in nanometers

Ip

photocurrent

in amperes

<j>e = radiant flux in watts

In Figure 4.1.2-1 the spectral quantum efficiency is shown

with dashed lines and the responsivity with solid lines.
These same values apply to all Hewlett-Packard PIN
photo diodes, regardless of their size or lens magnification.
This is because responsivity is area-independent,
being
defined for an incrementally small spot. Note that there is a
substantial difference between the responsivity of the
interior region and that of the edge region (see also Figure
4.1.1-2). Actually, the edge response is obtained only when
reverse
bias is applied. Interior response is nearly
independent of reverse bias at wavelengths shorter than
peak; at longer wavelengths interior responsivity does
increase slightly when reverse bias is applied - depending
on how much reverse bias is applied and how high the flux
level is. At reverse bias greater than punch through, the
variation of interior responsivity with reverse bias is nearly
zero for moderate flux level.
Another handy performance parameter is the incidance
response, RE, which takes account of the photosensitive

area (or apparent area, for photodiodes with magnifying

lenses). It is the ratio of photocurrent to incidance:

where:

R E = incidance response in amps per watt

per square millimeter*
Ee = radiant incidance in watts Pllr square
millimeter*
AD = effective photosensitive area in square
millimeters*
1p, R<j>
are defined in equation 4.1.2-1

*The SI units of area are square millimeters and square

meters but square centimeters are still frequently used
in describing incidance.
Incidance response describes the photodiode performance
when it is floodlighted (uniform incidance over the entire
device) so that edge effects are included along with interior
response. The incidance response varies with rever~e bias as
the depletion region spreads out from the P-diffusion area.
Consequently, the description of incidance respon.;e as the
product of responsivity times area is a useful appro.dmation
only. The perimeter-to-area ratio decreases as photo diode
area is increased, so the approximation improves as the area
is increased. With very large reverse bias (>lOOV) t le entire
surface of the chip can be photosensitive - that is, the
depletion region can be extended all the way from the edge
of the P-diffusion to the edge of the chip.
Speed of response also depends on what area of the
photo diode are irradiated and how much reverse bias is
applied, as seen in Figure 4.1.2-2. The speed observed also
depends on the load resistance. Photocurrent begins to flow
just a few picoseconds after flux is applied, but there is
junction capacitance, package capacitance, and stray wiring
capacitance to be charged. The rise/fall time constant
therefore depends largely on load resistance unless the load
resistance is so small that the internal resistance of the
photo diode limits the speed. The internal resistance is
mainly the sheet resistance of the very thin P- iffusion.
With flux applied at the very center of the interior region,
the resulting photocurrent
encounters maximum sheet
resistance and this may be as much as 50 ohms. Flux
applied to the interior at portions closer to the aluminum
contact ring produces photocurrent that encounters a lower
sheet resistance.
Low noise is another benefit resulting from the extremely
high resistivity of the I-layer in Hewlett-Packard PIN
photo diodes. Diode-noise-limited operation can be achieved
over a modulation frequency range extending fr m dc to
more than 10 KHz, as seen in Figure 4.1.2-3. The
horizontal dashed line is the noise calculated from the shot
noise formula:

_________

OPTICAL PULSE,
DURATION = tp ~

Q - SPOILER

R~-~~VCC
Ip AVG

1_

II
E

BYP6,SS CAPACITOR,
LOI/\ INDUCTANCE,
SUCH AS PIERCED
DISC SURROUNDING
PHOTODIODE

/e

C>

AT LEAST 5V
FOR HIGH SPEED

LOAD MATCHED
TO TRANSM ISSION
LINE IMPEDANCE

Ip.PK x tp
VCC

FLOODLIGHTED WITH
ZERO BIAS (Ee = 0)
OR
SPOTLIGHTED EDGE
WITH Ee < 50V
_I

10

20

30

40

t - TIME

L_
FLOODLIGHTED VVITH
2 V < Ee < 50V
Ip,E = RESPONSE FROIVI EDGE
Ip ,I

= RESPONSE FRm" INTERIOR

-J.1S

Ip ,E

~
Ip,1

~-'
0

10

20

I
30

t-TIME-J.1s

EDGE RESPONSE TAILS

DISAPPEAR WHEN:

.-

OR
OR

SPOTLIGHTING
INTERIOR ONLY
USING HIGH BIAS,
Ee> 100V
USING FAST SWEEP
(TAIL IS STILL THERE
BUT ONLY CAUSES
DC SHIFT)

I
2
t - TIME - os

(b) WAVEFORMS C BSERVED IN SPEED TEST

Figure 4.1.2-2

Speed of R<lsponse Observation

Waveforms.

4.4

Apparatus

and

f2, f1 = upper and lower 3 dB frequency in hertz

fN = flicker noise corner in hertz, typically less
than 20 Hz
LIGHT
SIGNALIS

OUTPUT

RL= lOlL

TOSS

Zero bias noise is just the thermal noise in the dynamic

resistance of the photodiode at zero bias. According to the
thermal noise formula:
IN THERM =
,

4kT
ROO
4

VR DO (Gm'
where:

(--!A-)
VHZ

IN/VB = bandwidth normalized noise current in

femtoamps per root hertz
k = Boltzmann's constant
1.38 x 10-23 joules per degree Kelvin
T = absolute temperature,

Modulation-Bandwidth-Normalized
Referred to Amplifier Input.

IN,SHOT
VB
where:

= v'2"'q'"I

dc'

= 17 9VI

(4.1.2-3)

IN/VB = bandwidth normalized noise current in

femtoamps per root hertz
q = electron charge = 1.602 x 10-19 coulombs
Idc = dc current flowing in the photodiode in
nanoamps -- usually taken as total dark
current.

The noise current calculated from equation 4.1.2-3 is

usually greater than the measured noise at frequencies
above the flicker noise comer, fN. This is because the dc
current used in the formula actually consists of two
components: leakage current and junction current. Only
junction current causes full shot nuise; the only noise
arising from leakage current is the thermal noise of the
leakage
resistance
and the flicker noise. The two
components are difficult to distinguish, so a worst case
value is obtained by applying the shot noise formula
(equation 4.1.2-3) to the entire dark current.
In a frequency band extending from fl at the low end to f2
at the high end, the total noise current is:

where:

ROO = photodiode dynamic resistance at zero

bias in gigaohms

Noise Spectrum

(nA) (f A )
dc
VHz

iN (f2,f1) = total noise current in amps

iNO = bandwidth normalized noise current in amps
per root hertz from equation 4.1.2-3

taken as 290 K

The zero bias dynamic resistance is difficult to measure

directly, but can be calculated from:
ROO (G

m = [k T (V)]
q

where:

_1_
IS (nA)

IS = "reverse saturation current", measured as in

Figure 4.2.2-2, in nanoamps
kT

- q "'"0.025V, defined in equations 4.1.2-3 and

4.1.2-5
Inserting Rno from equation 4.2.1-6 into equation 4.2.1-5
leads to the interesting result:
IN,THERM=~
= 25.3 ViS (nAf (fA/y'Hi)
Comparing this with equation 4.1.2-3 suggests that a low
leakage diode would be noisier at zero bias - and this is
true. However, at zero bias there is no flicker noise, so this
mode is usually preferred for low noise operation. Because
there is no flicker noise, the noise in a ba dwid th,
B= (f2-f1)

is that found from equation 4.1.2-4 with fN = O.

When signal flux is applied to a photodiode, the resulting

photocurrent
produces full shot noise. (This sometimes
comes as a surprise when the photodiode is used with
signals
of low modulation
depth.)
Nevertheless,
signal-to-noise
ratio is defmed as the ratio of the
photocurrent when signal is applied to the noise current
when there is no signal (i.e., dark):

The Noise Equivalent Power NEP is defined as the signal

flux levd for which SIN = 1.0 for B = 1 Hz:
A

NEP~---

I~
Rrf>

(fW )

,
I
60

v'HZ'

,,

\
\
\

'ReA
,1P21

50

Because NEP varies inversely as the responsivity, reversal of

the log scale of NEP in Figure 4.1.2-4 gives the log of
relative spectral response.

..,
CD

-110

cr:
w

...
a

Under
floodlight
conditions,
the
modulation-bandwidth-normalized
signal/noise ratio is:

IZ

30

W
...J

<l:

-100

20

:>

0
w
w

!!?

10

The SiN ratio increases as the square root of t 1e

photo diode area, AD' because the signal rises linearly as t1e
area while noise current varies only as the square root of
the area. A figure of merit called Detectivity, [)*
(DEE-STAR) characterizes the area normalized quality or a
photodiode surface by normalizing the SiN ratio with
respect 1:0 the square root of the area and the incidan;e
(flux per unit area). Thus from equation 4.1.2-10:

a
z
-90

-80
500

600

700

800

900

WAVELENGTH- NANOMETERS(nm)

1000

VISIBLE -I---INFRARED-------j

Figure 4.1.2-4

Then from equation 4.1.2-9, the relationship between Il*

and NEP is derived:
0* =

fA

~.JJ...
NEP

Because they both depend on responsivity, which vari,:s

with wavelength;
and on noise, which varies wilh
modulation frequency and bandwidth, NEP and D* a:"e
both
usually
given with
parenthetical
conditions,
e.g.: D* (A, f, f::,f)or NEP (A, f, f::,f).

Optical Spectrum of Noise Equivalent

for HP P-I-N Photodiodes.

Power (NEP)

bias is applied, the photocurrent will cause the anode to

become positive with respect to the cathode. Part of the
photocurrent will flow back through the photodiode, and
part will flow in the load resistance. If the load resistance is
open or extremely high, most of the photocurrent flows in
the forward direction through the diode. This may seem
paradoxical, but such a model does describe the results
obtained in an open circuit.

Operation with zero bias is called the photovoltaic mode

because the photodiode is actually generating the load
voltage. Photovoltaic operation can be either linear or
logari thmic depending upon the value of the load
resistance. Logarithmic operation is obtained if the load
resistance is very high (>10 11 0 hm)LO
s.
mear operatiOn IS
obtained if the load resistance is very low with respect to
the dynamic resistance of the photodiode. The upper limit
of linear zero-bias operation is at VL ~ 100 mV depending
on the precision of the linearity requirement. With higher
values of RV sensitivity can be increased for detecting very
low level signals, but the dynamic range of linear response is
decreased. The maximum practical value of RL ranges from
25MQ for large area photodiodes to 550MQ for the smaller
devices.
00

The simplest model of a photodiode is just an ordinary

diode having in parallel with it a current source as in FigUJe
4.2.1-1.
The magnitude
of this current source s
proportional to the radiant flux being detected by the
photodiode. The polarity of this photocurrent is frOJn.
cathode to anode. It is thus apparent that if zero external

...wI
Z

Ip ----..
~-I\)-----,

r----(

I
I
I

~_V_

~~

I
I
I

VD
'TOT = Ip + 'DARK

= (-ID)

c
I
I-

-20

a::
a::
::>

u
W
o

-40

o
I
-

PHOToeURI~ENT MODE
(PHOTOeO DUeTIVE)

-60

VD - DIODE VOI_TAGE - VOLTS

(b) VOLTAGE/CURRENT
CHARACTERISTICS
OF A
PIN PHOTODIODE (NOTE CHANGE OF VOLTAGE
SCALE THROUGH ZERO)

Figure 4.2.1-1

Electrical e ~aracteristics of HP P-I-N Photodiode.

5082-4207.

+.20
VF -VOLTS

Operation with reverse bias is called the photocurrent or

the photoconductive mode. As compared with photovol :aic
mode, the photocurrent mode offers:
(1)
(2)
(3)
(4)
(5)
(6)

speed logarithmic operation is required, it is best to use the

linear amplifier of Figure 4.2.2-1 followed by a logarithmic
converter.

higher speed
better stability
larger dynamic range
lower temperature coefficient
improved long-wavelength response over the inte 'ior
region
short-wavelength (ultraviolet) response in the eige
region.

The main drawback in photocurrent-mode operation is the

flow of dark current, due to the reverse bias. Dark current
is that which flows when no radiant flux is applied to the
photod:iode. Dark current flowing in RL produces an offset
voltage that varies exponentially with temperature. Then is
also likely to be some flicker noise due to reverse bIas.
Since offset and flicker noise disappear when zero bia; is
used, the photovoltaic mode is usually preferred unless the
application requires one of the six advantages enumerated
above.

For linear operation, the photodiode should be opera :ed

with as small a load resistance as possible. Figure 4.2.l-1
shows
the recommended
amplifier arrangement. The
negative-going input is at virtual ground; the dynanic
resistance seen there by the photo diode is R1 divided by
loop gain. If the op-amp has extremely high input
resistance, loop gain is very nearly the forward gain of lhe
o p-amp , R2 can be omitted if the photocurrent
is
reasonably high - its purpose is only to balance off lhe
effect of offset current. As shown, the output voltage vrill
rise in response to the optical signal. If it is preferable to
have the output drop in response to optical input, then
both the photo diode and Ec should be reversed. Ec may, of
course, 'Je zero. Speed of response is usually limited by the
time constant of R1 with its own capacitance, so it is
improved by using a string of two or more resistors in pIcce
of a single R1.
Logarithmic operation requires the highest possible load
resistance - at least lOCQ. With an FET-input op-amp, this
is easily achieved as in Figure 4.2.2-2. If the offset current
of the amplifier poses a problem, a resistor can be added
between the positive- and negative-going inputs. Its value
should not be less than lOCQ divided by loop gain. If ttle
amplifier has a very high input resistance, loop gain is eq'Jal
to the forward gain of the amplifier divided by (1 + R2/R1)
so making R2 = 0 allows the smallest possiQle resistar ce
between the inputs. The speed of response of this amplifier
will be very low, with a time constant 7 ~ O.1s. If high

Figure 4.2.2-1

Linear Response; Photodiode

Circuit Arrangement.

and Amplifier

PIN photodiodes
are extremely stable, have a zero
temperature
coefficient for A < 800nm, and operate
linearly over 100 dB with <1% distortion. The success
achieved with a servo system such as that in Figure 4.2.3-1
is limited more by mechanical stability of the components
than by the photodiodes. The loop consisting of amplifier
Al with optical feedback from D1 to D2 stabilizes the
intensity Ie2 and makes it linearly proprotional to V REF'
If the beam splitter is stable, then the ratio of Ie3 to Ie2 is

kT

2
VOUT=(1+-)-log
R1
q

Ip

(1+-)
IS

kT "" 25 mV
q

Figure 4.2.2-2

Logarithmic Response; Photodiode

Circuit Arrangement.

and Amplifier

R,=R2=--

VREF
Ip2 + Ip3

Ip2 R1 = VREF
VOUT= Ip3 R3

A1 A2 - CA3130

FOR IF ~ 10 mA. d = 100 mm: Ip2'" 3.2 nA

01 - HEMT-3300

THEN FOR VREF '" 100 mY:

02. 03 - S082-4207

Figure 4.2.3-1

R1 = R2 = R3 = R4 = 33M

Servo-Controlled
Linear Coupler.

Intensity for Densitometer

or

The very high dynamic resistance of PIN photodiodes raises

some interesting amplifier possibilities. Sum-and-difference
amplification, for example, is vastly simplified because the
diode impedance is high enough to isolate the "sum" side
from the "difference" side as in Figure 4.2.3.2. At the
input of amplifier AI' the photocurrents Ip2 and Ip3 are
summed to make the loop gain that stabilizes Ie 1 greater by
6 dB than it is in Figure 4.2.2-1. The expression for the
loop gain for Al is:

IF

A" A2 - CA 3'30

FOR IF

0, - HEMT -3300

THEN IF VREF = 640 mV {~ONE VaE'

10 mA, d

02' 03 - S082-4207

Rl

0"

R4 = 10M, RS = 33n

02 - 2N3904

RS = 10n

constant and VOUT is linearly proportional to V REF' This

remains true even if Dl degrades - the AI' D l' D2 servo
would simply force more current to Dl to compensate the
degradation. In this mode, since ground 2 need not be the
same as ground
1, it is a linear optical coupler
(optoisolator). With Ie3 held stable, the incidance at D3
would have a linear relationship to the transmittance of any
material inserted between the beam splitter and D3, making
a stable optical transmissometer. If material is instead
inserted between the beam splitter and D2, Ip3 would rise
in proportion to the attenuation. Then if, instead of a linear
amplifier, A2 were a log amplifier (Figure 4.2.2-2) V OUT
would be logarithmically related to the attenuation of Ie2,
thus making an optical densitometer.

_ Ip2+ Ip3
ALOOP 1 - (
)

R3 + R4 + R = 400 mV
s I
p2 + Ip3

Figure 4.2.3-2

= 100 mm: Ip2 + Ip3 = 6,4 nA

= R2 = '00 M, R3 = RS = 27 M

Sum-&-Difference Amplifiers Improve Sensitivity

in Obscuration Alarm.

assuming that Al has a forward gain AVI = 105. The extra

6dB loop gain is probably not needed with AVI = I 05, but
as long as loop gain is high, the cathodes of D2 an D3 are
at virtual ground. Consequently, the voltages at the anodes
of D2 and D3 may be as high as 100mV and still provide
linear operation. Thus the resistance from either anode to
ground may be as high as 100 mVj3.2 nA = 31.3MQ.
If A2 is connected as a linear differential amplifier, i would
be wise to keep the anode voltages below 50 mV to allow 0
to 100% excursion of Ip2 or Ip3' The system can then be
used to give linearly the relative transmission of the Ie2 and
Ie3 flux paths for the full range of 0 to 100%
transmittance.
As shown, however, A2 is connected as a very sensitive
obscuration alarm, with R4 adjusting the threshold. If the
obscuration should increase in the Ie3 flux path, or
decrease in the Ie2 flux path, A2 would turn on QI' Should
obscuration of both flux paths increase, LOOP 1 reacts to
raise If' and at approximately 20 mA Q2 will turn on.
Also, if the efficiency of Dl should degrade, Q2 will turn
on, indicating that something is wrong. Notice that if
obscuration increases in flux path Ie3 there will be
proportional increase of the flux in path Ie2 to keep
In + Ip3 constant. Thus the use of sum-and-difference
amplifier provides differential sensing with 6dB higher gain
than would be obtained by simply stabilizing flux path Ie2
and sensing a change in Ie3.

BEFORE CEMENTING
APPLY PATTERN
OF
METALLIZATION
TO UPPER
HYPOTENUSE

~~I

D4rlI

SUGGESTED METALLIZATION

PATTI:RNS:

HONEYCOMB

(a) BE.l\M SPLITTER

CUBE

-[]]==
x-

__

I [

. - '-1

COMPENSATING
EQUAL TO V

OFFSET

llJ==-

~I
(b) SPLITTER-TO-DETECTOR
DISTANCE
FOR PARALLEL BEAM OPERATION

EQUALIZATION

COMPENSATING

OFFSET EQUAL TO

-~~Y/n+tR(1-_1

)-tS(1-_1

REF.

nR

SA~:L1f-a=I--ts--;~I

(e) DISTANCE EQUALIZATION

ACCOUNTING FOR VARIATIO"~
REFRACTIVE INDEX THF:OUGH BOTH OPTICAL PATHS

_____
-

11----

_-

+'--J--- __
u----

II
JI

V/n

I~

Figure 4.2.3-3

Optical Configurations
Obscuration ,~Iarm.

for Use In Densitometer

and

)
nS

OF

Some suggested optical path considerations are shown in

Figure 4.2.3-3 to be used with the circuits of Figures
4.2.3-1,-2. Beam splitting to a very high precision can be
done by applying partial metallization to the hypotenuse
face of one prism, then cementing it to the hypotenuse face
of another, as in Figure 4.2.3-3(a). The repetition interval
of the metallization pattern should be very small to avoid
interference patterns with LED details. To obtain parallel
output beams, the incident beam can be applied parallel to
the hypotenuse face, thus causing the split beams to emerge
parallel (similar to a Dove prism), as shown by the dashed
lines. Separation between the two beams may not be

adequate, however, and the off-axis angle of incidence

introduces "tilted-plate" astigmatism. Alternative means for
obtaining parallel beams are shown in Figure 4.2.3-3(b,c)these require distance compensation for beam offset, which
(a) does not. All three, however, require compensation for
sample length, as in (c).
Figure 4.2.3-3(d) shows how sample cells with curved walls
can be applied with the densitometer or the obscuration
detector circuits to respond to changes of index of
refractior. of clear fluids, as well as relative opacity of
diffuse fluids.

Figure 5.0.1-1

A Collectic n of LE D Display Devices

5:

displays. The GaP transparent substrate LEDs will become

available in more and more products as this technology
advances.
The fU:lction of any display is to attract the attention of an
observn by some recognizable arrangement of obj{ cts,
illustra':ions, images or symbols. An electronic display has
the advantage of easy control of the various light emitting
or light
modulating
elements permitting substar.tial
information transfer across the man-machine interhce.
Electronic displays may be implemented using anyone )f a
large number of different technologies. A few of the IT.ore
commonly known are:

Incandescent
Cathode ray tube
Liquid crystal
Fluorescent (operation
CRT)
Electrochromic
Electrofluoridic
Light emitting diode

is analogous to that of the

Each of these technologies is characterized by distinct

capabilities and requirements with respect to element
density, color, power dissipation, packaging, unit size, drive
voltages, environmental capabilities, cost, availability and
other properties. The choice of anyone display technology
is generally a matter of optimizing some of these proper :ies
at the expense of others.

The LED display technology is relatively new compared to

most of the other technologies listed above. The f.rst
commercial LED products entered the marketplace in
1968. Smce that time, the scope of LED display devices las
expanded rapidly so that it now includes a wide variety of
distinctly different products. Based on use, size, and drive
requirements, these display products may be divided illto
four generalized categories:
1)
2)
3)
4)

Displays with On-Board Integrated Circuits

Strobable-Seven-Stretched Segment Displays
Magnified Monolithic Displays
Dot Matrix Alphanumeric Displays

Figure 5.0.1-1 depicts some of the different display

product~ available. All of these display types have been
originall:r developed using 655 nm red GaAsP type diode
technology. Technologically, it is feasible to manufacture
most of these different display categories utilizing the GaP
transparent substrate colors high-efficiency red, yellow a ld
green. However, at the present time, the availability of
displays utilizing these higher technology LEDs has, for the
most part, been limited to the stretched segment ty oe

The most prominent feature of any display product is the

physical
arrangement of the display elements. This
arrangement, or "font" as it is commonly termed, is
important not only from the standpoint of the type
information which can be transmitted but also important in
that it dictates the nature and complexity of the support
electronics required by the display.
Figure 5.0.2-1 depicts some of the common display fonts.
Fonts A and E are by far the most common in use in LED
technology. Both arrangements have some positive and
negative attributes. The seven segment technique is easy to
utilize from an electrical standpoint, however, it is limited
to displaying numeric and a small range of alphabetic
information. The 5x7 dot matrix can display a wide range
of numeric, alphabetic and other character but involves
some rather expensive electronic circuitry to implement the
other portions of the display subsystem. The sixteen
segment
approach
shown in Figure C has a full
alphanumeric
capability but has had rather limited
acceptance. This font is not capable of displaying some of
the specialized
characters required by sophisticated
systems. Font F illustrates a 9-segment display which has a
somewhat more pleasing font than a 7-segment display. It
can display the same numeric information and has more
alphabetic capability. The dot matrix fonts in Figures Band
D are abbreviated versions of the 35 dot matrix of Figure E
and are used primarily to display only numeric and
hexidecimal information.

The visible portion of a display (be it LED or some other

technology) is only a minor portion of the electronic
system necessary to convert electrically coded data into a
form readily comprehended by a human viewer. Figure
5.0.3-1 is a block diagram of what may be generally termed
a display subsystem. A display user can purchase one, some,
or all of the functions in the block diagram as a single
product. It is necessary, however, that all of the functions
depicted be present in some part of any information
display.
By way of example, a seven segment decoder/driver may be
combined with a data latch and a seven segment diode
matrix to form a display subsystem. Similarly, all four of
the above functions may be purchased in a single package
to perform the same end result. Often, the data latch
depicted in the block diagram will be present as a part of

DODD

I--I
I

o
o

[]

0
0

oeoo

00000
00000
00000

DCCDO
00000

DOCDO
00000

-1

o
DO

00000

[]

occe

[J
[]

OODD

00000

1\//1
-1//\1

CD

DD

00000

o
o

~]

(]

OOCOO

DD
DD

00000

1I I
I 1I

(]

I I

....

.....

....

CODED
DATA

.....

....

.....

DAfA
COI)E TO
DISPLAY
MAfRIX
DECODER

DATA
LATCH
FOR DIGIT
TO BE
DISPLAYED

.....
LATCH
ENABI.E

....

.....

DISPLAY
ELEMENT
DRIVERS

I--

N LINES
ONE PER
DISPLAY
ELEMENT

LED MATRIX

.,.,
.,.,
.,.,.,

.,.,

.,

the data source and does not have to be supplied as a

portion of the display. Price, digit size, viewing conditions,
power
limitations,
system architecture,
and otter
considerations must be taken into account when decidi 19
how to best design and partition a display subsystem.

The natural-partitioning of systems utilzing displays oft m

leads to a situation where the display subsystem is
physically separated from the data source. In these
situation:;, it becomes necessary to transmit coded da ta
from the data source to the display subsystem. Figure
5.0.4-1 cepicts two of the most common techniques f.)r
transmitting four line BCD data plus decimal point status.
The choice of technique usually depends on data source
and display system requirements. For long distances, H.e
cost of the full parallel (character parallel/bit parallel)
approach can become prohibitive. If full parallel data mu:;t
be converted to a character serial/bit parallel data forma:,
the circuit depicted in Figure 5.0.4-2 may be implemented
using the 3 state data bus buffers such as the Nation:
DM8095. A low true signal at the enable input of eaCl
buffer will apply the input data to the data bus. Timing of
the enabh signals will permit either RZ (return to zero) or
NRZ (noa return to zero) data formatting. In the RZ
system, clocking information can be extracted at th.:
receiving lmd and hence no additional transmission line;
need be utilized for clocking.
Numeric Displays with
Circuit (OBIC)

an On-Board

X,
X2
X4
X8

CHARACTER #1

DP

X,
X2
X4
X8
DP

X,
X2
X4
X8

) CHARACTER #2

CHARACTER #3

DP

CHARACTER PARALLEL/BIT PARALLEL

TRANSMISSION (FULL PARALLEL)

DATA

Integratec

For many applications, it is desirable to have the display

subsystem assembled into a compact configuration and
packaged s~parately at some distance from the data source.

CHARACTER SERIAL/BIT
TRANSMISSION.
Figure 5.0.4-1

Two Common

PARALLEL DATA

Techniques

BCD Data Plus Decimal

for Transmitting

4-Line

Point Status Information.

NATIONAL
OM 8095 (61

r-----'
I

ENABLE

The design of the dj.splay system is simp ified and

requires a minimum amount of engineering time.

I
I
23

IN 1

The manufacturing costs and the time required to

assemble each display system on a production basis
are reduced to a minimum.
BCD AND DP
DATA FOR
DIGIT ,

2'

The reliability of the display system is significantly

increased due to the reduction in circuit complexity
and the reduced component count.

I
IOUT5

L
HEX
THREE-STATE
BUFFERS

Less space is required to mount the display system

assembly.
6.
I
IOUT3

BCD AND DP
DATA FOR
DIGIT2

IN 3

2'

I
lOUT'

OP

IOUT5

IN 5

L
r--

----ENABLE

I
I
23

IN'

I
22

IN 2

IN3

2'

I
2

IN'

IOUT5
L

Figure 5.0.4-2

I
J

A Circuit for Character Parallel to Character Serial

Data Conversion.

One possible approach is to have the display PC board

contain all the necessary components that comprise the
display subsystem. However, a more efficient and in many
cases less costly approach is to use a display device that has
an in tegrated circuit containing the subsystem logic
functions, data latch/decoder/drivers, packaged on the same
substrate as the display matrix. Such a device is termed an
on-board integrated circuit (OBIC) display.
The cost per unit of an OBIC display device is more than
that of a device which does not contain an integrated
circuit. Even so, there are some specific advantages to an
OBIC device that off-set this initial cost.
1.

The disadvantage of an OBIC display module is that as

more functions are added, the module becomes unique in
its configuration
and a second source is us ally not
available.
A numeric OBIC display should provide a designer with a
device that operates directly from a 5.0 volt supply, is
compatible with commonly used data source logic such as
TTL, has the electrical characteristic guaranteed over a wide
temperature range, latches and decodes BCD data and
directly drives a large pleasing character that is easily
recognized at a distance.

BCD AND OP
DATA FOR
DIGITN

A pleasing font may be used, such as a modified

4x7 dot matrix, and need not conform to the font
codes available with discrete commercial decoders.

The pin connections between the data source and

the display assembly are reduced due to the
increased
level of integration of the display
subsystem.

The construction of an LED OBIC display typically begins

with a ceramic
substrate
that has printe
circuit
metallization on the face and external leads brazed to the
back surface. The integrated circuit and LEDs are die
attached and wire bonded to the metallization. A hermetic
device has a glass window covering the face of the display
with a hermetic seal at the substrate rim wall-to-glass
interface, as illustrated in Figure 5.1-1. A plastic device uses
a coating of silicone gell to protect the wire bonds and
reduce thermal stress to a minimum, then the substrate
assembly is encapsulated in tinted undiffused epoxy that
acts as an integral contrast fIlter and forms the display
package.
As mentioned earlier, OBIC displays are usually unique in
their design, thus the assembly technique described above
will necessarily be different depending upon the package,
character size and font, available functions incorporated
within the integrated circuit, pin out arrangement and lead
spacing. For this reason, Section 5.1.1 discusses a particular
OBIC LED display family that incorporates the features
described above.

TRUTH

The lIP 5082-7300 OBIC family offers a series of plastic or

hermetk devices that display either decimal or hexidecimal
numeric
information.
The plastic devices are epcxy
encapsulated and are designed for use in consumer md
commercial
equipment,
such as: control units for
household
appliances,
electronic office equipment,
computers and electronic measuring instruments.
The
hermetie devices, available in either industrial or military
grade packages, are designed for use in high reliability
applications involving the possible exposure to an adverse
environment,
such as: the instrument
panels of
transportation vehicles, controllers in numerical cont 01
milling machines, chemical processing equipment, milit;.ry
and aerospace systems. The construction of the herme tic
OBIC di:>playis pictured in Figure 5.1-1.

BCD INPUT
X4

X2

X1

H
L
L

(Blank)

(Blank)

OVERRANGE

..
..

...

...

... ,

...

...

..

H
r-----H
H

L
H
-H
-H
H

2.

--

'.;'-'--

(Blank)

(Blank)

--

...
...
..
-----':::....--

LEOs ARRANC ED IN
A MODIFIED 4.7 DOT
MA"RIX FONl

1. H = LOGIC HIGH; L = LOGIC LOW. WITH THE ENABLE

INPUT AT LOGIC HIGH CHANGES IN BCD INPUT LOGIC
LEVELS HAVE NO EFFECT UPON DISPLAY MEMORY
OR DISPLAYED CHARACTER.

DARK SURFACE
PROVIDI.S POSITIVE
ON/OFF ';ONTRAST
HERMETIC SE,IL AT
SUBSTRATE R M
WALLTOGLA ;S
INTERFACE

EXTERNAL LEADS,
BRAZE[' TO BACK
OF SUB~iTRATE

HEXIDECIMA

DECIMAL
NUMERIC

Xs

INTEGRATED CIRC UIT

CONTAINING BCD DATA LATCH/
DECODER/LED DRIVERS

TABLE

CERAMIC SUB~TRATE

The decimal devices are available with either right hand or

left hand decimal point and display the numeric characters
0-9, and a minus sign (-). The hexidecimal devices displ2y
the characters 0-9 and A-F, and incorporate a blankir g
feature that allows the display to be blanked (turned of:)
without affecting the information stored in the data latch.
A companion plus/minus one overrange display with rigl.t
hand decimal point (1.) is available in either of the
hermtic or plastic packages. The character font is a 7 Amm
(.29 inch) high modified 4x7 dot matrix as shown in Figure
5.1.1.1-1. The advantage of this font is that shapei
characters are formed, so that a "3" does not resemble a
reversed "E", a "B" looks different than an "8" and a "D '
can be distinguished from a "0". The result is a font that
forms characters which are easily recognized by an observer
standing at a distance of 4.6 meters (15 feet).

2. TEST PATTERN

FOR DECIMAL

NUMERIC

DEVICES.

Figure 5.1.1.1-1 Truth Table and Character Font for an OBIC LED
Display.

The integrated circuit is composed of a 5-bit data latch for

the decimal display (4-bit latch for the hexidecimal display)
a decoder and LED matrix drivers. The block diagram for
the integrated circuit is shown in Figure 5.1.1.2-1. On the
hexidecimal devices, the decimal point is replaced by the
blanking function. The data latch accepts logic high true
BCD information and logic low true decimal point status.
Information is loaded into the data latch when the enable
input is at logic low. This information is then latched when
the enable input is returned to a logic high. Changes at the
data inputs will then have no effect upon the display
memory or the displayed character. The blanking input
controls only the LED drivers and has no effect upon the
display memory. When the blanking input is at logic low,
the display is on (character is illuminated) and at logic high,
the display is blanked (character is not illuminated). A
summary of the logic input functions is given in the truth
table of Figure 5.1.1.1-1.

Equivalent input and LED driver circuits are shown in

Figure 5.1.1.2-2. Each data input represents one TTL load.
The display is blanked when a minimum threshold level of
3.5 volts is applied to the blanking input. This may be
achieved by using an open collector TTL gate and a pull-up
resistor. For example, 0/6) 7416 hexinverter/buffer/driver
and a 120n pull-up resistor will provide sufficient drive to
blank 12 displays. The size of the pull-up resistor required
to blank a quantity of N-digits may be determined from the
following formula:

PIN~
VCC

ENABLE

8
BCD
LOGIC
INPUTS

Xl

l_X2
2

X4

3_X8

When the hexidecimal display is blanked, the power

dissipation is reduced to 66% of that power dissip ted when
the maximum number of LED drivers are operating
(displaying the Figure "B"). As a result of this, a significant
power saving can be achieved by strobing the blanking
input of each display. The strobing rate should be a
minimum of lOO Hz or 5x the expected frequency of any
mechanical vibration of the display in order to avoid
flicker. Also, the time average intensity of the hexidecimal

LED
MATRIX
DRIVER

GROUND

6~

Pin 4 is the decimal point status input on

the decimal numeric devices and the
blanking

Figure 5.1.1.2-1

input on the hexidecimal

devices.

Block Diagram of the Four Subsystem

Incorporated into an OBIC Display.

Components

Vec

BCD, DP
AND
LATCH
INPUTS

Vec

LATCHED

DATA

TO MATRIX
DECODER
RL
BLANKING
CONTROL
INPUT

VCC

= ~

Figure 5.1.1.2-2

Equivalent Inp t and LED Driver Circuits for the

Decimal, Hexicecimal and Overrange Displays.

+1 LED

display can be adjusted by blanking to comply w.th

changing ambient lighting conditions. This is done by using
pulse width modulation of the blanking control to vary 1he
display on-time.

0 0

I
I
0 0
I
:
000

Constant current sources are used to drive the LED matrix.

Each constant current driver is connected to either a sin.~e
LED or two LEDs in series; the pattern is shown in Figure
5.1.1.2-3. A minimum VCC of 4.5 volts is required to
maintair. sufficient compliance within the constant current
drivers 10 produce an acceptably illuminated character. As
VCC falls below 4.5 volts, those drivers with two LEDs in
series no longer have sufficient compliance to sOUlce
current through two LED forward voltage drops, while
those drivers servicing a single LED will still have sufficient
compliance to source current through one LED forward
voltage drop. The consequence of this may be insufficient
illuminal ion of the series diodes.
A real benefit to a designer using an aBlc display is t1e
fact tha,: the electrical parameters are guaranteed on t1e
data sheet over a wide temperature range. This allows for
worst ca,e designing using known minimum and maximum
parameter values at the temperature extremes. In operati 19
a plastic aBIC device, it is the package temperature that is
of concern. Therefore, the electrical parameters for t le
plastic aBlC devices are guaranteed over a case temperature
range from -20C to +85C. The hermetic aBIC devices are
characterized
with respect to the operating ambient
tern per a t ure, with the electrical parameters for tjle
industrial devices guaranteed over the ambient temperature
range from OC to +70C, the same as for standard 7400
series n~L. The military grade devices are characterizl:d
over the full operating temperature range, guaranteeing :0
the designer the electrical parameters from _55C :0
+100C.

0= Seven constant

current sources driving one LED.

= Seven constant current sources driving two LEOs in

series. Maximum power dissipation for the decimal

numeric device occurs with nine LED drivers
operating, when numeral 5 and dp are displayed.
Maximum power dissipation for the hexidecimal
device occurs with ten LED drivers operating, when
digit B is displayed.

As is the case with the design of any data transmission

system, a designer should take into account the timing
requirements of both the data source and the aBIC display
in order to insure that the correct information is latched
and displayed. The tinling requirements for this series ,)f
aBIC di:;plays is given in Figure 5.1.1.2-4. These timing
requirements are valid over the full operating temperatu:e
range for either a plastic or hermetic device. Enable ri;e
times greater than 200 nsec may result in the latching of
erroneous information. Data may be clocked into the:;e
aBIC displays at data rates up to 10 MHz.

DESCRIPTION

Enable

Pulse Width

Time data must be held before positive transition

of enable line

Time data must be held after positive transition

of enable line

As indicated above, it is necessary to control the packa~:e

temperature of a plastic aBIC device, as measured at He
top of display pin number 3. For either a plastic (If
hermetic device, the primary thermal path for pOWI,r

Enable

pulse rise time

MAX.

SYMBOL

MIN.

'w

100

"see

'SETUP

50

osee

'HOLD

50

osee

'TLH

200

UNIT

"see

dissipation is through the device leads. The thermal

resistance
junction-to-Iead
is 15C/W. The maximum
allowed junction temperature for a plastic OBIC device is
100C. Therefore, a designer should establish the thermal
resistance to ambient of the display mounting structure in
order to determine if the plastic OBIC devices may be
operated in the maximum expected ambient temperature
without heat sinking. As an example, a mounting structure
consisting of DIP sockets that have been soldered onto a
printed circuit board which has sufficient metallization to
give a combined thermal resistance to ambient of 25C/W
per package, will permit the operation of plastic OBIC
displays in an ambient temperature of 60C without the use
of external heat sinking. A mounting structure that has a
thermal resistance to ambient of 35C/W per package is
sufficient to allow the operation of the hermetic OBIC
displays in ambients up to + 100C, without the need of
external heat sinking.
5.1.2

Intensity Control for Hexidecimal

Pulse Width Modulation

Displays Using

The optimal design operates a display at a light level that,

when compared to the ambient or background, is bright
enough to be seen without being so bright as to cause eye
strain
and fatigue. The fatigue becomes particularly
noticeable when an operator must view an overbright
display for long periods of time. An overbright display may
be defmed as one whose light level exceeds the optimal
value by more than 10 to 1. If the ambient light conditions
remain relatively constant, the selection of a good operating
light level is straight forward. However, there are
applications where the ambient can change by as much as
100: 1, such as in the instrument panels of transportation
vehicles, portable instrumentation, aircraft control tower or
shipboard applications. For widely changing conditions
such as these, the circuit designer may wish to add a
dimming feature. Pulse width modulation can be used to
control the percent time that a display is on (i.e., its "duty
cycle"). By varying this duty cycle, the average light level
can be varied over a wide range. Since the drive level is not
changed (only the duty cycle); close matching among
segments and digits is maintained even at very low light
levels.
It is convenient to have the display intensity automatically
track the ambient. In Figure 5.1.2-1 are two methods for
pulse width modulating the displays blanking input. If a
system clock is available, a photoresistor, R , can be used
x
to control the duty cycle of a monostable multivibrator.
The duty cycle varies from 80% in a bright ambient to 13%
in a dim ambient.
If a system clock is not available, an alternative method
employs a 555 timer as a free running oscillator. The
photoresistor, ~, controls the output pulse width resulting

in a 90% duty factor in a bright ambient and a 2% duty

factor in a dim ambient.
The frequency at which the blanking input is strobed to
turn the display off and on should be fast enough,
regardless of pulse width, to insure an observer sees a
continuous, flicker free, display. This minimum blanking
frequency should be at least 100 Hz or 5x the vibration
frequency, if mechanical vibration of the display is
expected.
5.1.3

Interfacing a Microprocessor
Display

to an OBIC Numeric

An OBIC decimal or hexidecimal display can interface very

easily to a microprocessor system. Since the OBIC numeric
display contains a latch, the only extra circuitry that is
required to interface to a microprocessor is some external
gating. An eight bit data word can be configured as two
BCD or hexidecimal characters. This allows the desired
display information to be stored compactly in a P AM and
easily manipulated by the microprocessor. Since the OBIC
display is dc driven, the microprocessor needs to update the
display only when the information needs to be changed.
Thus, the OBIC display requires only a minimum amount
of microprocessor time. If hexidecimal information is to be
displayed, an HP 5082-7340 display or equivalent should be
specified. This display has a blanking input which does not
affect the latch. When decimal information is to be
displayed, an HP 5082-7300 display or equivalent should be
specified. This display has a latched decimal point and has
the capability to display a minus sign, blank, or lamp test.
If a moveable decimal point on an OBIC decimal display or
selective digit blanking on an OBIC hexidecimal display is
required, some additional circuitry may be necessary. Since
the OBIC hexidecimal display does not have a latched
blanking control, an external latch is needed. The eight bit
data word can be configured as one character (four bits)
plus one additional bit for a decimal point or blanking
control. An example of this technique is shown in Figure
5.1.3-1. Each digit would be addressed by a different eight
bit address code. With the Intel 8080A microprocessor, an
output instruction is used to load the contents of the
accumulator into the specified OBIC numeric di play. In
this example, the lowest order four bits (D3, D2, D1, DO)
are used for the numeric information and D4 is used for
blanking control or decimal point control. When it is
desired to pack two hexidecimal or BCD characters in a
single eight bit word, a couple of bits from the address bus
can be decoded as decimal point or blanking information.
An example of this technique is shown in Figure 5.1.3-2.
Each pair of digits is addressed by four possible address
codes: n, n+l, n+2, or n+3. The two lowest order address
bus lines, Al and Ao are decoded as decimal point or
blanking information. The second byte of the output

microprocessor. At the beginning of each machine cycle,

the 8080A microprocessor strobes eight bits of information
into a status latch. These eight bits uniquely specify which
one of ten machine cycles will be executed by the
microprocessor. The sixteen address lines specify where the
information is to be read or written. The eight data lines
hold
the contents
of that
information.
If the
microprocessor is operated in a single step mode, then after
the end of each machine cycle, one of the ten LED lamps
will indicate which machine cycle was last executed and the
hexidecimal displays will indicate what was on the address
and data buses during that machine cycle. In this example,
a "Fetch" machine cycle was just executed and a JMP
instruction was read from (OOAF)16' Other microprocessor
systems can be debugged in a similar fashion, by using the
appropriate control signals from the microprocessor to load
the contents of the address and data buses into OBIC
hexidecimal displays and decoding the other control signals
to indicate what type of machine cycle was executed.

instruc1ion will determine which pair of characters is te be

updated and whether either digit is to be blanked or have
its decimal point turned on. When the decimal point inJ:uts
or the blanking inputs do not have to be controlled by the
microprocessor, only a single output address code would
uniquely specify each pair of digits. This can sim{ Ify
programming.
An OBIC hexidecimal display can also be used to simp:ify
microprocessor
program
debugging.
Hexidecimal
information
is much easier to decode than binuy
information by the software designer. Figure 5.1.3-3 she ws
a complete
debugging system for an Intel 8080A

vcc

4
RESETOUTPUT

SIGNETICS
1N914
R

555 TIMER

DISCHARGE

~~~:~~NDUCTo:r:1

C1
.01 J.lF

DISPLAY BLANKING INPUT

VB = LOGIC HIGH. DISPLAY IS OFF
VB = LOGIC LOW. DISPLAY IS ON
PHOTOCONDUCTOR
CLAIREX CL5P5L

..

5V

"
.01.tff

11

SCANNI G
CLOCK
INPUT
("" 1 KHz)

1
2
3
4

\\

')

'-~

'~5V
(

:::::~;oloB:::~

)@1rZ\,t1

R = 120n ~
FOR
)
13 N< 12

TIMING
OUTPUTS

Qf~
7416

FAIRCHILD 9601 MONOSTABLE

MUL TIVIBRATOR

/,
/

7 16
T4
VCCGND B (1)

X8X4X2X1E
3 2 1 8 5

"

T
7 16

T4
(2)

7 16

(N)

HP 5082-7359
H EXIDECIMAL
DI SPLAYS

3 2 1 85

3 2 1 8 5

r--

BCD DATA
SOURCE
LINES
I

HP 5082-7340 USED AS HEXIDECIMAL

MICROPROCESSOR

ADDRESS CODE
FOR OUTPUT (n)
"OUT" FROM 8080A
STATUS LATCH

r:

UTPUT FOR INTEL 8080A

HP 5082-7340

DO

X1

D1
D2
D3

X2
X4
X8
E
B

-l

%7474

I
I

I
I

D4 = 0

I
I
DISPLAY (D3 D2 D1 DO)

D4 = 1

BLANK DISPLAY

I
I

L_ - - - - "' O~~IONAL CI RCUITRY

D3
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1

D2
0
0
0
0
1
1
1
1
0
0
0
0
1
1
1
1

D1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1

DO DISPLAY
0
0
1
!
0
l~
~)
1
.J
t,!
0
C'
.J
1
b
0
"f
1
,.,
f:)
0
:3
1
;:'t
0
b
1
_.
I..
0
n
1
t.
0
F'
1

D3
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1

D2
0
0
0
0
1
1
1
1
0
0
0
0
1
1
1
1

D1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1

DO
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1

THAT CAN BE USED

TO BLANK DISPLAY

HP 5082-7300 USED AS DECIMAL OU PUT FOR INTEL 8080A

MICROPROCESSOR
HP 5082-7300

DO
TO DATA BUS

D1
D2
D3
D4

X1
X2
X4
X8
Dp
E

ADDRESS CODE
FOR OUTPUT (n)
"OUT" FROM 8080A
STATUS LATCH

DECIMAL POINT ON
DECIM L POINT OFF

Figure 5.1.3-1

Interfacing Obic r umeric Displays to a Microprocessor.

Five Bit Word SPI cities BCD Data and Blanking/DP.

DISPLAY
;j

c:
.'~

t1

:;
b

,:;
.J

8
(BLANK)
(BLANK)
.....
(BLANK)
(BLANK)

HP 5082-7340 USED tIS NUMBERIC/HEXIDECIMAL

OUTPUT FOR INTEl. 8080A MICROPROCESSOR

TO DATA BUS

HP 5082-7340
X,
X2
X4
X8
E
B

D4
~:
D7

ADDRESS CODE
FOR 0 TPUT (n)
"OUT" FROM 8080A
STATUS LATCH

HP 5082-7340

en

;:)

en

....

c
0
....

CO
D,

D2
D3

X,
X2
X4
X8
E
B

-------,
%7477

I
OPTIONAL CIRCUITRy-j
THAT CAN BE USED
TO BLANK DISPLAY

(n
(n
(n
(n

+ 0)
+ ')
+ 2)
+ 3)

Y47477
AO

[)

I
I

('

OUTPUT
OUTPUT
OUTPUT
OUTPUT

I
Q

I
I
.J

DISPLAY RESULT
BLANK LEAST SIGNIFICAI\lT DIGIT ONLY
BLANK MOST SIGNIFICANT DIGIT ONLY
BLANK DISPLAY

HP 5082-7300 USED AS A NUMERIC DISPLAY FOR INTEL 8080A MICROPROCESSOR WITH Dp,
MINUS, BLANKING, AND LAMP TEST ABILITY
HP 50827300

ADDRESS CODE
FOR OU PUT (n)
"OUT" FROM 8080A
STATUS LATCH

OUTPUT (1 + 0)
OUTPUT (~,+')
OUTPUT (11 + 2)
OUTPUT (11 + 3)

Figure 5.1.3-2

HP 5082-7300

X,

X,

X2
X4

X2
X4

X8

Xa

Dp
E

Dp
E

TURN ON Dp ON BOTH DIGITS

TURN ON Dp ON LEAST SIGNIFICANT DIGIT
TURN ON Dp ON MOST SIGNIFICANT DIGIT
Dp OFF ON BOTH DIGITS

Two BCD or Hexidecimal Characters can be Stored in

an Eight Bit Word. Address Bus is Decoded as
Blanking/DP.

HP50S2-7340

TO
ADDRESS
BUS

AI2
AI3

XI

AI4

AI5

HP50S2-7340

HP 50S2 -7340
AS
Ag

000
X20
0
0
0
0
0
4
o
0
X
000

000
0
0
0
0

A10

All

A4
A5

0
0
0
0

XI

AI
AI

0000
0
X20

A2

x4

A3

Xs

000
0
X2 0
0
0
0000
4 0
0
X 0
0
0
E 0

A6
A7

000

HP50S2-7340

SYNC
FROM
8080A
ADDRESS

BUS

HP 50S2-7340

0,

TO
DATA
BUS

000
X2 0
0
X4 0
0
X
000

0 5-

..-

FROM
8080A

HP 50S2 -7340

XI

00-

XI

0,-

0
000
X4
0
0
X 0000

O2-

03-

.--

X2

DATA BUS

DOD1D2D5D7-

Wcc

II

32x8
PROM
74188
OR
EOUIV.

I .

"FETCH"

,1

N "MEMORY READ"

,II

...,. "MEMORY WRITE"

I~

OUTPUTS
FROM
STATUS
LATCH

"STACK READ"
I~

"STACK WRITE"
,I~

01- ..,

~-7;
D3

~
,,,"

1
J
1

ILJ ~I
D6

"HALT ACO"

L_ ~J

MACHINE CYCLE

I~

"OUTPUT WRITE"

If.

'L..::.j

"INTACO IN
HALT"

./~

1:::::"1
"\...1

D4

"INTERRUPT
ACO"

I ..

"INPUT READ"

TYPE OF MACHINE CYCLE

STATUS WORD CHART

.~

"

cS:J ~

~~
'l>
"1"

i::~
~~"I"

f::'
~~ :"o.Y:o....V"

""

""

~<i. ~

!<' ..
t;' ~

.~
~.. /
"'~

YR

f!"'~~

!<;

~~

DO
D,
D2
D,
04
D~
D6
D7

INTA
WO
STACK
HLTA
OUT
M,
INP
MEMR

e 5.1.3-3

2
0
1
0
0
0
0
0
1

3
0
0
0
0
0
0
0
0

4
0
1
1
0
0
0
0
1

ON

"" ~~ <t ,~
~ ~ <t v!:;
"1"

,,'Ill"

~....,

#'

~
5
0
0
1
0
0
0
0
0

6
0
1
0
0
0
0
1
0

7
0
0
0
0
1
0
0
0

1..,,'Iil"

e:."

1
0
1
0
0
0
1
0
1

~ et i' i' ~
..

tj-

f::' ~.:f$(f'

8
1
1
0
0
0
1
0
0

"I"~V;

-. ~
~~

~"I"#,~

~
9
0
1
0
1
0
0
0
1

10
1
1
0
1
0
1
0
0

Use of LED Lar PS and OBle Numeric Displays

To Aid In Debu ging A Microprocessor System.

XI

000

000

The fIrst seven segment LED displays to enter he

marketplace were generally single digit devices with a SIT all
charact(:r height constructed
on a thick-fIlm ceranic
substrate with round pins, either swaged or epoxyed into
holes in the ceramic. Large rectangular slivers of GaP-sP
were either epoxy or eutectic die attached to the substrate
in the seven segment format. Standard Ie bonding
techniques were then used to connect the anodes of the
diodes to other traces on the substrate. The entire assemlly
was then encapsulated in a clear epoxy to provide
protectiiJn for the bonds and chips. This constructi)n
technique requires large amounts of relatively expensive
GaAsP, making the cost of the fmal product substantia:ly
higher than other display technologies. In addition, poor
alignment of the diodes, visible metallization patterns both
on the diode and on the substrate, and other problems,
made most of these larger digits less than desirable in
appearar.ce by comparison to other designs.
Recently, a new stretched segment assembly technique has
been de\eloped which combines advantages of substantial y
lower manufacturing costs with a markedly improwd
display appearance. In this stretched segment desipl
technique, a lead frame is utilized as the mechanical
support for and makes electrical contact with the LED
chips. Instead of encapsulating the assembly as before, a
cone shaped reflecting cavity is cast inside a rectangular
package above each LED, using glass filled epoxy. The tcp
of this cavity forms the stretched segment. light from an
LED is uniformly emitted from the stretched segment
resulting in a pleasing uniformly lighted rectangular area.
Utilizing this technique, characters of larger size in the ne'v
high-effIciency red, yellow and green colors, as well ~s
standard red, are assembled using LED chips an order (f
magnitude smaller than was used in the earlier design. Thu~"
the stretched segment display offers a variety of colon,
sizes, improved appearance, good on/off contrast, and
improved reliability at lower cost.

The large digit LED display devices are manufactured using

the concept of stretching the light from an LED by
diffusion
and reflection.
This is accomplished by
encapsulating the LED into a special cavity. This cavity is a
rectangular cone with the small end down at the LED, all(.
the large top end forming a segment of the display as showr
in Figure 5.2.1-1. The area of the top end surface of thf
cavity may be 30 times or more the surface area of thf
LED. The emitted light from the LED is diffused as i1
passes through to the top surface of the cavity producing an
evenly lighted segment. Optically, this segment will appeal
as an area source with a near lanlbertian radiation pattern.

DIGIT SEGMENT
FORMED BY
DIFFUSING
CAVITY

UNIFORMLY
LIGHTED
SEGMENT

DISPLAY
PACKAGE

Figure
5.2.1.1-1
is a cross
section
of a
strobable-seven-stretched-segment
display. The Lexan
housing, called a "scrambler", forms the display package
and contains the segment cavities. The sides of the cavities
are made to have as near perfect reflection as possible to
reduce light loss. The external top and side surfaces of the
scrambler are coated with an epoxy paint to match the
color of the LED. The colored scrambler helps provide
good segment on/off contrast.
The lead frame base metal is covered with a 50 microinch
copper flash and a 200 microinch silver plating. The pin size
and spacing match that of a standard 14 pin DIP.
SEGMENT DIFFUSING CAVITY,
GLASS FILLED,
TINTED EPOXY

LEXAN PLASTIC
PACKAGE
(SCRAMBLER)

J
LED CHIP
(CONDUCTIVE
EPOXY DIE
ATTACH)
LEAD FRAME
SILVER PLATED
ANODE LEAD
Figure 5.2.1.1-1

CATHODE LEAD

Cross-Sectional Diagram of One Segment in a

Stretched Segment Display.

The construction of a stretched segment display is a

follows: the LED chip is die attached to the cathode pin of
the lead frame with an electrically conductive epoxy. The
top contact of the LED is wire bonded to the anode pin; a
ball bond is formed on the LED top contact and a wedge
bond is formed on the lead frame.
The lead frame is inserted into the scrambler and the
scrambler is then filled with a glass filled tinted epoxy. This
glass filled epoxy forms the rectangular cone segments, the
glass acting as the light diffusing agent. The tinting in the
epoxy works in conjunction with the colored scrambler to
enhance segment on/off contrast.

The data sheet for a stretched segment display is divided

into the following basic sections:

Package dimensions and circuit diagrams

Absolute maximum ratings
Electrical/optical characteristics
Operational curves
Operational considerations

The data s eet is configured to give a display designer as

much helpful information as possible. The operational
considerations gives application information covering such
topics as determining V p, selecting a set of strobed
operating conditions, calculation of time average luminous
intensity, contrast enhancement suggestions and post solder
cleaning suggestions.

The process of establishing display drive conditions involves

the examination of LED junction temperature, based on
derated operating currents and device thermal resistance,
and the calculation of the LED forward voltage. If the
display
is strobed,
the process also includes the
determination of refresh rate, duty factor, pulse width and
peak current.
5.2.2.1 Maximum

The electrical/optical characteristics list typical values for

various device parameters at an ambient temperature of TA
= 25C.
The operati nal curves are used to determine display drive
conditions.

Temperature

The end of operating life of an LED is defined to be when

the light output has degraded to 50% of its initial value.
The maximum dc current is established from reliability
testing results which predict an operating life of lOOK
hours when the LED is continuously operated at that dc
current in an ambient temperature of 25C.
To keep within the package temperature limitations, the
maximum junction temperature for stretched segment
displays is TJ max = 100C. This value of TJ is based on a
o
thermal resistance junction-to-ambient
of 8 JA = 100 C/W
per package for a device soldered into a typical PC board.
One exception to this is the .3 inch common cathode
display. The thermal resistance per package is 110 C/W and
the TJ max = 105C.
The junction temperature
segment are illuminated.

The absolute maximum ratings give those limits beyond

which the device should not be operated. The device is
capable of being operating at the maximum ratings,
however, for best reliability it is suggested that the device
be operated below the maximum ratings.
The maximum dc forward current and dc power dissipation
are the maximum dc drive conditions, not temperature
derated, and are used in determining maximum strobed
operating conditions. These limits have been set based on a
projected operating life up to lOOK hours. The maximum
temperature limits are based on the display package
capability.

DC Current (I DC)' Junction

(T J) and Package Thermal Resistance (8 JA)

is calculated assuming all eight

TA+LlTJ
8 JA (8 SEGMENTS)
8JA (8) [IAVG

(PSEGMENT)

(VTURN-ON

+ IpEAK

RS)]

Example: The maximum average LED junction temperature

of a high-efficiency red device, that has a maximum
dynamic resistance of 33Q, being strobed at a refresh rate
of 1 kHz with a peak current of 60 mA and an average
current of 12 mA in an ambient of 50C is 84C.

TJ

= 50C
+ (100C/W)

TJ = 84C

(8) [.012A (1.55V + .060A (33m)]

5.2.2.2 Forward Voltage (V F) and LED Dynamic

Resistance (Rs)

Selies

The maximum forward voltage for a given peak forward

current may be calculated as follows:

When calculating LED power dissipation and junctJOn

temperz.ture, the variation in forward voltage with forward
current must be taken into account. Each data sh'let
contains a graph of the typical variation of V F with reSptlct
to IF (the familiar diode curve). A linear approximation can
be deri\ied from this curve to form an equivalent circuit :'or
an LED, see Figure 5.2.2.2-1.

The typical forward voltage is more accurately calculated

by measuring the typical Rs value from the VF point at IF
=5mA.

Where V 5 mA and Rs are obtained

table:

Device

SLOPE

from the following

V5mA

RS TYP

1.65V

21n

Yellow

1.75V

25n

Green

1.85V

19n

High-Efficiency

Red

Example: For a high-efficiency red device operating at 60

mA peak, V F = 2.81 volts:
VTURN-ON

V F TYP = 1.65V + 21 n (,060-.005A) = 2.81 V

-t>l-IIII-V\A-The typical forward voltage for an LED may be scaled off

of IF vs. VF curve or calculated. For example, the forwa"d
voltage of a standard red display may be calculated fro n
the following formula, Vturn-on = 1.55 volts:

MAJORITY
OF LED'S

80
~

70

VF = 1.55V + IpEAK RS
where:

RS TYP = 3n
RS MAX =7n

The forward voltage characteristics of a GaP transparent

substrate device varies considerably from lot to lot. This IS
due to the wide range of resistivity throughout the GaP
crystal and the variation in conductance of the reflecti\e
back contact. Figure 5.2.2.2-2 shows the measured forwad
voltage variation for the high-efficiency red display. FroIO
this kind of examination, the following table is derived:

Device

RSM1N

High-Efficiency
Red

17n

21n

33n

1.55V

Yellow

15n

25n

37n

1.60V

Green

12n

19n

29n

1.75V

RS TYP RS MAX

VTURN-ON

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

1.55V
VF - FORWARD VOLTAGE - V

Figure 5.2.2.2-2

Measured Forward Voltage Variation for a

High-Efficiency Red Display
(8 Segments x 25 Devices =-200 LEOS)

5.0

-3
w

0:

...

=>
I0:{

0:
W

0-

c.. ~-2

~ >
w

0:

IZ

0:{

I-

~
"1'-0

0: ~ -1
0 u..
u..
w
U

.......

"<J

-...u..
>
<l

.1

.2

.4

.6

10

20

40

60

100

150 2

IF-FORWARC CURRENT-mA
5.2.2.3 Variation of
Temperature

Forward

Voltage

with

Figure 5.2.2.3-1

Change in

The forwmrd voltage temperature coefficient (6vFte

=
-mVtC) i8 given in Figure 5.2.2.3. The magnitude of this
coefficient decreases in value with increasing forward
current. The change in forward voltage with temperature
should be taken into account when doing a worst case
design.

A power limiting criterion needs to be established in order

to defme the tolerable limitations with which an LED
device may be operated in the strobed mode. The criterion
which has been selected is the LED junction temperature.
Specifically, the maximum tolerable strobe mode operating
conditions are limited to a peak current (IpEAK), pulse
~urat!on (tp) and refresh rate (f) which produce a peak
JunctiOn temperature (TJ PEAK) equal to the junction
temperatur~ obtained when the device is driven by the
maximum temperature derated DC current (IDe MAX)' A
graph of the curves which define the maximum tolerable
strobe mode operating conditions for a high-efficiency red
display is reproduced in Figure 5.2.2.4-1.
The thermd time constant of an LED mounted on a lead
frame is about one millisecond. For this reason, refresh
rates less than 1 KHz produce significant peaks and valleys
in the absolute junction temperature with the average
junction temperature less than the dc junction temperature.
As the refresh rate increases, the average junction
temperature approaches the dc junction temperature. This
is illustrated by the constant duty factor line drawn across
the face of Figure 5.2.2.4-1. As the refresh rate increases,
the allowable operating peak current per segment also
increases. It is, therefore, worthwhile to use a refresh rate
of 1 KHz or faster.

Forward Voltage Temperature

Forward Current.

5.2.2.5Maximum
DC Current
Temperature Derating

(JDC

Coefficient vs

MAX)

and

The operational curves in Figure 5.2.2.4-1 are used in

conjunction with the temperature derated maximum dc
current
to determine the maximum tolerable peak
operating current per segment (IPEAK MAX)' The dc
current derating curve for the high-efficiency red device is
reproduced in Figure 5.2.2.5-1. Figure 5.2.2.5-2 shows the
maximum tolerable dc junction temperature
and the
maximum
tolerable
dc power dissipation for the
high-efficiency red device.

5.2.3

Sample Calculation of a Typical and Worst Case

Design, Strobed Operation

2.

Maximum
ambient
temperature
as measured
between two digits just above PC board is TA =
60C (l40F).

1.

Maximum
segment.

tolerable

operating

conditions

per

LOCUS OF OPERATING
POINTS FOR A CONSTANT
IDC MAX = 10 mA AND
IpEAK MAX = 60 mA

LOCUS OF OPERATING POINTS

FOR A CONSTANT DUTY
FACTOR OF 1/4 (25%) WITHOUT
DERATING IDC MAX

20

III

~o

UJ
UJI15
a..
C,?a:
zUJ
_0
I-UJ
10
a:
a::>
9
UJI-I~2
8
a:UJ
~UJa:
:>a..a: 7
~~:>
6

" \.

00
~I-0
LLI-~

5
4

02:>

OUJ~
_a:

\,

"\

X
X
~

~ ~

'\

\.

~~
a:o~

, "
\.
,

"\'\.

\.

X~O

\.

", \
'"

,,

~~

r\t'\

~~,

--~r\

"

'"I

?at:I

i\
~

IDC MAX

~Ul
1S-"9

'f\.

r~

'"

"" "9~

1'1

-0

i\.
\

OPERATION IN THIS
REGION REQUIRES
TEMPERATURE
DERATING OF

'"

I\.

\.

7'\

'\.

_~_O

" \.1

I'\.
'\.

\.

1.5

......

'\.

I
~

t:I~

"1\1'

100

DC OPERATION

10000

tp - PULSE DIJRATION

-,i{sec

Figure 5.2.2.4-1 Maximum Tolttrable Peak Current vs Pulse Duration

for a High-Efficiency Red Display.

0:
0:

:>

u
u

18

~
:>
~
X

a:
UJ

s:

10 __

0
a..

~Q G<;;'
~<;;,~Q.'t>-~
cp~",\\~G
.~<;;,.'t>-

12 -

Cl

s:

./'I' ~

Cl

~
:>
~

\
\

O~

8 -

6 -

~
u

Cl
2

~
u

Cl
a..

UJ

o
30

40

50

60

70

I 90

80

85
T A - AMBIENT

TEIVIPERATURE

Figure 5.2.2.5-1 Maximum Allowable DC Current and DC Powel

Dissipation per Segment as a Function of Ambi mt
Temperature for a High-Efficiency Red Display

BASED ON A
= 100oC/W PER DIGIT
= 800oC/W PER LEAD

9JA

s:
E

:>
I-

90

wa..

80

~
w

70

-Icn
z

0:

I-

a..
~w

60

60

I-

50

50

:>

-,
0
w...J

40

40

~
:>
~

30

30

20

20

10

10

Cl~

a:"
ww

s:cn
00:

a..w
~a..
:>

~
X

20

100

0:

10

T CURVE
T~PICAL

16
14

r\

20

E
Iz
UJ

X
IDC MAX

=6

mA

1--'

10

20

30

T A - AMBIENT

40

50

60

0
a..

70

TEMPERATURE

Figure 5.2.2.5-2 Maximum Tolerable LED Junction Temperature and

Maximum Tolerable Power Dissipation for a HighEfficiency Red Display.

The duty factor is set at 1/4. It is necessary to select a

refresh ra": which will allow the maximum ratio of IpEAK
MAX/IDe MAX and stay within the timing constraints of
the systen. Refering to Figure 5.2.2.4-1, a minimum
possible fl~'resh rate of 100 Hz gives a ratio of 2.5 for a
pulse dural ion of t = 2.5 msec, a refresh rate of 1 KHz
P
gives a rat.o of 2.6 for t = 250 J.1sec. and 10 KHz gives a
ratio of 2.H for t = 25 J.1fec. A good refresh rate is 3125 Hz
which give~,a ratfo of 2.7 for t = 80 J.1sec.
p
In selectilli~ a minimum refresh rate in order to avoid
observable flicker, a designer should not use less than 100
Hz or less t'lan 5X the expected vibration frequency.

Next deter nine IDC MAX from Figure 5.2.2.5-1, IOC MAX
atTA=6
"Cis 16mA.

t:N F "'" (-.0014VfC)(S1.2-25C)

= -.079V

V F = 2.452 - .079 = 2.373V

The typical
power dissipation and typical junction
temperature can now be calculated with good accuracy:

= (.010SA)(2.373V)
= .026 W/SEG
SEG
POIGIT = (S SEG)(.026 W/SEG) = .20SW

TJ TYP = (100C/W)(.20SW)

+60C "'" S1C

Now the worst case is calculated using the maximum Rs

value. Equation 5.2.2.2-2 is used to calculate VF MAX' The
same iterative process, as above, is used to take into
account the -1.4mVtC change in VF MAX'

VF MAX = 1.55V + (.0432A)(33Q)

TJ MAX = (100C/W)(S

= 2.976V

SEG)(.010SA)(2.976V)

+ 60C

TJ MAX = 25.7 + 60 = S5.7C

VF MAX = 2.976V -(.0014VfC)(S5.7-25C)=2.S91V
PSEG MAX = (.010SA)(2.S91V)
POIGIT MAX = (S SEG)(.0312
TJ MAX = (100C/W)(.250W)

= .0312W
W/SEG) = .250W
+ 60C

TJ MAX = 25 + 60 = S5C

The typica segment power may be calculated using the VF

from equati:m 5.2.2.2-3:
Comparing TJ MAX and TJ TYPICAL with the curve in
Figure 5.2.2.5-2, the worst case TJ is 5C below the limit
and TJ TYP is 9C below the allowed limit for TA = 60C.

VF TYP = ; .65V + 21st (.0432-.005A)

VF TYP = 2 452V
The value iF = 2.452 volts assumes TJ = 25C. The
junction temperature is above 60C so the change in
forward vollage due to junction temperature is now taken
into accoun . This is an iterative calculation, however, one
iteration is t sually sufficient.
From Figur(' 5.2.2.3-1, I::NFtC = 1.4 mVtC for IF = 43
mA. An apI roximate TJ is' obtained and used to calculate
an adjusted' 'F:
TJ APPROJ< = (100C/W)(S

SEG)(PSEG APPROX)
= (100C/w)(SSEG)(.010SA)(2.452V)

TJ APPRQ}( = 21.2 C + 60 C = S1.2 c

+T A
+60C

The efficiency of an LED (photons/electrons)

increases
with forward current until the junction reaches saturation.
If the efficiency at a given current is taken as a reference,
then the efficiency at other currents may be compared to
the reference and a plot of relative efficiency can be
generated as shown in Figure 5.2.4-1. A relative efficiency
curve is actually generated by measuring the time average
luminous
intensity
at various peak currents while
maintaining a constant average current. The curve is
normalized to a relative efficiency of 1.00 at that data sheet
current where the minimum luminous intensity is specified.

1.8

>
u
z

1.6

~
u..

>

1.3

<l:

1.2

a:

1.1

i=
w

~ ~

/'

1.4

w
w

1.59

1.5

u..

I
I

,/

II

Standard Red

-.0188

High-Efficiency Red

-.0131

Yellow

-.0112

Green

-.0104

For an increase in temperature

change of the LED is as follows:

I
I

~ 1.0

<l:
w
c..

0.9

LED

0.8

f:"'

0.7

43

<>

I
5 10

20

30

40

50

60

Ip - PEAK SEGMENT CURRENT - mA

Figure 5.2.4-1

Relative Luminous Efficiency (Luminous Intensity

Per Unit Current) vs Peak Current Per Segment for a
HighEfficiency Red Display.

The time average luminous intensity can be calculated from

the following equation:
(5.2.4.1)

" TIM' AVG

~~~:PE~~::::: SPE;]['SPE~]

where:
=

Operating point average current

IAVG SPEC = Average current for data sheet luminol! s

intensity value, Iv
SPEC
1}1PEAK = Relative efficiency at operating peak current.
1}1P::A K SPEC = Relative efficiency at data sheet
peak current where luminous intensity Iv
is specified.
SPEC
I
Data sheet luminous intensity, specified a1
vSPEC
IAVG SPEC and IPEAK SPEC'
The light output of an LED also varies with temperature.
As the temperature of an LED increases, the lightoutpllt
decreases
according
to the following exponentil1
relations1ip:

K(T A-25C)
Iv (TA)=lv(25 C) e
where:
Iv (TA) = Luminous intensity at operating ambient
temperature.
Iv (25<C) = Luminous intensity at TA = 25C
TA

Operating ambient temperature,

Exponent constant 1fc,

dependent
2.7183

of 1C, the light output

+1C Factor

% Change

Iv

LED

1.7

LED material

Standard Red

.98141v

-1.86

High-Efficiency Red

.98701v

-1.30

Yellow

.98891v

-1.11

Green

.98971 v

-1.03

Equation 5.2.4-1 applies to luminous intensity as perceived

by the eye and is not applicable to radiant power output.
5.2.4.1 Sample Calculation
Intensity

of Time Average Luminous

The time average luminous intensity for the example in

Section 5.2.3 can be calculated using the value of ~ = 300
mcd at IF = 5 mA. The relative efficiency calculation does
take into account junction temperature rise above ambient
within reasonable error. The resultant IvrIME AV~ is then
corrected for the increased ambient above TA = 25 C.
The relative efficiency for Ip = 43 mA is 1.59, as obtained
from Figure 5.2.4-1, and the time .average luminous
intensity
at TA = 25C is calculated to be 1030
microcandelas/ segment.
I TIME AVG

r10.8 m~

n5~l

L 5 mAJ L 1 J

=
At TA = 60C, the
microcandelas/ segment.

[300 /-lcd]

1030/-lC;d/SEGMENT
light

output

decreases

to

651

I (60C) = 1030 (.0131) (6o-25C)

v

Iv (60C)

1030 (.6322)

= 651

/-lcd/SEGMENT

5.2.4.2 Digit-to-Digit and Segment-to-Segment

Intensity Ratio

Luminous

The luminous intensity value listed on a 7-segment display

data sheet is a digit average luminous intensity for the seven
segments. The ratio of the maximum Iv segment to the
minimum Iv segment within a digit needs to be held to a
reasonable value. Also, the digit to digit luminous intensity
ratio needs to be held to a reasonable value so that digits
may be stacked side by side with a pleasing appearance.

Tests have mown that when a display is lighted with a

string of 8' , an observer may scrutinize the digits with a
critical eye, fmding all sorts of faults. However, this is not a
normal mo: e of operation. The exact same display lighted
with a strirt" of random numbers may well be acceptable to
the same 01 server. In fact, the observer reading the same
display to )btain information during normal usage will
NOT notice the minor faults he detected when scrutinizing
the string C' 8's. This is true when the segment-to-segment
and digit-toligit ratios are kept within certain limits.

The forward current through an LED must be limited in

some fashion to obtain the desired illumination and to
prevent high current damage. The two most popular
methods of accomplishing these two objectives is to (1)
drive the display from a decoder/driver with switching
outputs that control the flow of LED current through
external limiting resistors and (2) drive the display from a
decoder/driver
with constant current outputs. Figure
5.2.5.1-2 illustrates both methods.
IF

10 mA/SEGMENT

An observer scrutinizing an LED display is able to detect a

difference i I light output at a threshold level of about
1.6: 1. A change in light output becomes quite noticeable at
the ratio of ~.3:1.
BCD
DATA
INPUT

A typical Ie sic system will produce output information in

the form ()' BCD data. This BCD data must then be
converted ':') the 7-segment matrix code format of the
display. Drivers are then required to drive each segment of
the display. The two methods of driving a 7 -segment
display are 1) direct dc drive of a single digit and 2)
strobing a sl ing of digits.

DIRECT DRIVE CIRCUIT FOR A

5082-7650 COMMON ANODE DISPLAY
IF

= 6 mA/SEGMENT

The simple~' method of driving a 7-segment display is to

have one decoder/driver for each digit. This is usually cost
effective up to four or five digits depending upon the
trade-off bel ween component cost and circuit complexity.
Figure 5.2.5. 1-1 illustrates the direct dc drive concept. Each
digit has i1, own decoder/driver and is continuously
illuminated. Though LED displays are more efficient when
strobed, thf advantage with direct drive is that the drivers
need not han dle high current levels. Also, the display circuit
complexity i simple as extra timing is not required.

BCD {~~
DATA
INPUT

DECODER/
DRIVER

LOW CURRENT SOURCE DIRECT DRIVE ON A

50827653 COMMON CATHODE DISPLAY

7SEGMENT
DISPLAY

BCD
DATA
INPUT
LOGIC WITH
BCD DATA
OUTPUT
(CLOCK)

HP 5082-7653

NATIONAL
OM 8856
DECODER/
DRIVER

IF
LOGIC
WITH
BCD DATA
OUTPUT

COMMON
CATHODE

15 mA/SEGMENT

4511
LATCH/
DECODER/
DRIVERS

DECODER/
DRIVER
DIRECT DRIVE CIRCUIT FOR A 50827760
COMMON CATHODE DISPLAY

Figure 5.2.5:

COMMON
ANODE
+5V

I Block Diagram for Direct DC Drive Scheme for

7.segment Displays.

COMMON
CATHODE

Figure 5.2.5.1-2 Example Methods for Direct DC Driving 70Segment

Displays.

Wher using resistor current limiting, the current lirr iting

resistor may be determined from the following equatio:l:
Multiple function devices that will direct drive LED
displays are becoming more available. An example is the
Texas Instrument SN74143 BCD counter/4 bit latch/BCD
to 7-segment decoder/driver. The outputs are 15 mA
constant current sink. The SN74144 is the same device with
a 25 mA sink capability with external current limiting
resistors. A 4digit high speed counter can be made using
the SN74143 and the yellow .43 inch common anode
displays as pictured in Figure 5.2.5.2-1. The 15 mA LED
drive current is sufficient for the display to be read in most
bright ambients.

R=-VCC-VF-VCE _
IF
whel'e:
R
VCC

Series current limiter

Supply voltage

VF

LED forward voltage at IF

VCE

Saturation voltage drop across driver

transistor
LED forward dc current

II:

If a decoder/driver with constant current outputs is teing

used and the ambient temperature around the display
requites dc current derating, shunt resistors at the d :iver
outputs may be used to bleed off the excess current.

The data strobe line may be held low to obtain a

continuous display of the counter state, or strobed to latch
specific counts. A decimal point input is provided in each
SN74143 so that any desired decimal point position may be
obtained by providing a logic high true input on the
appropriate dp select line.

VOH
RSHlINT = ---IOH- IF

As an example,
the Fairchild
9368
7-segment
latch/decoder/driver will source a maximum of 22 mA over
a wide temperature range. A display is to be composed of
standard red common cathode displays operating in an
ambient temperature of 70C. The data sheet derating is
0.43 mA dc /0C above TA = 50 C. Therefore, at TA = 7CC,
IDC MAX = 16.4 mA. A worst case design requires shun :ing
5.6 rilA from each output of the 9368. The LED fornard
voltagf: is 1.60 volts. The shunt resistor value is 286Q:

The most efficient method of driving an LED display is to

strobe it at a high peak current and a low duty factor. This
takes advantage of the increased efficiency of an LED at
high peak currents and the reduction in average power
dissipation when compared to dc operation. A multiplexed
display
design
becomes
cost effective when the
decoder/driver and display PC board size and complexity
required for dc drive become significant when compared to
the timing circuitry and digit driver cost and size of PC
board needed for strobed operation. The power saving that
is attainable with strobed operation may show up as a need
for a smaller sized power supply, contributing to cost
reduction.

1.60V
RSHUNT=

(.022 _ .0164)A=

286Q

DIGIT #2

DIGIT#1
OST
IGNIFICANT
IGIT

a bed

CLOCK INPUT vDATA STROBE ~

INPUT
CONSTAI\T

9 dp
I

A A A A A A A A
a bed
e f 9 dp

CURRENl'
SINK@15mA

STROBE

C RBO
C

INPUT
INPUT ~

1-1
1 I

HP 5082-7730
(5)

74143

~tJ~H

RBI
SCE1

1r

RBO
MC

74143

CK
RBI
SCE1

CLR
PCE1
B,OAOBOCOoDP

p-<
P-<
PI r

RBI

MC

SCE1

CLR
B,O
I

PCE1
DP

1-'
I

AAAAAAAA
STROBE

CK

74143

CK

I>--(
I>--(

RBO

lr

CLR
PCE1
B, 0AOB 0cOo DP

MC

LEA ST
SIG NIFICANT
DIGI T

i-I

i-I

STROBE
RBO

DIGIT #5

1-'

AAAAAAA

I
I

:>---c
I>--(

1-'
l I

AAAAAAAA
STROBE

CLOCK

CLEAR
PCE1
B 0AOe0COODP

DIGIT #4

DIGIT #3 ~

74143

RBI
SCE1

Firr
-=-

111111
STROBE
RBO

74143

CK

f-

RBI

P-

p...
p...
PCE1
SCE1

CLR
MC

On DP

~rAO

"7.-

L POINT
~
SELECT ~

()

(
I

Figure 5.2.5.2-1 High Speed 5-Digit Counter Using Direct DC Drive

LED DiSplays.

1 LATCH LOGIC OUTPUT

BCD TO
7-SEGMENT
DECODER

c
7-SEGMENT
DRIVERS

e
f
9

LOGIC INCL UDING

6-DIGIT (.lATA
STORAGE BJFFERS
AND
MULTI PLE XING
CIRCUIl RY

I~
1_1

7-SEGMENT
LED DISPLAYS

DIGIT
DRIVERS

1 of 6
DIGIT
SCANNER
Figure 5-2.5.3-1

Block Diagram of a Strobed (Multiplexed)

LE
Display.

A strobed .:ix digit display is illustrated in block diagram

form in Fi! ure 5.2.5.3-1. The like segments of each digit in
the display are tied to a common segment bus line which is
connected ,0 one of seven segment drivers. For example, all
of the "a" segments are tied common to the segment "a"
driver and ,lIl of the "b" segments are tied common to the
segment
"b"
driver.
This allows
one BCD to
7-segment/( ecoder driver to be time shared between all
digits, with BCD data for each digit stored in one of six
digit data sorage buffers. Each digit is enabled by its own
digit driver.
The ope rat on of this multiplexing scheme is as follows:
First Clock Pulse: The multiplexing logic selects digit data
buffer number 1. This first digit BCD data is presented to
the decode /driver and the required segment lines are
activated.
['he digit scanner enables only digit driver
number I a.ld the first number is displayed. Second Clock
Pulse: The multiplexing logic selects digit data buffer
number 2.
s second digit BCD data is presented to the
decoder/dri"er and again the required segment lines are
activated. lile digit scanner now enables only digit driver
number 2 a ld the second number is displayed. This process
is continued through digits 3, 4, 5 and 6, and then the
whole proce<;srepeats.
The strobin!
a minimum
vibration. H
a 7-segment
or faster.

rate to produce a flicker free display should be

of 100 Hz or 5X the expect mechanical
)wever, to optimize the overall performance of
display, it is best to use a refresh rate of I KHz

When desig ling a multiplexed display system, timing

consideratio lSbecome important. The timing parameters to
take into ac :ount when designing a display of a string of
N-digits are:

6-Digit

N = quantity of digits

The first consideration is to select a desired refresh rate.

The reciprocal of the refresh rate is the refresh time period
in which all N-digits are enabled, one at a time:
1
Refresh

Period .=

T =

The maximum allotted digit on-pulse duration is the refresh

period divided by the quantity of digits to be strobed:
'T

Maximum

OnPulse

= t p max = --N

Figure 5.2.5.3-2 illustrates a basic timing diagram for

strobing N=6 digits. Each digit is enabled in sequence
without any overlapping of the digit enable time, t max' A
segment line, however, may remain active if it is re~uired in
two or more successive digits. A variation of this basic
timing scheme is to use Pulse Width Modulation (PWM) to
vary the operating on-pulse duration, t , to values less than
the allotted pulse duration, tp max'
may effectively
be used to vary the apparent brightness of a display to
ma tch changing ambient lighting conditions. As an
example, for low level ambient conditions, t is reduced to
dim the display. The segment peak cJrent
remains
constant to take advantage of the increased LED efficiency
at high peak currents, yet the average current is decreased
which reduces the time average luminous intensity of the
display, see equation 5.2.4-1.

r\VM

TYPICAL
SEGMEIIT

ON
OFF

I L.J L.J

Pmax.,
ON
OFF

ON
OFF

DIGIT 4

DIGIT 5
DIGIT6

ON
OFF
ON
OFF
ON
OFF
ON
OFF:

"

---

tp OPERATING

ON-TIME

,
I

OFF

DIGIT2

DIGIT
EI\ABLE
TIIVIING

II

DIGIT 3
DIGIT4

!--.--...Jn~

DIGIT 5

OFF
ON
ON
OFF

- --------

ON

OFF

---I

ON
OFF

~r
n

ACTIVE
SEGMENT
TIMING
:

REFRESH--~
PERIOD

M~

~tb

ON
DIGIT

rLJLJI
rr;tp

n'---------...!

OFF

TIMII\G

PERIOD

_,

I
~

ON

TYPICAL
SEGMENT

n
I

DIGIT 3

ION-PULSE

DIGIT2

I--REFRESH

DIGIT 1

r-1 r-1 ACTIVE

SEGM ENT

DIGIT
BLANKING
TIME

Ul

__ n

___ n _

DIGIT
ENABLE
TIMING

L-JU
~
...rl.J

DIGIT6

ON
OFF

I
HI

tp

LOW
DIGIT ENABLE

ON-PULSE = tp
=N
max

DUTY FACTOR =

Figure
5.2.5.3-3
illustrates a basic timing diagram
employing PWM. The digit enable on-pulse,. tp' is. varie~ to
obtain the desired display time average lurmnous mtensl.y.
The maximum allotted pulse duration is now composed of
the digit operating on-pulse, tp' and a blanking time, tb:
Allotted Pulse Width

tp max

p + tb

= t

The du':y factor is now defined as the ratio of the operating

on-pulse to the refresh period:

The active segment on-pulse duration need not he

modulated and can remain at t max' It is usually sufficient
and easier to modulate only thf digit enable operating tim e,
though a designer may choose to modulate either the
segment on-time, digit on-time or both.

+ tb

A Basic Timing Diagra'!1 for Strobin~ N=6 Digit~.

(The Typical Segment IS Shown Active When Dig ts
1, 2, 5 and 6 are Enabled.)

= tp

t
DUTY FACTOR = ~

Figure 5.2.5.3-3
Figure 5.2.5.3-2

max

A Basic Timing Diagram Using Pul~;eWidth

Modulation to Vary Digit Enable Time, tp. The
Active Segment Pulse Duration is not Modulated
and Remains at tpMAX.

An additional use of PWM is applied when a fast strobing

rate in the neighborhood of 10 KHz is used. At this fast
refresh rate, the turn-off time of driving transistors may not
be fast enough to prevent overlapping of digit and segment
on-pulse durations. This timing overlapping is called
ghosting, since a digit that should be off is still partially
turned-on. A timing overlap of approximately 3% is usually
sufficient to have perceptable ghosting in moderate ambient
lighting conditions. To prevent this situation from occuring,
PWM is employed to effect inter-digit blanking. In this
mode of operation, a fIxed digit (and possibly segment)
blanking time, tb, is built into the display timing to insure
sufficient time for the driving transistors for one digit to
turn completely off prior to turning on the driver
transistors for the next digit.
The value of the segment current limiting resistors may be
calculated from the generalized schematic shown in Figure
5.2.5.3-4.

VCC
2000
IpEAK LIMITATION

en
::l

- --

- -I

i'
.~

DIGIT
DRIVER

COMMON
ANODE
DISPLAY

'B::l. 1000

:E
::l

I-

800

600

z
..J

w
Cl

<X:

Cl

w
c..

>

Z
(,W

SEGMENT
DRIVER

--

I
I
I
560,ucd

- 1-

I I

\ - 380 ,ucd

300

~
~

II

>1<X:~

>
100
1

~
"\

..-

't'

II""I)

DC OPERATION

/I
/ I
:.

,J

200

I-~

II'

I-

en

I~~

STROBED OPERATION
1/6 DUTY FACTOR

0:: en 400
w 0::

>
<X:

I III

8.2 mA DC
I

I
6mAAVG

I
,

I
8

10

20

IF AVG - AVERAGE CURRENT PER SEGMENT - mA

SEGMENT
DRIVER

COMMON
CATHODE
DISPLAY

DIGIT
DRIVER

From Figu-e 5.2.5.3-4, the following equation may be used

to calculat( the vaue of the current limiting resistor, R:
R

= -VCC -- (VF

+ VCE SEG + VCE DIGIT)

_
IpEAK

where:
V CC

Supply voltage

VF

LED forward voltage at IpEAK

V CE SE:3

Saturation voltage drop across segment

driver at IpEAK

VCE DIGIT = Saturation voltage drop across digit driver

at (IPEAK x the number of segments
multipled)
IpEAK = Peak LED segment current

Figure 5-2.5.3-5

Comparison of Time Average Luminous Intensity

Obtained from Strobed and DC Operation, HighEfficiencly
Red Display.

as

The advantage of strobe operation over dc operation is

illustrated in Figure 5.2.5.3-5. This is a graph comparing the
light output obtained by strobing a 6-digit display vs. dc
driving each digit individually. The design constraint is to
maintain 6 mA average. At 6 mA dc the time average
luminous intensity per segment is IV AVG = 380Mcd. By
strobing the displays at IpEAK = 36 mA and a duty factor
of 1/6, the light output is increased to IV AVG = 560MCd.
This is an increase in light output of 1.47X. over dc
operation for the same average current.
At 6 mA average, the strobed operation pro uces an
average power dissipation within the display of 10.3 mW
per segment vs. 10.0 mW per segment for dc operation. To
obtain the light output of 560 Mcd per segment by using dc
operation would require a forward current of 8.2 mA dc
and a power dissipation of 14.1 mW per segment. The total
power saving obtained by strobing is considerable. This
power saving is realized by the reduction of the required Ie
components needed to drive the display. For example, in
this illustration, only one decoder/driver is required for
strobed operation vs. six needed for dc operation.
The human eye is a time average detector and, therefore,
will easily discern the increased light output obtained by
strobing.

5.2.6

I nterfacing
Displays

Microprocessors

to

Seven

Segment

Seven segment
displays
can be interfaced to a
microprocessor with only a few external components. fhe
microprocessor can be used to control the refreshing of a
multiplexed display or simply to periodically up-date" dc
driven display. In either case, one or more eight bit latches
are required to hold the the seven segment and digit
information. If" the seven segment display is multipleJl ed,
then some timing circuitry is also required to periodically
request new information from the microprocessor. . he
timing circuitry could consist of either a monosta ble
multivibrator or an oscillator. This circuitry would reqt est
an intE:rrupt and use the already available hardware to
handle the proper interrupt decoding. Several commercidly
available seven segment decoder/drivers are available t lat
can directly drive the seven segment LED displays.
When the seven segment display is driven on a dc bash., a
seven segment decoder/driver is required for each disphy.
Figur.
5.2.6-1
shows
an example
of a seven
segmen t-microprocessor interface. Each display is driven by
its own seven segment driver. The Fairchild 9374 has
current source outputs that drive each LED segment at 15
mA dc. The decimal points are driven by a National
DS8859, a hex latch with current source outputs. The Intel
8080A microprocessor updates each display, with an
OUTPUT instruction.
The second byte of the OUT
instruction specifies an eight bit address which determir es
the display that is to be updated. Upon execution of t e
OUT instruction, the lowest four bits of the accumulator
are loaded into the appropriate latched-decoder/driver.
A
separate OUT instruction also updates the decimal poir ts
and overflow digit. Since the upper six states of tle
Fairchild 9374 are decoded as (-, E, H, L, P, and blank), it
is possible to blank each digit selectly under microprocess)r
control, indicate an overflow by a row of minus signs, and
use the word "HELP" to catch someone's attention.
A microprocessor can also be used to multiplex a seVfll
segment display. Figures 5.2.6-2 and 5.2.6-3 show two wars
by which this can be accomplished. In Figure 5.2.6-2, tie
microprocessor outputs two bytes per digit. The first byle
contains segment information for the display. Since the
seven segment
information
is decoded
by the
microprocessor, the programmer can program the charactfI
font to include any special symbols that are desired. The
second byte turns on the proper digit driver. At the same
time, thE: RCA 4047 monostable multivibrator is triggeretl
and requests another interrupt in 2 milliseconds. UpOll
requesting an interrupt, the interrupt circuitry (not shown)
must force an RST(7) instrument to be executed by tht:
Intel 8080A microprocessor. Then the refresh progranl
shown in Figure 5.2.6-2b is executed. The program reads a

pointer from RAM and then outputs two bytes of

information to the display. The pointer is incremented by
one and compared to the final address of the data me. If
they are equal, the pointer is reset to the address of the first
byte in the me, otherwise the pointer is set to the address
of the next byte in the me. As shown, the circuit refreshes
five seven segment displays on a 20% duty factor at 60 mA
peak/segment. This technique can be expanded to eight
displays without additional hardware. The time required to
refresh the display can be determined as shown below:

(165N

+ 7)R

MICROPROCESSOR

CLOCK RATE

where N is the number of digits to be refreshed, and R is

the strobing rate.
For example, suppose five digits are refreshed on a 100 Hz
rate and the microprocessor uses a 2MHz clock, then 4.2%
of microprocessor time is required to refresh the display.
For the remaining 95.8% of the time, the microprocessor
can be performing countless other tasks.
Figure 5.2.6-3 shows another techniq11e that can be used to
multiplex a seven segment display. With this technique, the
microprocessor outputs only a single byte of information to
the display. Four bits contain the BCD character to be
displayed, one bit contains the decimal point information,
and the remaining three bits determine which digit should
be turned on. The Motorola MC 14511 seven segment
latched decoder driver can source up to 25 mA per segment
to a common cathode LED display. The high efficiency red
displays that are shown can be satisfactorily operated at 25
mA/segment on a 20% duty duty factor. This power
requirement is considerably below those required for
standard red seven segment displays. The interrupt request
circuitry is similar to the previous example. After receiving
an interrupt request, the interrupt hardware must force a
RST(7) instruction to be executed by the Intel 8080A
microprocessor. Following the RST(7) instruction, the
program shown in Figure 5.2.6-3b will be executed. This
program is sinlilar to the program shown in Figure 5.2.6-2b
except that only a single byte is outputed to the display.
This technique can also be expanded to eight displays
without additional hardware. The time required to refresh
the display can be determined by equation 5.2.6-2:
(143N +7)R
MICROPROCESSOR

CLOC

where N is the number of digits to be refreshed, and R is

the strobing rate.

~--'+--T~-

r1
b

Ti

j,3

,J

HP S0827:6'

11

I DIG 3

DIG4

HP S082.17S0

I-I

IADJ

06 Os 0403

'-I

'-I

o/~

I DIG

HP S082-17S0

00

1-'

dpabede'

'I

10

HP S08217S0

"

)~I
dpabede'g

I DIG 2

HP S082-17S0

doabede'

d abede'

O2 01

NATIONAL
D588S9

ST 16 IS 14 13 12 11

FAIRCHILD
9374

FAIRCHILD
9374

FAIRCHILD
9374
AOA1 A2A3

EL AOA,

FAIRCHILD
9374

e;:

L AOA1 A2 A3

A2A3

AO A, A2 A3

DO

-r_

..

DI

Ir>

01

0
03

04
,OS

74LS138

PROPER ADDRESS
CODE FOR OUTPUT
PORTS (nl, (n+1I,

{:~
A

(n+2I, (n+3I,(n+41
"OUT" FROM 8080A
STATUS LATCH

:
C

.0

~
2

3
1

4" :>-

5)
6)
7)

UPDATE DIGIT 1:
UPDATE DIGIT 2:
UPDATE DIGIT 3:
UPDATE DIGIT 4:
UPDATE DECIMAL POINTS
AND OVERFLOW

OPCODE

ADDRESS (A1 AOI

DATA WORD (07 DOl

OUT (n)
OUT (n+11
OUT (n+21

A7 A6 AS A4 A3 0 0 0
A7 A6 AS A4 A3 0 0'
A7 A6 AS A4 A3 0 1 0
h_~6_~S_~4h~_1_1_
A7 A6 AS A4 A3 1 0 0

XXXX
XXXX
XXXX
XXXX

Q\:!IJ.':!~31
OUT (n+4)
(~

I
I

DCBA
DCBA
DCBA
DCBA

-X XfMTOVtDp~ID~3iD-P2fD;1

AS A4 A31 AS DECODED BY HARDWARE

X = DON'T CARE

Figure 5.2.6-'

Seven Segment Displays can Interface Directly to a

Microprocessor with Standard Latched Decoder/
Drivers.

0v = OVERFLOW:
Dp= DECIMAL

0v = 0 OFF, 0v = 1 ON

POINT:

Dp= 0 OFF, Dp = 1 ON

+VDD =5V
~

FIr.I-lT P~!P TRANS!STORS

WITH COMMON EMiTTER

2711.

ADDRESS CODE FOR

OUTPUT (n)
"OUT"
FROM 8080A
STATUS LATCH

DO

00

D1

01

D2

02

D3

03

DO

D1

01

D2

02

D3
ST

f
g
dp

ST
MR

,-,

MR
DIS

HP5082-7760

'-I

DIS

I=J

NATIONAL
MOTOROLA

14508

ADDRESS CODE
FOR OUTPUT (Q)
"OUT"
FROM 8080A
STATUS LATCH

DO

00

D1

01

D2

02

D3

03

DO

00

D1

01

D2

02

D3

03

ST

REOUEST
FORCES
RST(7)
INSTRUCTION

DIG I

10

II

14 +VDD

MR

DIS

'-'

DIG 2

COMMON CATHODE

OUTPUT
DATA

OUTPUT

(n)

WORD (D2"DO)

ST
MR

INTERRUPT

OS8863

SEGMENT
SEGMENT

DATA

(Q)

WORD (D7'

DOl

X X X D5 D4 D3 D2 D1

= 0, SEGMENT
= 1, SEGMENT

ON
OFF

Dn = 0, DIGIT
Dn = 1, DIGIT

OFF
ON

Figure 5-2.6-2

A Microprocessor can be Used to Strobe a Seven

Segment Display. Microprocessor Outputs Decoded
Digit and Segment Information to Display.

DIS

Q
RCA
4047A

--

+VDD = 6V
MOTOROLA
14511

TO DATA

BUS {

~~

O2
ADDRESS CODE ~n
OUTPUT (n)
"our' FROM
STATUS LATCH

l.-

03

(5)
HP 5082-7653

DIG 2
MOTOROLA
14042

D4
TO DATA

BUS
+VDD

~L4

10
INTERRUPT
REOUEST
FORCES
RST (7)
INSTRUCTION

..11

..-

DO
01

001 -

O2 02-

07

03 03-

l1
-

4047A
RCA

~5
6

MOTOROLA
14028

COMMON
CATHODE

Or--1 r--2 r--: r--5

.rn
~

r
.J...

6
7

DATA

WORD (07 - DO):

O2 01 DO
0201

i i
y

DO = DIGIT

C B A
SELECT

Y = DECIMAL

POINT SELECT

Y = 0, DECIMAL

POINT ON

Y = 1, DECIMAL

POINT OFF

DC B A = BCD CODE OF SEGMENT

Figure 5.2.6-3a

A Microprocessor
Can Be Used To Strobe a
Seven Segment Display. Microprocessor
Outputs
Single Byte Which Is Decoded By Display
Circuitry.

I
INFORMATION

CLOCK
CYCLES

ADDRESS

OP CODE

(0038)16

PUSH PSW.
PUSH HL

1111
(111

LHLD

1161

COMMENTS

HL - POINTER

AL
AH
MOVA.M

(7)

A= IHL)
STORES NEW SEGMENT INFDRMATION

1101

OUT
INX HL

HL"'HL+l

151
(7)

MOVA,M
OUT

A= IHL)
TURNS ON DIGIT

1101

For example, 3.6% of the microprocessor's time would be

required to refresh five displays at a 100 Hz refresh rate
with a 2 MHz microprocessor clock. While these examples
use an Intel 8080A microprocessor, these techniques can be
used with
other microprocessors
by modifying the
hardware and programs to reflect a different instruction set
and different ways to output information to peripheral
devices.

DRIVER

A-L

(51
(7)

MOV A, L
CPI

COMPARE L TO ADDRESS OF LAST MEMORY

LOCATION

(19116
JNZ LOOP

JUMP IF A"

1101

5.2.7

Detection and Indication of Segment

Seven Segment LED Displays

Failures in

119)16

(ADDRESS OF
LOOP)
MVI, L

171

IOF)16
INX HL

LOOP

L = IOFI16
HL=Hl+l

(51
(16)

SHLD

POINTER = HL

AL
AH
POPHL

1101

POPPSW

1101
(41

EI

!.lQl

RET

MEMORY
ADDRESS

CONTENTS

AH AL

POINTERL

DOl COMMENTS

(07 .

POINTER = NEXT DIGIT TO BE DISPLAY

~~~~~--~~~~~~=~~
(X X 1 01, 6
dp 9 fed
c b a
DIGIT 1
(X
IX
(X
(X
IX
(X
(X
IX
IX

X
X
X
X
X
X
X
X
X

Q.. 9
dp 9
CL Q
dp 9
Q.. Q
dp 9
Q.. Q
dp 9
0 0

1 1116
1 2116
13)16
1 4116
1 5)16
1 61, 6
1 7116
1 8116
191,6

Figure 5.2.'-2b

99Q Q
fed
c
Q Q Q Q
fed
c

Q 1
b a
~ Q
b a

DIGIT 2
DIGIT 3

Q Q Q ! Q Q
fed
c b a
Q Q 1 Q Q 9
fed
c b a
0 I 0 0 0 0

DIGIT 4
DIGIT 5

Refresh Program Used to Strobe the Display Shown

in Figure 5.2.6-2.
CLOCK
CYCLES

ADDRESS

()P CODE

(00381, 6

PUSH PSW

(111

PUSH HL

(11)

ll-tLD

1161

COMMENTS

HL = POINTER

~L

"H
IVOVA,M

MOV

(7)

A. L

DISPLAY IS UPDATED
A=L
COMPARE L TO ADDRESS OF LAST MEMORY
LOCATION

(51
17)

CI
(14116
Jr.Z LOOP

110)

JUMP IF A"

17)

L = (OF116

(51

HL=HL+l

(141,6

VIDDRESSOF
LOOP)
MIll, L
(e

FI,6

I~X HL

POINTER

(16)

SHLD

HL

At.
All
PCP HL

1101

PCPPSW

1101
(41

EI

!.lQl

REf

MEMORY
ADDRESS
AH AL
AH AL + 1

IJ(XT6i16
(X
IX
IX
IX

X
X
X
X

11116
12116
13116
14)16

CONTENTS (07' Dol

PO NTERL
PO NTERH
o-'''jj1Y
o
1:Y I
BCD
o 1 0, Y,
DATA
01 1, Y,
1 0 O,YI

r-r

DIGIT 1
DIGIT2
DIGIT 3
DIGIT4
DIGIT5

X = DON'T CAR E
Y = DECIMAL POINT INFORMATION
Y = C DECIMAL POINT ON
Y = 1, DECIMAL POINT OFF

Figure 5.2.6-3)

Segment failures in seven segment LED displays, although

rare, do have a measurable probability of occurrence.
Characteristically,
the failures are electrical opens or
unusual degradation of light intensity. Shorted segments
have such an extremely low probability of occurrence that
they do not need to be considered. Fortunately, it is a
rather simple matter to detect the occurrence of an open
segment in a seven segment display matrix.
The circuits on the following pages depict two possible
techniques for the detection and indication of such failures.
5.2.7.1 Seven-Segment
Anode Displays

Self-Test

Circuit

for

Common

A= (HLI

(101

OUT

The failure of a single element in a seven segment display

device could result in the presentation of erroneous
information which may not be detectable to the user. Such
failures may merely be annoying in some equipment
applications but may, on the other hand, be quite serious in
the case of point
of sales terminals or medical
instrumentation.

Refresh Program Used to Strobe the Display Shown

in Figure 5.2.6-3.

The circuit in Figure 5.2.7.1-1 will detect an open segment

in any of the digits of a multi-digit 7 segment display.
Basically, the test consists of turning on all segments for a
short period not detected by the eye. Should the segment
current corresponding to an "ON" segment not flow, a
logic circuit blanks the entire display until reset. The
portion of the circuit inside the labeled dashed line
represents the failure detection system.
The scanning circuitry for the strobed display consists of a
clock, a counter and a decoder. The 9601 one-shot
multivibrator produces a 1 microsecond pulse out for each
clock period input to the scan counter. The Q output of the
one shot enables the lamp test function of the BCD to
seven segment decoder driver, turning on all of the segment
outputs. The status of each of the segment lines at nodes A
is monitored by the 8 input -7430 NAND gate. If all
segments are intact, the voltage of all the nodes A will be a
logic high (V CC-~CE SAT V FLED) = ~2 volts. The output
of the 8 input NAND will be a logic low and therefore no
clock pulse will be gated through to the -7474 flip flop. If a

r-----V-

1---------------,
TYPICAL DIGIT SELECT
I
I

STROBING CIRCUIT

I
I
I
I8
I
I

6
SIGNETICS
8280
COUNTER

I
I

I
0

14 FAIRCHILD
1 9301 1/10
2 DECODER 9

I
I

15

I
IL

R
470n

I
I

~VCE SAT

I
I
I
I
I
I
I

COMMON ANODE ,
7SEGMENT LED I
DISPLAY
I
a

I
I
I

...1

1
6

THIS EXAMPLE)'
DIGIT
I
DRIVER
I

4.7k

OCD
DATA {
INPUT

cc

~I

---(10Floi'N-i

7447
7-SEG.
DECODER

1= 50 mA
NOMINAL

TO SEGMENT
LINES OF
} OTHER DIGITS

R
470n

LIT
BI/RBO 9

r----

- ---

------VCC
1000 pf

2K

VCC
2
3
~lms

------

2
: 7430
9 : 8-INPUT
NAND
VCC 10

----

----1
I

I
I
3

RESET
CLK R
7474
FLIP
FLOP

FAIRCHILD
9601
1
ONE SHOT
6
Q

II
-

I
I

I
I
I
I

I
FAILURE DETECTION CIRCUIT

----------------~
Figure 5.2.7.1-1 Self

segment should be open, the low true output of the -7447

decoder will produce a logic 0 at the node A corresponding
to the open segment. The resultant logical 1 output of the 8
input NJ ND will be gated through the 2 input NAND by
the Q outuput of the one shot, thereby clocking the
flip-flop. The 470 ohm segment current limiting resistor R
is the largest value which can be used and still product a
logic low at node A when using a standard TTL 8 input
NAND gate.
In the circuit shown, the output of the flip-flop is used to
blank the display on the occurrence of an open. The
flip-flop is externally reset.
5.2.7.2 SI~lf Test Circuit for Seven-Segment Displays and
Associated Decoder Driver
The circuit in Figure 5.2.7.2-1 will detect an open segment
in any of the digits of a multidigit seven segment display.
Also, through the use of a redundant BCD to seven segment
decoder, a malfunction of the decoder can be detected.

The circuit functions through the use of exclusive OR

gating. The 8 gates labeled V 1 will only produce a logical 1
output when a segment is designated "ON" (-7447 decoder
output low) and current is flowing through the segment so
as to produce a logical 1 at node A and a logical 0 at node B.
This is condition 1. The other two conditions are: (2) a
segment turned off; nodes A and B are high, and (3) a
segment is turned on by the decoder but no current flows
due to an open; nodes A and B are low. Both of these
conditions will result in a logical 0 at the output of gates
V l' Failure detection occurs at gates V2. One input of
each of these exclusive OR gates is the output of the gates
VI' The other input is the respective output of the -7448
BCD to 7 segment decoder. The -7448 has inverted logic
outputs when compared to the -7447 (high true instead of
low true). The three conditions defined above will produce
the following results at the outputs of V .
2
Condition:

r- - - - - VC-;;- -

r~~A~~IT~ucr---------1
STROBING CIRCUIT

15
8

SIGNETICS
8280
COUNTER

I
I

14
1
2

o
FAIRCHILD
9301 1/10
DECODER

I
I
1

1
I
~I

- ("'~f101N---,

THIS EXAMPLE)

I
4JK

I
I

I
I

~
jVCESAT

I
1

COMMON ANODE:
7-SEGMENT
I
LED DISPLAY
I

I
I
I
I

DIGIT
DRIVER

:L 1----------

I
I

Ij

...1

1 OF 7 (TYP.)

7
1
BCD {
DATA
INPUT

2
6

7447
7-SEG.
DECODER

TOSEGMENT
LINES OF
} OTHER DIGITS

R
B 470n A
BI/RBO

7
BCD {
DATA
INPUT

1
2
6

CLK
7448
a
7-SEG.
DECODER

7474
FLiP-FLO
2 D

U4

111f

I
Figure 5.2.7.2-1

Self Te~t Circuit for 7-Segment Displays and

Associatt!d Decoder/Driver.

C and D both logical 0; U2 output logical O.

C at logical 0; D at logical 1; U2 output at a logic,Ll
1.
Co~dition 3 ~ll produce a clock input pulse to U4, thereby
settmg the flip-flop; Q to logical 1.
A fourth condition will occur if either one of the decodels
fails to function properly. Unless the respective 'segmen t
outputs of the two decoders are always at opposite logi;
states, a logical 1 will be produced at the output of one of
the gates U2, causing the flip-flop to set; Q to logical 1.
~ the circuit shown U2 and U3 are combined to form a :~
mput exclusive NOR with open collector limiting resistor:;
R to be as great as 1.1K. A capacitor is included on th(:
clock inp'Ut of U4 to fIlter logic transition spikes off of th(
clock line allowing only a true change in the state of U tc
2
clock th flip-flop. In this circuit, the Q output of thf
flip-flop is used to blank the display. The flip-flop must bf
manually reset; bringing Q to logical O.

5.2.8.1 Suggested
Drive Currents for Stretched Seven
Segment Displays Used in Various Ambient Light
Levels
The level of ambient light and the display contrast fIlter
must be considered together when establishing the proper
LED drive current. The LED drive current must be of
sufficient magnitude to illuminate a display segment so that
it can be easily recognized behind the contrast fJ1ter in the
expected user ambient. As the ambient light level increases,
the LED drive current must also increase to offset any
masking of the display emitted light due to elevated
ambient.
Table 5.2.8.1-1 lists suggested drive currents for 7.62mm
(.3 inch) and 10.92mm (.43 inch) stretched seven segment
displays when used in a given ambient light level with a
specific fIlter. The currents listed should be considered as
starting points and should be adjusted as necessary for each
specific application. The currents listed allow the 7.62mm
devices to be read from a distance of 4 meters (6 meters for
the 1O.92mm devices) in the corresponding ambient.

AMBIENT
LIGHTING
CONulTlON

.4: STD RED

.3 STD RED

LED
DRIVE
MODE

CATEGORy2

CA' EGORy2

AA

CATEGORy2

CATEGORy2

7mA

5mA

3mA

4mA

DC DRIVE

CATEGORy2

.43 GREEN

.43 YELLOW

.43 HER

DIM
(HOIVIE OR LOW
LIGHT LEVEL
CONTROL ROOM)
(10-'100

3.4mA
RH 2423
P60
H1605
3MV
3M655

FILTERS

lux)

DC DRIVE
MODERATE
(TYPICAL
(100--1000

2.1 mA

P27
H1720
3M590

P65
H1670
3M625

6mA

P48
H1440
3M565

11 mA

6mA

STROBED
DRIVE1

OFFICE)

4.9mA

3.8mA

11 mA

8.0mA
P65
H1670
3M625

RH2423
P60
H1605
3MV
3M655

lux)
FILTERS

15mA

P27
H1720
3M590

7.6mA
P48
H1440
3M565

16mA

9mA

BRIGHT
(OUTDOORS,
VERY BRIGHT
OFFICE)
(1000-10,000

14.25 mA

lux)

1 FIGURE

IS AVERAGE

P60
P65
3M625L
3MNDL

P63
3M655L
3MNDL

FILTERS

CURRENT.

13.53 mA

12 mA

ASSUME Ip

2DRIVE CURRENT FOR OTHER CATEGORIES

OF 1.5:1 PER CATEGORY WITH ADDITIONAL
EFFICIENCY.

P27
H1720
3M590L
3MNDL

H1425
3M565L
3MNDL

= 5 x IAVG

HOULD BE VARIED BY A RATIO

CORRECTION FOR RELATIVE

31pk = 4 x IAVG
PANELGRAPHIC

SGL H MALITE

3M COMPANY

P63
P60
P65
P27
P48

H1605
H1670
H1720
H1440
H1425

3MV 3M655
3M625
3M590
3M565
3MND

DARK RED 63
RUBY RED 60
SCARLET RED 65
YELLOW 27
GREEN 48

RED H100-1605
RED H1001670
AMBER H100-1720
GREEN H100-1440
GREEN H100-1425

RH2423 = ROHM & HASS PLEXIGLAS

The monolithic seven segment display represents one of the

largest ~egments of the optoelectronics industry today.
Monolithic LED displays are found in many handheld
calculators and digital watches. Monolithic LED displays
are used because of their low cost, small size, and low
power requirements. Monolithic displays have a very high
sterance which allow them to be used outdoors and in other
areas of high ambient lighting. While monolithic displays
ha ve g~'nerally been used for low cost applications,

(L = WITH LOUVERS)

VIOLET
- RED 655
- RED 625
- AMBER 590
- GREEN 565
- NEUTRAL DENSITY
2423 OR 2444

improvements in character size and viewing angle have also

made monolithic displays suitable for portable instruments
and other high quality applications. Table 5.3 ,1-} compares
the relative merits of a premium monolithic display to
other large seven segment displays that could be used in a
portable instrument application.
Monolithic displays differ from other types of LED displays
in that the individual light emitting segments are formed by
diffusing separate LED junctions on a single chip of GaAsP.
Usually the GaAsP substrate is n doped material and each
LED junction is formed by a p+ diffusion, so monolithic
displays are normally of common cathode configuration.
Although most monolithic displays are constructed with

1) (HARACTER

SI ZE

2) [IIGIT SPACING

3) I\IAXIMUM

VIEWING DISTANCE

4) VIEWING ANGLE

5) TYPICAL

STERANCE

6) POWER REQUIREMENTS

7) D :CIMAL

LOCATION

MONOLITHIC

4.45 mm

(.175 inches)

5.84 mm ( 230 inches)

LARGE SEVEN SEGME

7.62 mm (.300 inches)

10.92 mm (.430 inches)

10.16 mm (.400 inches)

12.70 mm (.500 inches)

~m

4m

6m

380

75

75

300 cd/m2 (90 fL)

@2mA avg'seg 1/5 DF

120 cd/m2 (35 fL)

@15mA avg/seg 1/5DF

130 cd/m2 (38 fL)

@6mA avg/seg 1/5 DF

1.5-2.0 m,\ avg/seg

10-15 mA avg/seg
(Standard Red)

4-6 mA avg/seg
(High Efficiency Red)

RHDp' CDp

RHDp' LHDp

RHDp' LHDp

Colon Jwailable

8) EJ\SE OF USE

Single Multidigit

Package

Inherent Digit Alignment

Inherent Lurr inous
Intensity/Matching
Must be Multiplexed
Common Cathode

Single Digit Package

Single Digit Package

DC or Multiplexed
Operation
Common Anode or
Common Cathode

DC or Multiplexed
Operation
Common Anode or
Common Cathode

the standard seven segment format, almost any number of

segments or character shapes can be diffused into the
GaAsP s bstrate. Figure 5.3.1-1 shows the construction of a
monolithic GaAsP chip. Since several segments are diffused
into a single GaAsP chip, monolithic displays cannot be
constructed
on a GaP substrate. GaP is relatively
transparent to red light, so the contrast between on and off
segments would be inadequate. Because GaAsP is relatively
expensive, most monolithic displays are magnified to keep
chip sizes small, while attaining a viewable character size. In
most cases, the monolithic display is magnified by an
external lens. The resulting apparent character height is
equal to the actual character height times the magnification
of the lens.

MONOLITHIC
GaAsP
N+ EPITAXIAL
LAYER ON
GaAs SUBSTRATE

The external lens has a very important impact in the

appearance of the final display. Besides increasing the
apparent character height, the lens influences the radiation
pattern, the axial luminous intensity, and the viewing angle
of the display. A lens increases the axial luminous intensity
by the square of the magnification:
IV (MAG)

=m

IV (UNMAG)

The addition of a lens may also increase the fresnelloss of

the optical system. Fresnel loss, which is explained in more
detail in Section 2.1.2, is the loss of light due to reflection
as light passes from a medium with one index of refraction
to a medium with a different index of refraction. The
efficiency of light transmission between two mediums can
be expressed as:
4

ELECTRICAL

EOUIVALENT

SEGMENT ANODES
d

dp

T = 2 + "2/"' + ",7"2
where T is the transmission coefficient, nl is the index of
refraction of the first medium, and n2 is the index of
refraction of the second medium.
Lenses for monolithic displays fall into two categories -immersion lenses and non immersion lenses. Immersion
lenses are formed by molding a lens directly over the LED
chip, non immersion lenses have at least one layer of air
between the LED chip and the lens assembly. The axial
luminous intensity for either type of lens is equal to:
IV (MAG)

= ( n"

Tj) m

IV

j=,

where IV (MAG) is the magnified axial luminous intensity,

T. is the transmission coefficient from medium i-I to
niedium i, n is the total number of medium interfaces, m is
the magnification of the lens, and IV is the axial luminous
intensity just beneath the surface of the LED chip.

Figure 5.3.2-1 shows examples of immersion and non

immersion lenses and compares bare chip axial luminous
intensity
to magnified axial luminous intensity for
immersion and non immersion lenses. This increase in axial
luminous intensity allows a reduction in display drive
currents to achieve the same luminous intensity as an
unmagnified display.
The maximum off axis viewing angle at which any display
can be observed is limited either by the radiation pattern of
the display or by the maximum amount of distortion
before the display becomes unreadable. An unmagnified
LED device generally has very close to a lambertian
radiation pattern, that is the luminous intensity varies as
the cosine of the off axis angle:

d~!

J"

G.AsP
LED

T1

APPARENT
CHARACTER
HEIGHT

T3
ACTUAL
T2
CHARACTER
HEIGHT

m' V'IV

FIRST
MEDIUM
GaAsP n ""3.4
Plastic n - 1.5

BARE CHIP:
lV'

1.702) IV

IMMERSION LENS:
IV'= (.850)1.960) m2 IV

Figure 5.3.2-1

= 1.8161

m21V

SECOND
MEDIUM
Plastic
.850

Air

.702
.960

NON IMMERSION LENS:

IV' (.702)1.960)2 m2 IV = (.6471 m11v

Comparison of the Axial Luminous Intensity of Bar

Chip, Immersion Lens, and Non-Immersion Lens
Systems.

where I\' (8) is the off axis luminous intensity, 8 degrees

off axis, ,md IV is the axial luminous intensity.
Thus at Ln angle of 60 (total included angle of 120) th,~
luminous intensity will be 50% of the axial luminou>
intensity. The maximum viewing angle for a display of thi;
type is :imited for a practical purposes to about 75"
because of the requirement that the display be recessed
behind a fIlter. The radiation pattern of a magnified system
mayor may not be lambertian. However, in a magnifiec
display, t he viewing angle is generally limited by character
distortior. The maximum viewing angle of a seven segmenl
display can be defined as the off axis angle when tht
display is rotated that causes any of the segments to narro\'.
or disapp ~ar. The maximum viewing angle can further be
defined as the vertical viewing angle (segments a, d), the
horizonta
viewing angle (segments b, c, e, f), and the
decimal point viewing angle. The maximum viewing angle is
determined by magnification, type of lens, shape of lens,
distance tetween the lens and the monolithic GaAsP chips
and the >pacing between characters. The calculation of
maximum viewing angle is beyond the scope of this text.
Figure
:'.3.2-2 summarizes the relationships between
magnification, luminous intensity, and viewing angle for
magnified displays.
The earlie;t magnified monolithic displays used cast epoxy
immersion lenses. The immersion lens served to increase the
character height of the display, increase the axial luminous
intensity end still provide a very acceptable viewing angle.
Because of the relatively high cost of manufacturing this
type of display, manufacturers began to experiment with
other low~r cost methods to manufacture a magnified
monolithic display. The most commonly used technique
that was d~veloped was to die attach the monolithic GaAsP
chips to some kind of substrate and then fasten an injection
molded lells to the substrate. This technique was able to

reduce the manufacturing cost, improve the monolithic

chip alignment, and improve the optical consistency
between individual lenses in the display. The lens shape has
evolved from spherical immersion lenses to plano-convex
non immersion lenses, and then to positive meniscus non
immersion lenses. The first non immersion lens consisted of
several spherically shaped plano-convex "bubble" lenses
that were molded into a linear array such that each
monolithic digit was magnified by a separate "bubble" lens.
Lens designers discovered that a positive meniscus lens with
two spherically shaped surfaces allowed a much wider
viewing angle than the simple plano-convex lens. The
present state of the art is a positive meniscus lens with a
spherical inner surface and an aspheric outer surface. This
lens design can be optimized to allow almost as large a
viewing angle as an immersion lens with the same
magnification or to reduce off axis distortion of the
monolithic character. An additional cylindrical magnifier
can also be attached over the primary magnifying lens to
further increase character height with the reduction of
some viewing angle.

Monolithic displays can be classified into two basic types

according to whether the lens is an immersion or non
immersion lens. Monolithic displays constructed
with
immersion lenses are manufactured by die attaching the
monolithic GaAsP chips to either a ceramic substrate or to
a lead frame. The die attach pad also forms the common
cathode electrical connection to the monolithic chip. The
aluminum contact for each segment on the monolithic chip
is then wirebonded to the appropriate anode contact. A ball
bond is formed on the LED top contact and a wedge bond
is formed on the lead frame. The completed device is then
encapsulated in epoxy. The magnifying lens is formed
during encapsulation.
Figure 5.3 .3-1 shows the typical
mechanical
construction
of a monolithic display
constructed on a lead frame with a cast epoxy immersion
lens.
Monolithic displays with non immersion lenses are usually
constructed by epoxy die attaching the monolithic GaAsP
chips to a special high temperature printed circuit board.
The length of the printed circuit board depends on the
desired number of digits and the desired digit spacing, since
this type of display is generally not end stackable. Then the
aluminum contact for each segment on the monolithic chip
is wire bonded to the appropriate anode trace on the
printed circuit board. A precision injection molded lens is
then aligned and attached to the printed circuit board using
holes that were drilled during fabrication of the printed
circuit board. This insures accurate alignment of the lens
over the LED chips. A secondary cylindrical lens can also
be attached for added character height. Figure 5.3.3-2
shows the typical mechanical construction of a monolithic

EFFECTS OF MAGNI ICATION ON INTENSITY,

VIEWING ANGLE, AND APPARENT SIZE

Object
Height Y

Lens

Ray Path
With Lens

....----

,....-

....---_

--------- ~,....-

1-'-

-I

-.:::::--"'"

Apparent
Image
With Lens
Height Y'

,- .....

Effects of Magnification:

Figure 5.3.2-2

.....

--- -- ---.,.

Enlar es Apparent Image Size

Incre~ses Intensity
Limits Viewing Angle

@
Y'

= Ym

Iv' = Ivm2
ex: 11m

Effects of Mag. ification on Luminous Intensity

Viewing Angle, and Apparent Character Size.

display manufactured
with these techniques. Displays
constructed with these techniques are commonly available
in digit strings from five to fifteen digits. Table 5.3.3-1
compares the differences between these two packaging
techniques with respect to manufacturing costs, ease of use
by the customer, optics, and package reliability.

Because of their construction, seven segment monolithic

displays are generally connected as an 8xN x-y addressable
array, where N refers to the numer of digits in the display.
Each of the eight anode lines addresses a different segment
(a to dp) and each of the N cathode lines addresses a
particular digit. This allows a five digit cluster to be
constructed on a 14 pin lead frame and a fifteen digit
monolithic display to be constructed with only 23 active
pins. While substantially reducing the number of wires used
to interconnect to the display, it also requires that the
display be strobed on a 1 of N or lower duty factor.

The voltage and current characteristics of each monolithic

LED are similar to the voltage and current characteristics of
a single diode LED lamp. Each monolithic LED is
effectively isolated from adjacent LED segments below the
VBR of the display. Because monolithic displays are
constructed on a GaAsP substrate, they have a very low
dynamic resistance in the forward region. This allows the
monolithic display to be multiplexed at relatively high peak
currents for higher luminous efficiency with forward
voltages typically less than 1.8 volts.
Every monolithic display has upper and lower limits on the
peak and average currents at which the device can be
operated. The maximum peak forward current per segment
has been chosen by the manufacturer to limit the maximum
current density through the semiconductor junction. The
average current per segment is limited by the ambient
temperature and the thermal resistance of the display. In
general, the junction temperature should be prevented from
exceeding the maximum allowable storage temperature for

the particular display. Typically monolithic displays with

cast epoxy immersion lenses are limited to 100C storage
temperature and monolithic displays with non immersion
lenses are limited to 85C storage temperature. Above these
temperatures,
the reliability of the display may be
impaired, or the lens may permanently be deformed. In
most applications, monolithic displays can be driven at
considerably lower average currents than this maximum
average
current. One exception is in high ambient
applications where it is desirable to have a very high
sterance display and a very dense mter to achieve an
acceptable contrast ratio. The minimum peak current at
which the display can be operated is determined by the
junction area of each segment. Below a minimum current
density, the luminous intensity matching between segments
and digits may be unacceptable.
On most monolithic displays, the decimal point is located
either in the center of the digit or in the lower right hand
corner of the digit. For low cost applications, the right
hand decimal point is recommended. The center decimal
point requires a full digit location and time frame to be
allocated to display the decimal. The center decimal point
is an asset in quality instruments because it reduces the
chances of an error caused by misreading the decimal point
location. The center decimal point also allows a slightly
wider viewing angle for the display since the monolithic
chip can then be centered under the magnifying lens.

MONOLITHC
SEVEN SEGMENT
LED CHIP

Most monolithic displays emit light with a peak wavelength

of 655 nm. This wavelength represents the best compromise
between the human eye response curve and the quantum
efficiency of the GaAsP material.
Figure 5-3 3-1

Mechanical Construction
an Immersion Lens.

of a Monolithic

Display

PRIMARY
ASPHERIC
LENS
.

wi--h

Monolithic displays are usually characterized for luminous

intensity under drive conditions that simulate typical usage
conditions.
A graph showing the relative luminous
efficiency vs. peak forward current allows the designer to
determine the time averaged luminous intensity for other
peak and average forward currents. An example of a
"relative luminous efficiency" curve is shown in Figure
5.3.4-1. The time averaged luminous intensity can be
determined for any other drive condition by equation
5.3.4-1:
IVTIMEAVG=
[IPEAK] [DUTY CYCLE] [f/(IPEAK)]

[IV SPEC]

[ISPEC] [DUTY CYCLE SPEC] [f/(ISPEC)]

Figure 5-3.3-2 Mechanical Construction

of a Monolithic
Display
Constructed
on a PC Board With a Non-Immersion
Lens.

where IpEAK is the desired peak current, DUTY CYCLE is

the ratio of time the LED is "on" to total time, f/ is the
relative luminous efficiency of the LED at IpEAK or at
ISPEC and ISPEC and DUTY CYCLE SPEC are the test
conditions
under
which
IV SPEC was originally
characterized.

Lead Frame C nstruction

Printed Circuit Construction

Cast Immersior Lens

Injection Molded Non-I mmersion Lens

2. 3. 4. 5 - en
clusters)
.3" DIP
Wave. or Hand
Vapor Degreasi
Clusters should
Clusters should
Replace Cluster

Special sockets available

Hand Solder
Hand Clean Only
Pre-alignment included
Intensity Matching included
Replace Display

1) MANUFACTURING COST
2) E SE OF USE
a) Digit ~umber
b)
c)
d)
e)
f)
g)

Socketing
Solderability
Cleaning
Display Alignment
Luminous Intensity
Repairability

stackable (multiple

g
be aligned
be matched

3) OPTICS
a) Viewing Angle
Varies slightly ue to lens casting
2.79 mm (.11")

4) RELIABILITY
a) Temp Storage
b) Temp Cycling
c) oisture Resistance

_40C to +100'C
Good
Good
Very Good

d) Mechanical Vibration

Depends on Lens design - can be as

good as Immersion Lens
Precision (may be increased by
Optional Magnified
2.54 mm (.10") to 4.45 mm (.175")

_20 C to +S5 C
Very Good
Condensing atmosphere should be
avoided
Reasonably Good

Table 5.3.3-1 Comparison Between \/Ionolithic Displays Constructed

with Cast Immersion
e nses and Monolithic Displays
with Non-Immersion
e nses.

(SO)(.025)( 1.55)(70)

.... l-./

I--'"

(10)(.20)( 1.00)

V
/

I
I

20

40

Figure 5.3.4-1

60

120
80
'00
IpK - PEAK CURRENT PEA SEGMENT mA

'40

160

'80

Relative Luminous Efficiency vs Peak Current Per

Segment for Hewlett-Packard 5082-7265, 7275,
7285,7295 Monolithic Displays.

For example, suppose a monolithic device is tested for

luminous intensity at 10 mA peak per segment, 20% duty
cycle and the typical time averaged luminous intensity is
70llcdjsegment. If the same display is strobed at 80 mA
peak per egment on a 2.5% duty cycle, the typical time
averaged 1 minous intensity will be:

2(10

Another
representation
of the "relative luminous
efficiency" curve is shown in Figure 5.3.4-2. In this figure,
equation 5.3.4-1 and Figure 5.3.4-1 have been used to
calculate typical time averaged luminous intensities for a
wide variety of drive conditions. This figure graphically
shows the advantage of strobing at higher peak currents to
obtain a higher time averaged luminous intensity for the
same average current. For example suppose a five digit
display is strobed on a 20% duty cycle at 1.5 mA average
current per segment (7.5 mA peakjsegment), then the
typical time averaged luminous intensity would be 50
Ilcdjsegment. This represents a very acceptable luminous
intensity for this type of display under ambient conditions
similar to an office. If the display is strobed at higher peak
currents, the average current per segment can be re uced to
get the same time averaged luminous intensity or the time
averaged luminous intensity can be increased. At 200 mA
peak current per segment, the average current can be
reduced to .82 mA per segment and still obtain the same
50llcdjsegment;
or by maintaining the same average
current, the time averaged luminous intensity can be
increased to 90 Ilcdjsegment.

1!1200UTY)
FACTOR

SI GMENT

STROBE

,
"/

I"

'-

[)

I.~/ III,
7. IIL~
/

;I,

,," ~/~
'11'

Figure 5-~ .42 Typical Time Averaged Luminous Intensity Per

Segment vs Average Current Per Segment for HewlettPackard 50827265,7275,7285,7295
Monolithic
Displays.

Monolitl:ic displays are driven in the same way as any othn

commor
cathode strobed display. The advantage cf
monolithic seven segment displays is that they can be
driven at considerably lower average currents than large
seven ~egment
displays for a satisfactory luminous
intensity For example, a monolithic display designed for 1
low cost calculator can be satisfactorily driven at averag~
currents as low as .25 mA per segment and a somewhat
larger monolithic display such as one designed forL
portable .Jlstrument can be driven at average currents as lovi
as 1 mA per segment. By comparison, a high efficiency re(l
large seven segment display probably requires 4 to 6 ml
average current per segment for a comparable luminou;
intensity. Since the monolithic LED display can be driven
at such low average currents, it is particularly suitable fo'
battery towered applications. Furthermore, a monolithic
LED disp ay can often be driven directly from many readil)
available j ntegrated circuit devices.
Monolithic displays are commonly driven either with digi t
or segmf'nt strobe techniques. When a display is digi:
strobed, each digit is sequentially enabled and at the sam(:
time the appropriate combination of segments is turned on.
For an N digit display, the maximum duty factor for a digi:
strobed display is l/N. If a display is segment strobed, thell
each digit is sequentially enabled and while that digit i::
enabled, each segment is sequentially turned on. Th(
maximurr duty factor for a segment strobed system i:.

1/( 8N). Thus for the same time averaged luminous

intensity, the segment strobed display must drive each LED
segment at a higher peak current where the LED has a
higher quantum efficiency. This allows the display to be
driven at a lower average current for the same time averaged
luminous intensity. While more logic is required for a
segment strobed system, segment strobing may also. b.e
preferable in custom MOS designs because the peak dI~t
driver currents can be reduced. For example, suppose a five
digit monolithic display can be described by the :'rela~ive
efficiency" curve shown in Figure 5.3.4-1. The typIcal tIme
averaged luminous intensity for this display is shown in
Figure 5.3.4-2. When this display is digit strobed at 7.5 mA
peak current per segment on a 1/5 duty factor, the typical
time averaged luminous intensity is 50 J.Lcdper segment.
Each segment driver must be able to source 7.5 mA and
each digit driver must be able to sink 60 mA from the
display. When the display is segment strobed on a 1/40
duty factor, the same time averaged luminous intensity can
be achieved at 39 mA peak current per segment. Thus each
segment driver would be required to source 39 mA and
each digit driver would be required to sink 39 mA from the
display, a 35% reduction in average current per segment.
The primary difference between driving a monolithic
display
with a center decimal point and driving a
monolithic display with a right hand decimal point is that
the segments surrounding the center decimal point should
be blanked when the decimal point is turned on. With a
right hand decimal point, each digit is displayed regardless
of the decimal location. Figure 5.3.5-1 shows a typical
CMOS logic interface to a five digit monolithic display. The
outputs of the National 74C90 counter sequentially enable
each of the five digit cathodes and at the same time enable
circuitry inside the digital subsystem to output the desired
segment and decimal information to the display. If a
display with a center decimal point is used, the proper digit
can be blanked either by outputting a BCD code that is
decoded as a blank state by the seven segment decoder
(state 15 for the National 74C48)
or using the blanking input of the seven segment decoder. A
fixed decimal point can be implemented
simply by
connecting the Dp input to the desired cathode driver.
Then whenever the selected cathode driver is on, the
decimal point will be turned on. Several LSI integrated
circuits are available that combine all the circuitry shown in
Figure 5.3.5-1 into one or a few integrated circuit packages.
5.3.6

Interfacing Microprocessors
Segment Displays

to Monolithic

Seven

Microprocessors can be interfaced to monolithic displays

with only a few external components. In general, these
external components will include one or more latches to
hold the information
to be displayed and additional
circuitry to periodically request new information from the
display. For displays of up to eight digits, either of the

+VDD
OPTIONAL
CONNECTION
FOR CDp

r---------------,
DIGITAL

SUBSYSTEM

NATIONAL
74C48

A
B
BCD + Dp
DATA FOR
SELECTED
DIGIT

DIGIT

SELECT
~DPI

C
D
1

+V

I
I
I
I

4.7K

-1

LT

RBI
A
B

180n

b
d

C
D

BI

270n
HP 5082-7265
HP 5082-7285
(RHDp)

10K

(CDp)

dp

fed

L__-?~!.I~

c b a

BBBBB

(LOW TRUI)

A_L_~<?.~~!.I~~YE~Y.!.X~E_D"p
~

75492

i5-NPN DARUNGTO"Nl
RCA
C[ 4028
I\!ATIONAL
74C90

I
I

TRANSISTORS

I
I
I

1
CLOCK
f ;;;. 500 Hz

2
3
4

5
6

IL

I
I
J

8
9

circuits described in Figures 5.2.6-2 or 5.2.6-3 can be used.

Monolithic displays of up to 16 digits can be driven from
the circuit shown in Figure 5.3.6-1.
This circuit differs from any of the circuits shown in
Section 5.2.6 in the way the decimal point and interrupts
are handled. The microprocessor outputs only a single byte
of information to the display. Four bits contain the
segment and decimal point information and four bits select
the proper digit enable. The upper six states (10 to 15) of
the RCA CD4511 seven segment decoder are decoded as
blank states. Four of these states are decoded externally to
select the decimal point or minus sign. In this example,
states 10 and 11 are decoded as a minus sign, states 12 and
13 are decoded as the decimal point select, and states 14
and 15 are decoded as blank states. Every 66711s, the
refresh clock requests an interrupt from the Motorola 6800
microprocessor. Upon requesting an interrupt, the interrupt
circuitry (not shown) must force a vectored interrupt to
address "REFRESH". The interrupt request can also be
decoded by a software polling program but will increase the
microprocessor time that is required to service the display.

Then the refresh program shown in Figure 5.3.6-lb is

executed. The program reads a pointer from RAM and then
outputs the contents of the memory location specified by
the pointer to the display. Then the pointer is compared to
the address of the fmal byte of information in the data fIle.
If they are equal, the pointer is reset to the address of the
first byte of information in the data me, otherwise, the
pointer is set to the address of the next byte of information
in the me. Since the decimal point is decoded separately
from the other seven segments, a separate time frame is
required to display the decimal point. Thus, a 15 digit
monolithic display with a right hand decimal point will
require 16 bytes of information in the data me. The
microprocessor time required to refresh any display of up
to 16 digits can be determined as shown below:

REFRESH TIME

= --------------

(52N +1)R

MICROPROCESSOR

OR CLOCK RATE

+VOO

+VOO
~A

~On

DO
01
O2
03

C04511
LT
BI
A
B

P
0 1
DO
01 1
O2 2
03
2
C
3

b
c
d

C
0
LE

f
9

3
RCA
CO 4042

33n

+VOO

47.0

RCA

1 4023
3

4.7K
82(1

HP 5082-7275 (COp)
HP 5082-7295 (RHOp)
dp

RRRRR~f:lP

2'h - 754~?

o..Jo..JL...IL...IL-IL-IL..ILJ

15 r-N-;;~A-;~N~~~

2345678

fed

(::Jf:JQQQQ

L.JUL.JL.JUU

10 11 12

13

14 15

TRANSISTORS COMMON1
EMITTER
:

o I
ST
PROPER
n
ADDRESS CODE

ez

MOTOROLA
MC 14501

1
2 I

I
I

04
0
5

3
4

06

5 1----6
7 I

07

1500 Hz
CLOCK
REOUESTS
INTERRUPT
EVERY 667.t{s

I
I

I
I

I
I

8 I
9 I
10 I
11 I
12
13
14
15

RCA
CD 4514L

-l

Figure 5.3.6-1 Interface Between Motorola 6800 and Monolithic

Display.

ADDRESS

OPCODE

REFRESH

LOX
AH
A
LbAA.X

CLOCK
CYCLES

COMMENTS

IX=

POINTER

AA=(IX)

STORES NEW SEGMENT AND DIGIT

IX - (XXIE)16

JUMP IF IX

AA = (10)16

POINTER = (XX10)16

2
10
6

CLEAR INT FLAG

RETURN
POINTER = POINTER

2
10

CLEAR INT FLAG

RETURN

0
STAA
DISPH
DISPL
CPX=
(XX)16
(1E116
BNE
(07116
LDAA=
(10)16
STAA
AH
AL + 1
CLI
RTI
INC
AH
AL + 1
CLI
RTI

INFORMATION

* (XXIE)16

+1

BCD DATA TABLE

AH AL
AH AL

POINTER

(NEXT

TO BE

01 GIT
01SPLAYED)

+1

(XX10)16
(XXll116
(XX12116
(XX13)16
(XX14)16
(XX15)16
(XX16)16
(XX17)16
(XX18)16
(XX19)16
(XX1A)16
(XX1B)16
(XX1C)16
(XX1D)16
(XX1E116

0000
0001
0010
0011
0100
0101
0110
01 1 1
1000
1001
1010
1 01 1
1100
1 101
1 1 10

OUTPU

BCD

POINTERH
POINTERL

"
-,

0000

U
I
I

0001
0010

:3

0011
BCD
DATA

0100

0101

5
,5
-,

0110
01 11
1000

,-,
I

1001

0
CJ

1010

101

Seven segment display devices are capable of transmitting

only a limited number of useful different states of
information. Theoretically, up to 49 different information
states could be depicted utilizing the 7 segment font,
however, less than 1/3 of these have any meaning in
normally utilized communications context. By increasing
the number of elements in a display character from 7 to 9,
it becomes possible to depict in rough form most of the
Roman alphabet and arabic numerals. Utilizatio
of a 14
or 16 segment
character'
allows
better
symbol
differentiation and use of some special symbols. However,
the lower case alphabet cannot be represented, and asthetic
quality and readability are still poor. By increasing the
number of elements still further, it is possible to add
further refmement to the display quality. By utilizing an
infinite number of elements, virtually any 2 dinlensional
character and character shape could be achieved. In
practice, it is found that 35 elements in a 5x7 array or 63
elements in a 7x9 array offer satisfactory resolution for
most alphabetic language communications. The trade-off is
one of the cost of supplying and addressing the individual
elements
vs. the improvement gained from higher
resolution. For LED alphanumeric dot matrix displays, the
industry standard product is the 35 element 5x7 array.

1100
1 101
1 1 10
11 1 1

Figure 5.3.6-1 b Refresh Program Used to Strobe the Monolithic

Display Shown in Figure 5.3.6-1

where N is the number of digits on a CDp display or one

plus the number of digits on a RHDp display, and R is the
strobing rate.
For example, suppose a fifteen digit RHDp display is
refreshed at a 100 Hz refresh rate and the microprocessor
uses a 1 MHz clock, then 8.3% of the microprocessor's time
is required to service the display. For the remaining 91.7%
of the time, the microprocessor can be performing other
functions. While this example used a Motorola 6800
microprocessor, this technique can be used with any other
microprocessor by modifying the hardware and software to
reflect a different instruction set and different ways to
output information to peripheral devices.

Figure 5.4.1-1 depicts schematically one digit of a 5x7

array. The row lines are tied to the anodes of all of the
diodes of a particular row and are designated by the Roman
numerials I-VII. The column lines tie together the cathodes
of each diode of a particular colunm and are designated
A-E. In multi digit arrays, the columns of digit one are
designated lA-IE, digit two 2A-2E, etc. The row lines are
extended in common to all digits in an array. The effect of
this type of organization is to form a 7x5N crosspoint
matrix (N = number of characters) of diodes where an
individual diode is energized by selecting the appropriate
row an~ column lines. Using this technique, a 4 digit array
of 140 diodes may be addressed using only 27 lines (7 row
lines + 4x5 digit lines). If decoding or data storage circuits
are included inside the display package, the external pin
count can be even further reduced.

The generation of character information in a 5x7 array is

somewhat more complex than the techniques utilized for 7
segment displays. In the 7 segment display, all elements of a
digit can be simultaneously addressed so that the entire
character is formed during one address cycle. In a 5x7
array, each character must be made up of 5 or 7 subsets of
data which are presented during sequential address cycles.

1A

18

1C

1D

1E

00000
.000.
00000
00000
00000
00000
00000

) )

00000
00000
00000
00000
.000.
00000
00000

000.
.000.
.000.

.000
00000
00000
00000
00000
00000
00000

.000.
.000.
.000.
II --(

00000
00000
.000.
00000
00000
00000
00000

By the nature of its electronic organization, then, a 5:0

display character must be operated on a strobed basis.
Operating the display on this basis, though it increases the
system Gomplexity somewhat, does offer advantages whi(;h
clearly outweigh the negative aspects, even for a single
character display.

A minimum number of interconnects are used 10

interface between the data source and the displa:r.

The cost of the character generation and associated

C rive circuitry is shared over many digits.

00000
00000
00000
00000
00000
00000

00000
00000

00000
00000

00000)
00000
00000
.000.
00000

00000)
00000
00000
00000
.000.

.000.
.000.
.000.

.000.
.000.

.0000
.0000
.0000
.0000
.0000
.0000
.0000

00000
00000
00000
0.000
00000
00000
00000

00000
00000
00000
00.00
00000
00000
00000

00000
00000
00000
000.0
00000
00000
00000

0000.
0000.
0000.
0000.
0000.
0000.
0000.

.000.

Clock, timing and data storage elements require1

hr the strobed display can be shared with other
portions of the system.

There ar~ two common methods for addressing the diodes

in the 5x 7 array in order to generate character information.
Both of these methods depend on the eye's response to a
strobed c.isplay. Figures 5.4.2-1 a,b depict the sequence 0 r
steps necessary to form the character "H" using either rov'
or columa strobing techniques.

Fig. B.
Figure 5.4.2-1a,b

5.44

..

Column Strobing

Character Generation
Strobe Methods.

Using Row (a) or Column

(b)

In row strobing, data for one row of the desired character is

applied to the column lines and the row is then energized.
This process is repeated for each of seven rows. For row
strobing of a multicharacter string, row data for each
character in the string is decoded and stored in a data latch
associated with each character. The row line is then
enabled, displaying simultaneously one row of data for all
of the characters in the string.
In column strobing, data for one column of the desired
character is supplied to the row lines and the column is
then energized. The process is repeated, varying the data,
until all of the columns of the character have been selected.
Multicharacter displays are treated as simply an extension
of the single character case such that the number of subsets
of data generated will be equal to 5N, where N is equal to
the number of digits in the display.
If either of these sequences is repeated at a rate which
insures that each of the appropriate matrix locations is
re-energized a minimum of 100 times per second the eye
will perceive a continuous image of the entire character.
The apparent intensity of each of the display elements will
be equal to the intensity of that element during the "ON'
period multiplied by the ratio of "ON" time to REFRESH
PERIOD. This ratio is referred to as the display DUTY
FACTOR.

From a conceptual standpoint, column scanning is the

easiest method to understand; however, it is more limited as
to the number of digits which may be displayed using only
one decoder. The block diagram in Figure 504.2-2 depicts
the elements of a column strobing system Circuit operation
is as follows:
Coded character information from the keyboard is first
stored in the input storage buffers, generally six bits per
character. The timing circuitry then selects the first
character word, from storage buffer number one, presenting
this data to the seven line output ROM. The timing
circuitry also inputs the column one select code to the
ROM causing the ROM output states in turn to present
column I, character I data at its output terminals. These
states turn on the row drivers for the appropriate diodes in
the first column of the display. The first column is then
enabled by its column driver for an appropriate "ON"
period. The second column data is then selected and
displayed, etc., until all five columns of the first character
have been displayed.
The timing circuit then selects the data for the second
character from buffer number two, and displays each of the
columns. This procedure is repeated until all columns of all
characters have been displayed. If each column is refreshed
at 100 times per second, the eye will not perceive flicker in

I
I

KEYBOARD

ARRAY
SELECT

I
I

INPUT STORAGE
BUFFER NO.1

INPUT STORAGE
BUFFER NO.2

-1

INPUT STORAGE
BUFFER NO.3

1 I
1 I

CLOCK &
TIMING

CHARACTER
DATA IN
COLU MN
DRIV ERS

..L

10

I
I
I

I
I

I
I
I

I
I

I
I

I
I

I
I
I

11 12 13

14 15

2-

COLUMN
SELECT

DECODED
DATA OUT

ROM

.!..L

ROW
DRIVERS

2..

LED ARRAY
# 1

Figure 5.4.2-2

Block Diagram of a Column

Display.

Strobed Alphanumeric

-..-I

LED ARRAY
#2

-I

I
I

I
I
I

I
I

LED ARRAY
#3

the display. Herein lies the problem with column strobil1g.

Since the maximum peak current per diode is limited to
about 100 mA and the desired minimum average curwnt
per diJde is around 1 mA, the maximum number of
columrs which could be displayed is 100 columns or 20
characters. Given the practicalities of timing and switchi 19,
this number is more realistically 16 characters.

the five bit output storage buffer No.1.

Each of the
remaining words are selected, decoded into first row
character information and loaded into the five bit output
storage registers corresponding to each of the characters of
the display. At this point in the sequence, each of the five
output storage buffers has been loaded with the appropriate
display information for the first row of each character. The
timing circuitry now turns on the top row driver so current
flows through and lights all of the appropriate LEDs in the
top row of all of the characters. The complete cycle is
repeated to decode and display row II, then row III, and so
on through the seventh row. Repeating this scan at a rate
above 100 Hz in most applications gives a flicker free
alphanumeric character at each of the digit locations. The
only practical limitation to the number of characters which
can be operated from a single character generator in this
mode is imposed by the maximum clock rate of the logic.
In general, character string lengths well in excess of 100
digits should not pose a problem.

The row strobing scheme eliminates the problem of h gh

peak currents and consequent severe limitations on the
numbeJ of digits which can be strobed in one display string
utilizing only one decoder. However, there is a substant ial
increase in circuit complexity. The block diagram for a row
strobing scheme is shown in Figure 5.4.2-3.
Coded alphanumeric information is sequentially entered
and stored in the input storage buffers. Next, the timil1g
circuitry enables the ROM and selects data word No.1. An
additional input code selects row I output data from t e
ROM.'
e first row of character information is stored in

6 LINE ASCII
T

11

ARRAY SELECT
1

r--

~
[

MASTER
CLOCK

11

CHARACTER
DATAIN

GriMING
CIRCUITRY

6 BIT INPUT STORAGE BUFFERS

READ ONLY
MEMORY

ROW
SELECT
DATA
/

5 BIT OUTPUT ~TORAGE BUFFERS

D ECODED
DATAOUT

I
-

COLUMN
DRIVERS

I
LED
DISPLAY

f----

f------

f----

r-LED
f-----JISPLAY r--

f-----f------

!---

I
2

I
f----

ROW
l;:RIVERS

I
r-f----

r-LED
DISPLAY c--r--

f------

c---

Figure 5.4.2-3 Block Diag am of a Row Strobed Alphanumeric

Display.

LED
DISPLAY

LED
DISPLAY

5.4.3

Implementation
Display

of

a 16 Character

Row Scan

The circuit
Figure 5.4.3-1 depicts a 16 character
alphanumeric display utilizing the row strobing technique.
One 7496 five bit shift register is used as the output data
storage buffer for each character of the display. Data from
the RO is loaded into a parallel to serial shift register and
then shifted into the data storage buffers on each positive
transition of the clock. After five clock pulses a new input
data word is selected from the ASCII data storage RAM
(1 ~ Fairchild 93403). After all sixteen words (one word
per character) have been decoded and shifted into the
output data storage buffers, the shift register clock input is
gated off and the row line is enabled to light the display. At
the end of the display period, the row is turned off, the row
counter is incremented by one count, and the second row
of data is loaded into the output storage buffer. The row
scanner is then enabled while the shift register clock is again
gated off. This process is repeated for all rows of the
display.
The high true output from the parallel to serial shift register
is inverted before loading into the output data buffers. The
shift register outputs will then be low for a dot "ON"
condition. The output low state of the 7496 is rated to sink
16 mA and can, therefore, be used to directly drive the
display columns at 16 mA peak. A 500 KHz input clock
rate will give a refresh rate of about 93 Hz. The DUlY
FACTOR expressed in Equation 5.4.3-1 is set by the ratio
of the loading time to the display time multiplied by 11m
where m is equal to the number of rows to be addressed;
seven in this case:

24
D.F. = (1-7)
2

-;:;; =

12.5%

The circuit in Figure 5.4.3-1 utilizes a shift register to store

decoded row data during the display cycle. If the shift
register and current limiting elements are included along
with the display matrix in a single hybrid package,
significant improvements are realized in the total cost of
display implementation,
packing density, and reliability.
The HP HDSP-2000 display is designed to provide on-board
storage of decoded column data plus constant current
sinking row drivers for each of the 28 rows in the 4
character display. This approach allows the user to address
e a c h d i splay
package
through
just
11 active
interconnections
vs. the 176 interconnections
and 36
components required to effect a similar function using
conventional LED matrices.

Figure 5.4.4-1 is a block diagram of the in ternal circuitry of

the display. The device consists of four LED matrices amI
two l4-bit serial-in-parallel-out shift registers. The LED
matrix for each character is a 5x7 diode array organized
with the anodes of e:ich column tied in common and the
cathodes of each row tied in common. D,e row cathode
commons of each character are tied to the constant current
sinking outputs of 7 successive stages of the shift register.
The like columns of each of the 4 characters are tied
together and brought to a single address pin (i.e., column 1
of all 4 characters is tied to pin 1, etc.). In this way, any
diode in the four 5x7 matrices may be addressed by shifting
data to the appropriate shift register location and applying
a voltage to the appropriate column.
The two on-board
shift registers act as a single
serial-in-parallel-out (SIPO) register. The SIPO shift register
has a constant current sinking output associated with each
shift register stage. The output stage is a current mirror
design with a nominal current gain of 10. The magnitude of
the current to the reference diode is established from the
output voltage of the VB input buffer applied across the
current reference resistors, R. The reference current flow is
switched by a transistor tied to the output of the associated
shift register stage. A logical 1 loaded into the shift register
will turn the current source "ON" thereby sinking current
from the row line. A voltage applied to the appropriate
column input will then turn "ON" the desired diode.
Data is loaded serially into the shift registe: on the high to
low transition of the clock line. The data output terminal is
a TTL buffer interface to the 28th bit of the shift register
(i.e., the 7th row of character 4 in each package). The Data
Output is arranged to directly interconnect to the Data
Input on a succeeding 4 digit display package. The Data,
Clock and VB inputs are all buffered to allow direct
interface to any TTL or DTL logic family. The display is
organized so that column strobing is utilized in the same
manner as in a row strobed system; (i.e., the data for all of
the like columns in the display string is loaded into shift
register and then the column is energized to generate
character information).
For a four character display, 28 bits representing the first
column of each of the four characters are loaded serially
into the on-board SIPO shift register and the first column is
then energized for a period of time, T. Tills process is then
repeated for columns 2 through 5.
If the time required to load the 28 bits into the SIPO shift
register is t, then the duty factor is:

D.F.

T
5(t + T)

4.7K

(7) 7406
Ao

00

TMS2400 AI

01

ROM TI

ASC II

I~~~

S:~E~T

A2

02

A3
A4

03
W

X
y
Z

AS

AO

01

AI

02

A2

A2 03

A3

FAIR

93403
AI
Ao

A3

ASC II

DATA
INPUT

AO AI

00

(1/2)7407

DO

A2A3

FAIR

01 93403

IOF7

DECODER
9301

3
4

HP 5082-7101
(4)

5
ENABLE
u

1;
'7

IN

IN
ClK

QA

B
ClK

ClKC

QB

74163

Qc
QD

D
CARRY

OUT

ABC
Enp 74163

elK

VCC

GND

_DATA

IN

~nlnrl
I I I T

1
1rl 1
I I I r

1 rl

'1 1 '1

rJ

14-81T SERIAL IN
PARALLEL OUT
SHIFT REGISTER

?/
~

V
1
1 rl 11 fJ
11 t111 fJ
rJ
I

"1 1 h 1
I T T T

r~
\s
\s f r~'I'f f rf

11 11T 1T 1T hT

10

11

'1

1111T 1T 1f hIf

.\S

'1

II'[ T
~ T T I rf

12~

*
f
?/ "f
.\;
6

~1'[ T
1 1

r1 T
1

rf

11

h
11 11r "1T 1T IT
r

11 'l

11

'lh

T 1 T

'I
t11tlT '-]T lh
IIII
1 T
1 I I I~
r r
I

~if

b. b. 'l
T

I T

I 1

L-

C~~

5 x 7 LED
MATRIX

LV", 71~

Irflf ~f

5 x 7 LED
MATRIX

rlrllln

f\s

11ilJ.
I

5 x 7 LED
MATRIX

L 7"'7 I~
1

14-81T SERIAL IN
PARALLEL OUT
SHIFT REGISTER
5 x 7 LED
MATRIX

VB

_CLOCK

V.
rJ
3

I~~~[ff
.I\s
\s f I~'f f
~

r~ I~'f 'f

f rf I~'f 'f
"f\s I~
ff r~:f 'f

10

V.
V ..I\s tl
V "I\s i*f rr
kf, .\;
5

r r r

1 I I

'X 'I X

T T T I

L-

-I

COL3

COL4

COL5

Figure 5.4.4-1 Block Diag am of the HDSP2000.

the term 5(t+T) is then the refresh period.

satisfactory display, the refresh period should be:

For

IN:AL
CONNECTIO

a
(T+t) - (#of bits to be loaded) (1/1 MHz)
5(tH)
(2 ms) - 700 (1 ~s)
5x 2ms

Two milliseconds then is the maximum time period which

should be allowed for loading and display of each column
location. For t~,
the duty factor will approach 20%.
The number of digits which can be addressed in a single
string is then dependent upon the minimum acceptable
duty factor and the choice of clock rate. For instance, at I
MHz clock rate, a 100 character string of 25 packages could
be operated at a duty factor of

For most applications, a duty factor of 10% or greater will

provide more than satisfactory display intensity. In brightly
illuminated ambient environments, a larger duty factor may
be desirable whereas, in dim ambient situations, the duty
factor may have to be reduced in order to provide a display
with satisfactory contrast.

enabling the output of the 1/5 column select decoder and

disabling the clock input to the display. The information
now present in the shift registers will be displayed for a
period, T, at the column 1 location. At the end of the
display period, T, the divide by 5 counter which provides
column select data for both the display and the character
generator is incremented one count and column 2 data is
then loaded and displayed in the same manner as column 1.
This process is repeated for each of 5 columns which
comprise the 5 subsets of data necessary to display the
desired characters. After the fifth count, the 1/5 decoder
automatically resets to "one" and the sequence is repeated.
The only changes required to extend this interface to
character strings of more than 4 digits are to increase the
size of the refresh memory and to change the divide by four
counter to a modulus equal to the number of digits in the
desired string.

A practical display system utilizing the HP HDSP-20 0

display requires interfacing with a character generator and
refresh memory. A block diagram of such a display-syst'lm
is depicted in Figure 5.4.4.1-1. In explanation, assume that
this system is for a four character display. Therefore, the
l/N co nter becomes a 1/4 counter where N is equal to the
number of characters in the string.
The refresh memory is utilized to store the information to
be displayed. Information can be coded in anyone
of
several different standard data codes, such as ASCII or
EBDIC, or the code and the display font can be customized
throu h the use a custom coded ROM. The oIlly
require ent is the output data be generated as 5 subsets of
7 bits each. The character generator receives data from the
refresh
memory
and outputs 7 display data bits
corresp nding to the character and the column select d:.ta
input. This data is converted to serial format in the parallel
to serial shift register for clocking into the display shift
register. In the typical system, the right most character to
be disp ayed is selected first and the data corresponding to
the O~' and OFF display elements in the fust column is
clocked into the first 7 shift register locations. In a similar
manner, column 1 data for characters 3, 2, and 1 is select~d
by the I/N counter, decoded and shifted into the disphy
shift register. After 28 clock counts, data for each character
is located in the shift register locations which are associat ~d
with the 7 rows of the appropriate LED matrix. The I, N
counter overflows, triggering the display time count<:r,

STEM
OCK
IN

RESET

.....--

Since data is loaded for all of the like columns in the

display
string and these columns are then enabled
simultaneously, only five column switch transistors are
required regardless of the number of characters in the
string. The column switch transistors should be selected to
handle approximately 110mA per character in the display
string.
The collector
emitter
saturation
voltage
characteristics and column voltage supply should be chosen
to provide a 3.0V';;;; VCOL .;;;;V CC. To save on power
supply costs and improve efficiency, this supply may be a
full wave rectified unregulated dc voltage as long as the
PEAK value does not exceed the value of V CC and the
minimum value does not drop below 3.0 volts.

CLOCK

IN

77

CI.OCK
IN

OUT

IN

f5

~
~

CLOCK

7.lINE

OUT

(COLUMN)

CLOCK

DATA
IN

IN

SELECT

DATA

IN

GENERATOR

TRANSISTORS

Sf LECT INPUTS

IN

COLUMN TOENABLE

DATA
OUT

HD:;P2000

COLUMN

OUT

COLUMN

DATA

(~

IT)

I I II II
ASCII DATA

CHARACTER

IN

DISPLAY TIME
COUNTER

COLUMN SELECT
COUNTER
75

REFRESH
MEMORY

:OLUMN
7BIT
PARALLEL
IN
SERIAL OUT
SHIFT
REGISTER

START

OUT

II II

IN

CLOCK

1/N COUNTER
No. OF DIGITS
IN DISPLAY
STRING

AOC"
DATA

1
f2
f3
f-

!~
----<

1/5 DECODER

ENABLE

f--

Figure 5.4.4.1-1

Block Diagram of the Support

for the HDSP-2000.

Electroni

Since large current transients can occur if a column line is

enabled
during
data shifting operations, the most
satisfactory operation will be achieved if the column
current is switched off before clocking begins. ICC will be
reduced by about 10-15% if the clock is held in the logical
1 state during the display period, T.

There are many possible practical techniques for interfacing

to the lIP HDSP-2000 alphanumeric display. Two basic
approaches will be treated here.

The circuit shown in Figure 5.4.4.2.1-1 is for a 16 character

display and is designed to function primarily as a readout
for general instrumentation systems. CMOS logic circuitry
is utilized in this design; however, it should be a simple
exercise to substitute TTL functions if CMOS is not
desired. In this circuit, a CD 4022 and CD 4520 are
combined to perform the functions of the divide by 7,
divide by 16 (l/N) and display time counters as depicted in
Figure 5.4.4.1-1. The timing diagram, Figure 5.4.4.2.1-2,
demonstrates the relationship of the various critical outputs
and inputs. The CD4022 actually acts here as a divide by 8
counter with the first count used to latch data into the
parallel-in-serial-out (PISO) shift register and the other 7
counts shifting data out of the PISO and into the display
shift register. The CD4520 is a dual 4 bit counter wired as
an 8 bit binary ripple counter. The NAND gate, U 1'
establishes the ratio of loading time to displ~ time. In this
case, loading will occur once in every 8 x 2 clock counts
for a period of 8 x 24 clock counts. Duty factor is then
from Equation 5.4.4-1
(ax27)

(ax24)

D.F. =-----5

(ax27)

=17.5%

The four least significant bits of the CD4520 counter are

used to continually address the CD4036 refresh memory.
Data can be written into the desired memory address by
strobing the WRITE ENABLE line when the appropriate
memory address appears on the WRITE ADDRESS lines.
This fWlction can occur simultaneously with a read from
memory.
Two counters, a CD4029 and a CD4022, are used for the
column data generator and the column select decoder,
respectively.
Note that the Signetics 2516 character

generator requires column select inputs of binary coded 1

to 5 instead of 0 to 4. For this reason, the CD4029 is preset
to a binary 1 by the same pulse which is used to reset the
CD4022 column select decoder. To minimize ICC' the VB
terminal is held low during data load operations, turning
"OFF" the current mirror reference current. The column
current switch is a PNP diulington transistor driven from a
buffered NAND gate. The 1N4720 serves to reduce the
column voltage by approximately 1 volt, thereby reducing
on board power dissipation in the display devices. Due to
maximum clock rate limitations of the CMOS logic, clock
input should not exceed 1 MHz.

The HDSP-2000 alphanumeric display is ideally suited for

interface to a microprocessor. There are several different
ways in which the hardware/software partitioning can be
arranged in such a system. The choice of the technique will
depend on how much of the microprocessor time the
designer wishes to devote to supporting the display. The
additional microprocessor software time required is traded
off against
additional
hardware
costs.
Figures
5.4.4.2.2-1a,b,c
illustrate three different partitioning
techniques which can be utilized in interfacing the HP
HDSP-2000 to a microprocessor.
In 5.4.4.2.2-1a, the
display and its interface hardware appear as an autonomous
peripheral to the microprocessor. This system accepts data
into a local RAM whenever the microprocessor updates the
information to be displayed. A good example of this type
of system is illustrated in Figure 5.4.4.2.1-1. This system,
however, duplicates some of the functions available in the
microprocessor system.
Figure 5.4.4.2.2-1 b illustrates a technique in which the
coded data RAM and character generator ROM have been
removed from the display interface and included as a
fraction of the microprocessor hardware. In this approach,
the entire message is decoded and the data is sent to the
display where it is stored and accessed by the display
scanning
logic.
This approach is ideal where one
microprocessor system may be utilized to service several
displays.
The display interface may be even further simplified by
developing a system in which the microprocessor supplies
the display refresh information periodically in response to
an interrupt request from the display. The lIP HDSP-2000
due to the onboard storage of decoded data, is ideally
suited to the implementation .of this third type of interface.
Depending on the number of digits in the display string,
and the microprocessor clock rate, the display can be
updated at a time burden to the microprocessor of less than
1% per digit. Figure 5.4.4.2.2-2 illustrates the practical
implementation
of the technique presented in Figure

ASC II

WRITE
ENABLE

WRT

C.
INH

-=-

1
2

MEM.
BYPASS

3
4
5

BITS
IN

CD4036

-b
"REFER

TO TIMING

DIAGRAM,

WRITE
ADDRESS
!

FIGURE 4.
I

CD4036

l"

CD4036

T
.J.

CD4036

OUT

ClK

"0"

ClK

EN

-=NODE"
C

1/4 CfJ4001

A alA

Vcc-

EN A

01C
CD4520
OlD
EN B

CD4022

BRESET
ClK

l'/6
CD4049

'"""""~

1/6

BINI
DEC

UI

COL. 2

COL. 3

COL. 4

~~1N4720
Vcc

110F
I

a;

PRESET

03

J F

RESET

ClK

COL. 5

CD4022
0

5
3

_~_I_J_I

1/6 CD4050
5
(TYP.)

L ___________

-I

1/4

CD4011

.J

r
C.EN A.
A,
A2
A3
2

2
1

1
T

___~__ J ___~___ I
I
I

- -=-

.13 .I.
a

CD4029

ClK
EN

HP HDSP2000 (4)
16 DIGITS
COL. 1

.I, .1

VB

DIN

T T I I T 1 1

~Vcc

CLK

1/4
CD4011

1/6 CD4049

1/4 CD4011

-.

-. -.
QY
ClK

1/6 CD4049

A, C.EN

1/4
CD4011

1/4
CD4011

114
CD4011

NODE"
B
NODE"
D

AO

fq0

F""'~~
Q2C-

Ao A, C.EN

A, C.EN

01B

1/6 CD4050

~~

6
AO

AO A, C.EN

NODE"
A

-=-

As

1/6 CD4049 ~

1/3 CD4023

A7

As

Ag
Vcc c--

SIG.2516
0.

5
6
CD4014

PIS
Os CONTROL

+5V

Voo c-- -5V

VGG c-- -12V
Os Os 07 OS

I
SER.
IN

As

ClK

Display strings of other lengths can be accomodated by

changing the number of words which are written to the
display from the RAM. This involves a simple one
instruction software change.

5.4.4.2.2-1c, interfacing a 16 digit string of HP HDSP-2000

displays to an 8080A microprocessor. In this circuit, an
astable timing element generates an interrupt request once
every 2 msec. The interrupt subroutine outputs 16 bytes of
data to the display via a parallel to serial converter. A local
clock shifts this data into the display between output cycles
of the microprocessor. Only 7 bits of each output word are
shifted to the display. A column select word is sent to a
separate address. Prior to data loading, this word is used to
turn all column enable transistors off. After all of the data
has been sent, this word gates the appropriate column data
"ON". Decoded data for the display is stored in a memory
stack in the microprocessor RAM. A memory pointer in the
programmed
subroutine
is utilized to indicate the
appropriate block of data to be sent to the display.

In this situation, the microprocessor load time can be

calculated from Equation 5.4.4.2.2-1 using

where:

For a display refresh rate of 100 Hz (interrupt every 2

msec), the microprocessor time required to service the
display as a percent of total time is expressed in the
following equation:

Microprocessor clock frequency

Number of digits in the display !;tring

tLOAD

% tSERVICE=

tlNTERRUPT

X 100%

518 x .5 /ls

2 ms
= 12.55%

23456
8 x 27 CLOCK

COUNTS

REPEAT
FROM
ZERO

Figure 5.4.4.2.1-2

Timing Diagram for the Circuit Depictel

Figure 5.4.4.2.1-1

in

REPEATS 16 TIMES
TO LOAD 16 COLUMNS
OF 7 BITS EACH

I
RELATIVE

COST -

RELATIVE
MICROPROCESSOR TIME
BURDEN

I
I
I
I
I
I

GREATEST

LEAST

o.

CODED
DATA
RAM

CHARACTER
GENERATOR
ROM

DISPLA

SCANNING
LOGIC

DISPLAY

INTERFACE

I
MP

RAM
RELATIVE

COST-INTERMEDIATE

RELATIVE
MICROPROCESSOR T I ME
BURDEN

MODERATE

DISPLA

SCANNING
LOGIC

RELATIVE

COST-

LOWEST

RELATIVE MICROPROCESSOR TIME

BURDEN ----LARGEST

c.
Figure 5.4.4.2.2-1a,

I
I

PARALLEL

TO

SERIAL
CONVERTER

~J~------,
I

INTERRUPT
GENERATOR

b, c Mic oprocessor Interface Partitioning

Tec niques for Implementation of Display
USillg the HDSP-2000.

IIS
ClK

74LSI12

ADDRESS CODE
FOR OUTPUT (n)
OUT FROM eOeOA
STATUS LATCH

1/2

1.5 MHz

CIR

CLOCK
I INPUT

+v eOL PNP

,-DATA BUS,

% THRU

DARLINGTON
TRANSISTORS WITH
COMMON EMITTER

-1

COLUMN I

COLUMN 2

COLUMN 3

ADDRESS CODE
FOR OUTPUT (I)
OUT FROM 8080A
STATUS LATCH

COLUMN 4
COLUMN 5

I
L
Figure 5.4.4.2.2-2

..J

Practicle Interface Between the HDSP-2000 and

an 8080A Micjroprocessor.

CLOCK
CYCLES
(0038)16

PUSH PSW
PUSH HL
ORI
(FF)16
OUT
I
LHLD
AL
AH
MOV A, M
OUT
n

LOOP

(7)

A ~ (FF)16

._---(10)
(16)
------

POINTER

-----(7)

(10)

A ~ (HL)

DIGIT 16 ~ A
HL~HL+1

(7)

A~(HL)

(10)
(5)

HL~HL+1
A ~ (H2)

(10)

-----(5)

INX HL
MOV A, M
OUT

(10)

------

(7)

DIGIT 14~ A

HL~HL+1
A~ (HL)
DIGIT 1 ~ A

(5)

HL~HL+1

(7)

A~

(10)

DISPLAY)

DIGIT 15 ~ A

(7)

INX HL
MOV A, M
OUT

TO HDSP20 00

(5)

------

eONTENTS (DTDO)

AH Al

POINTER l

AH Al + 1

POINTER H

(XX10)16

X R1 R2 R3 R4 R5 R6 R]

(XX1F)16
(XX20)16
(XX21 )16

X R1 R2 R3 R4 R5 R6 R]
X X X 1 1 1 1 0
X R1 R2 R3 R4 R5 R6 R]

DIGIT 1, eOl 1
eOl1
ENABLE
DIGIT 16, eOl 2

(XX30)16
(XX31)16
(XX32)16

X R1 R2 R3 R4 R5 R6 R]
XX X 1 1 1 0 1
X R1 R2 R3 R4 R5 R6 R]

DIGIT 1, eOl 2
eOl2
ENABLE
DIGIT 16, eOl 3

(XX41)16
(XX42116
(XX43)16

X R1 R2 R3 R4 R5 R6 R]
XX X 1 1 0 1 1
X R1 R2 R3 R4 R5 R6 R]

DIGIT 1, eOl 3
eOl3 ENABLE
DIGIT 16, eOl 4

(XX52)16
(XX53116
(XX54116

X R1 R2 R3 R4 R5 R6 R]
XX X 1 0 1 1 1
X R1 R2 R3 R4 R5 R6 R]

(XX63)16
(XX64)16

X R1 R2 R3 R4 R5 R6 R]
XX X 0 1 1 1 1

(POINTS AT N EXT
WORD TO BE SENT

------

ADDRESS

TURNS OFF COLUMNS

------

INX HL
MOV A, M
OUT

INX HL
MOV A, M
OUT
I
MOV A, L
CPI
(64)16
JNZ
(ADDRESS
OF LOOP)
MVI L
(OF)16
JNX HL
SHLD
AL
AH
POP HL
POP PSW
EI
RET

COMMENTS

(11)
(11)

(HL)

TURNS A COLUMN ON

-----(5)
(7)
------

(10)

-----------

A~L
COMPARE L TO ADDRESS
OF LAST MEMORY LOCATION
JUMP IF A (64)16

(7)

-----(5)
(16)
------

-----(10)
(10)
(4)
(10)

Figure 5.4.4.2.2-2

Contim ed. Microcade and RAM Contents

Used to Support Microprocessor Display
Interface.

DIGIT1,eOl4
eOl4 ENABLE
DIGIT 16, eOl 5

BASIC
DEVICE
NUMBER

DEVICE DESCRIPTION

IMA)

mA

MANUFACTURER
FAIR, MOT, TI
NS, TI
NS
ITT

491

50 (si k/source
25 (scurce)
19(sc urce)
40(sc urce)
34(sc urce)
18(sc urce)
16 (s( urce)
14.5 source)
12 (scurce)
10 (scurce)
50 (scurce)

75492
75494
8870
8877
8892
500
502
506
510
492

250 ( ink)
150 ( ink)
350 ( ink)
50 (si ,k)
200 ( ink)
250 ( ink)
200 ( ink)
200 ( ink)
160 ( ink)
250 ( ink)

FAIR, MOT, NS, TI

NS, TI

7-Digit Driver, MOS to LED

Cathode

75497
8866
546
552
554
556

150 (
50 (si
50 (si
500 (
500 (
500 (

ink)
k)
k)
ink)
ink)
ink)

TI
NS
ITT

8-Digit Driver, MOS to LED

Cathode

8863/8963
8865
8871
514/525

500 (
50 (si
40 (si
40 (si

ink)
k)
k)
k)

NS

526
548

40 (si k)
60 (si k)
40 (si k)
150 (~ink)
50 (si k)
50 (si k)

ITT

Quad Segment Driver, MOS to

LED A ode

HEX Digit Driver, MOS to LED

Cathode

9-Digit Driver, MOS to LED

Cathode

75491
75493
7895/8895
501
503
507
517
518
522
523

558
75498
8855
8874/8876/
8879
8973/8974/
8976

100 (l ink)

12-Digit Driver, MOS to LE 0

Cathode

8868
8973

110 Is ink)
40 (sil k)

Segment Driver, MOS to LED

8861
(5-Seg)
8877
(8-Seg)

50 (source)

TABLE 5.5-1

ITT

ITT

TI
NS

NS

NS
14 (soIJrce)

List of LED Displa Iv to Logic Interface Devices

5.57

IOL T mA

FONT: 6 AND 9
WITH OR

SOURCE/SINK

WITHOUT TAILS

BASIC
DEVICE
NUMBER

DEVICE DESCRIPTION
BCD to 7-Segment Decoder/Driver
OUTP TS: Active High, Interal
Resistive Pull-up

6.4 (sink)
2 (s )urce)
6.5 (sink)
2 (s,)urce)
6 (s nk)
2 (st)urce)
15 (sink)
4 (st)urce)

W/O

7856

FAIR, ITT, MOT, NS,

SIG, TI
TI

W/O

NS, TI

W/O

SIG

7.5 (source)

W/O

NS

7857
7858

60 (rource)
50 (lource)

W/O
W/O

NS
NS

7449
74LS49
74249
74LS249

10 (link)
8 (sink)
10 (;ink)
8 (sink)

W/O
W/O
W
W

FAIR,MOT
NS, TI
TI
TI

7447

40 hink)

W/O

74L47
74LS47
74247
74LS247
9317B
9317C
8T04

20 (~ink)

W/O
W/O
W
W
W/O
W/O
W/O

FAIR, ITT, MOT, NS,

SIG, TI
TI
NS, TI
TI
TI
FAIR
FAIR
SIG

7448
74248
74LS48
8T05

BCD to 7-Segment Decoder/Driver

OUTPUTS: Active High, Current
Source
BCD to 7-Segment Decoder/Driver
OUTPUTS: Active High, Open
Collector

BCD to 7-Segment Decoder/Driver

OUTPUTS: Active Low, Open
Collector

MANUFACTURER

12 bink)
40 (link)
24 (!ink)
40 !link)
20 (sink)
40 Silk

BCD to 7-Segment Decoder, CMOS;

OUTPUTS: N-channel sink and
NPN Bipolar Sources

74C48

3.6 hink)
50 (s>urce)

W/O

HAR, NS

BCD to 7-Segment Latch/Deco~er/

Driver; CMOS with Bipolar Outputs

4511
(14511)

25 (st)urce)

W/O

FAIR, HAR, MOT, RCA,

SSS

BCD to 7-Segment Decoder/Driver,

CMOS

14558

0.28 :sink)
0.35 source)

W/O

MOT

BCD to 7-Segment Latch/Decoder/

Driver

9368

22 (source)

6(W)

FAIR

OUTPUTS: Constant Current

9374

18 (silk)

8673/
8674

18 (si ,k)

9370

40 (sink)

BCD to 7-Segment Latch/Decoder/

Driver

9 (W/O)
6(W)
9 (W/O)
6(W)
9 (W/O)
6(W)

FAIR
NS

FAIR

9 (W/O)

OUTPUTS: Active Low, Open

Collector
LIST OF MANUFACTURERS
FAIR

ITT = ITT SEMICONDUCTOR

F IRCHILD SEMICONDUCTOR

HAR = HARRIS SEMICONDUCTOR

TABLE 5.5.9

MOT

MOTOROLA

NS = NATIONAL

SEMICONDUCTOR

SEMICONDUCTOR

List of LED Display to Logic I terface Devices

5.58

RCA = RCA SOLID STATE DIVISION

SIG = SIGNETICS
SSS = SOLID STATE SCIENTIFIC
TI = TEXAS

INSTRUMENTS

ao

CONTRAST
DISPLAYS

ENHANCEMENT

FOR

Contrast ratio ranges in value from one, where L = L '

vS
vB
to a value tending towards infinity, when LvS ~ LvB'
Contrast and contrast ratio are related by the following two
expressions:

LED

The most important attribute of any equipment utilizing

digital readout is the ability to clearly display informatio
to an observer. An observer must be able to quickly an
accurately recognize the information being displayed by th
instrument. The display, usually front panel mounted, mus1
be visible without difficulty in the ambient light condition'
where the instrument will be used.

C = 1 __ 1_
CR

(6.1-3)

CR = _1_

(6.1-4)

1-C
As shown in the curve of Figure 6.1-1, the range of
attainable contrast for LED displays is between a useable
minimum of 0.90 (CR=lO) to 0.98 (CR=50) with a typical
fIlter providing a contrast of 0.95 (CR=20).

Since most ambient light levels are sufficiently bright t

impair the visibility of an LED display, it is necessary t
employ certain techniques to develop a high viewin!
contrast between the display and its background. Since th
quality of visibility is primarily subjective, it is not easil)
measured or treated by analytical means. Thus, humar
enginee ring plays a very important
role in displa)
applications. The best judge of the viewing esthetics of ,
display is the human eye, and the final display design mus
be pleasing to the eye when viewed in the end user ambient

A spot photometer is used to measure luminous sterance.

The display should be located behind the contrast filter,
installed in its mounting assembly and packaged as the final
product in order to obtain realistic values. Ambient
lighting, which represents the end use envirorunent, should
be incident to the fIlter during the measurement. Care
should be taken to prevent shading of the spot being
measured by the photometer.
With stretched segment displays, it is difficult to achieve a
high value of segment on/off contrast while effectively
concealing the display package from view. For example, a
display with a black package is easily concealed from view,
however, the "OFF" segments will be visible. This is due to
the difference in reflectivity between the "OFF" segments
and the black package.

The objective of contrast enhancement is to maximize th

contrast
between display "ON" and display "OFF'
conditions. This is accomplished by (1) reducing to
minimum the reflected ambient light from the face of th
display and (2) allowing a maximum of the display'
emitted light to reach the eye of an observer. The goal is t<
achieve a maximum contrast between "ON" segments anI
"OFF" segments as well as a minimum contrast betweel
"OFF" segments and display package and background.
Both contrast and contrast ratio are used as the measure 0
the difference in the luminous sterance of a source wit!
respect to the surrounding background. Contrast, C, for aJ
LED display is defined as:
C=

-L

vS
vB ; 0
LvS

where:

<; C <;

-LB

vS
v
LvS

(6.1-1

LvS = Source luminous sterance (cd/m2)

LVB = Background luminous sterance (cd/m2)

I-

.6

lz

.5

en

a:

u
U

.4

Contrast varies from zero, where LvS = LvB' to a value 0

one, where the source intensity is at a level such that LvS ~

11.illl:ilii.I:ilii~IJII~

~B'

CR

lllllllll:!iliili.iili:llll]'

L
= ~

LVB

1<; CR <;

00

"lililiililili::I,ii:.lIIIIJII
5

Contrast ratio, CR, is the ratio of the source luminou

sterance to the background luminous sterance.

Figure

6.1-1

10
20
50
CR" CONTRAST RATIO

The Relationship
Ratio.

Between

Contrast

and Contrast

A reduction in the reflectivity difference between the

"OFF" segments and the package of a stretched segment
display IT ay be obtained by adding a small amount of dye
to color 'int the segments, and the display package may be
colored 1.0 match the "OFF" segment color. With the
addition of an appropriate optical fIlter placed in front of
the dis ;Jlay, the "0 FF"
se gmen ts tend
to be
indistingl ishable from the background. The trade-off is that
a colorec package is more visible than a black package.
Because of this trade-off, a designer has to decide which is
more imrortant, concealing "OFF" segments or concealing
the display package. Since the usual choice is to conceal
"OFF", segments, Hewlett-Packard is using this colored
packageechnique
on its high-efficiency red, yellow and
green stretched segment displays.
Contrast enhancement under artificial lighting conditions
may be accomplished by use of selected wavelength optical
filters.
Under
bright
sunlight
conditions contrast
enhancement
becomes
more difficult and requires
additiona: techniques such as the use of louvered fIlters
combined with shading of the display. The effect of a
wavelength optical ftlter is illustrated in Figure 6.1-2. The
ftltered portion of the display can be easily read while the
"OFF" segments are not apparent. By comparison, reading
the unfIltered portion of the display is difficult.
6.2

Eye Response.
W velength

Peak Wavelength

and Dominant

The 1931 CIE (Commission Intemationale De L'Eclairage)

standard observer curve, also known as the photopic curve,
is shown in Figure 6.2-1. This curve represents the eye
response of a standard observer to various wavelengths of
light. The vivid color ranges are also identified in Figure
6.2-1 to illustrate the sensitivity of the eye to the various
colors. The photopic curve peaks at 555 nanometers (nm)
in the yellowish-green region. This peak corresponds to 680
lumens of luminous flux (1m) per watt of radiated power
(w).
Two wavelengths of the LED emission are important to a
user of LED displays; Peak Wavelength and Dominant
Wavelength. Peak wavelength (A ) is the wavelength at the
peak of tfJ.e radiated spectrum. PThe peak wavelength may
be used to estimate the approximate amount of display
emitted lIght that passes through an optical ftlter. For
example, if an optical fIlter has a relative transmission of
40% at a given \~, then approximately 40% of the display
emitted light at fhe peak wavelength will pass through the
fIlter to the viewer while 60% will be absorbed. This gives a
designer an initial estimate of the amount of loss of display
emitted light he should expect.

,
ii

'.

8"." 0
0 B, ,"
.

".

B"

Figure 6.1-2 Effect of a Wavelength

Display.

,f

-,

Optical

=' ,;,
,=,_,
Filter on an LED

LED display. Specifically, the dominant wavelength is that

wavelength of the color spectrum, which, when additively
mixed with the light from the source CIE illuminant C, will
be perceived by the eye as the same color as is produced by
o
the radiated spectrum. CIE illuminant C is a 6500 Kcolor
temperature source that produces light which simulates the
daylight
produced by an overcast sky. A graphical
definition of Ad and color purity is given on the CIE
chromaticity
diagram in Figure 6.2-2. The dominant
wavelength is derived by first obtaining the x,y color
coordinates from the radiated spectrum. These color
coordinates are then plotted on the CIE chromaticity
diagram. A line is drawn from the illuminant C point
through the x,y color point intersecting the perimeter of
the diagram. The point where the line intersects the
perimeter is the dominant wavelength.
The color purity, or saturation, is defined as the ratio of the
distance from the x,y color point to the illuminant C point,
divided by the sum of this distance and the distance from
the x,y point to the perimeter. The x,y color coordinates
for LEDs plot very close to the perimeter of the
chromaticity
diagram.
Therefore, the color purity
approaches a value of 1, typical of the color saturation
obtained from a monochromatic light source.
The dominant wavelengths and corresponding colors for
LEDs are shown on the CIE chromaticity diagram in Figure
6.2-3. As defined by Ad' the color of a standard red LED is
red, a high-efficiency red LED is reddish-orange, a yellow
LED is yellowish-orange and a green LED is actually
greenish-yellow.
It is of value to know the actual colors of each LED when
selecting a contrast fIlter, as the optimum fIlter will have
the same color as the device. Both Ap and Ad are listed on
LED data sheets.

The contrast filter must meet certain basic requirements. It

should be mechanically stable with temperature,
have

reasonable chemical resistance, be free from visual defects,

have a homogeneous index of refraction and have sufficient
transmission of the radiated spectrum of the LED. It is the
last two requirements
that are of primary optical
importance.
The relative transmittance of an optical fJ.1terwith respect
to wavelength is defined to as:
(6.3-1)
UJ

0.1

en
a:

UJ
UJ

UJ

a:

"~...J

UJ

>
2

I<l:
UJ

a:

"

a:

0
...J

...J
UJ
>

:i:
en

0 2

UJ
UJ

...J

2>

T(A)

UJ

en

2
0
0..

...J
UJ
>

UJ
UJ

a:

"

"a:
<l:

2
UJ
UJ

UJ

a:

0)

:i:

UJ

:)
...J

!!?
...J

aJ

0..

a:

:)

0..

.001
370 400

UJ
UJ

a:

"
500

450

Ad FOR HP DISPLAYS

Figure

6.2-1

"en

:i:

0
...J

...J
UJ
>

en

~
0
...J

...J
>
UJ

UJ

The index of refraction determines the amount of incident

light that is reflected at the filter/air interface. The amount
of reflected light is given by the following ratio:

"a:
2

<l:

:i:
!!?

0
UJ

a:

0
0
UJ

a:

550 1\600 [ 650

572
626
585
639

700

750

where:

UJ

I<l:

0.6

Ad,

DOMINANT
WAVELENGlH
= 585 nm

a:

0
0
(.)

a:
0
...J

n1

Index of refraction of the filter material.

n2

Index of refraction for air

1.0.

A plastic fJ.1terwith an average index of refraction equal to

1.5, for the range of wavelengths encompassing the LEDs
radiated spectrum, will reflect 4% of the normal incident
light at each fJ.1ter/air interface. Thus, 8% of the LED
emitted light, passing through the fJ.1ter to an observer, is
lost due to reflection.

CIE Standard
Observer Eye Response Curve
(Photopic
Curve), Including CI E Vivid Color
Ranges.

0.8

2
0

Luminous Flux with Filter at Wavelength A

Luminous Flux without Filter at Wavelength A

The relative transmittance is a function of the index of

refraction of the fJ.1ter material and the transmission
through the material as determined by the coloring. Both
are a function of wavelength. Specifically, the transmission
through the fJ.1ter is dependent upon the amount of
incident light reflected at the fJ.1ter/air interface and the
amount of wavelength absorbtion within the fJ.1termaterial.

:i:

.01

:)
...J

0.4

Inside the filter, light is lost due to absorbtion by the tinted

material. The amount of absorbtion is a direct function of
wavelength and is determined by the dye coloring and dye
concentration. If the dye coloring is held at a constant
density, the transmission through the fJ.1termaterial at any
given wavelength, TA' is an exponetial function of the
thickness of the material:

0
(.)

>

where:

0.2

0.4

0.6

x - COLOR COORINATE
Figure

6.2-2

Definition
of Dominant
Wavelength and Color
Shown on the CI E Chromaticity
Diagram.

Purity.

The quantity
material.
2.71828

of unit thicknesses of filter

Absorbtion coefficient and is equal to -Int,

where t is the transmission for a unit
thickness.

As an ~xample, the transmission through 1.0mm thickn,:ss

of fIlter material is 0.875 at a wavelength of 655 nm. At} =
2.5, th~ thickness is 2.5mm and the transmission is 0.716:
-In

Dim ambients are in the range of 32 to 215 Ix (3 to 20

footcandles), moderate ambients are in the range of 215 to
1076 lux (20 to 100 footcandles), and bright ambients are
in the range of 1076 to 5382 lux (100 to 500 footcandles).
Lux = (lm/m2) and footcandle = (lm/ft2).

(J:75) = 0.1335

T A = e-(.1335)(2.5)

= .716

The relative transmission of a contrast fIlter at a particular

wavelength, T(A), may be calculated with reasonatle
accuracy by using the following relationship:

where:

n = The index of refraction of the filter material.

TA = Transmission through the filter material
at wavelength A.

For a 1.0mm thick plastic fIlter with n = 1.5 and TA = .8i 5

at a wavelength of 655 nm, the relative transmission is
0.808:
T (A = 655 nm) =

f?

(1.5)

levels much less than display emitted light. The display can
now be driven at a low average current.

[.875]

= .808

Listed on each fIlter transmittance curve are empirically

selected ranges of relative transmittance values at the peak
wavelength which may give satisfactory fIltering. For
example, a fIlter to be used with a yellow display in
moderate ambient lighting could have a transmittance value
at the peak wavelength [T(Ap)] between 0.27 and 0.35.
The fIlter wavelength cut-off sfiould occur between 530 and
550 nm for best results.
When selecting a fIlter, the transmittance curve shape,
attenuation at the peak wavelength and wavelength cut-off
should be carefully considered in relationship to the LED
radiated spectrum and ambient light level so as to obtain
optimum contrast enhancement.

~1.5)2 +1J

The sane fIlter material at a thickness of 2.5 mm has a

relative transmittance of 0.661, as shown in Figure 6.3-1.

Due to their low cost, ease in machining to size an i

resistance to breakage, plastic contrast fIlters are being use i
in a majority of display applications. The filter requiremen;s
for dim, moderate and bright ambients are different for each
lighting ~ondition. Therefore, it is advantageous to becom:
familiar
with
the various
relative
transmittanc,:
characte -istics that are available in plastic fIlters.
Most manufacturers of wavelength fIlters for use with LED
displays provide relative transmittance curves for thei'
products. Sample transmittance curves are presented ir
Figures 6.3.1-1, -2, -3 and -4. These curves represem
approximate fIlter characteristics which may be used ir
various ambient light levels. The total transmittance curve
shape and wavelength cut-off points have been chosen in
direct relationship to the LED radiated spectrum. Each
fIlter curve has been empirically determined and is similar
to commercially available products. The higher the ambient
light,
the more optically dense the fIlter must be to
absorb refected light from the face of the display. Because
the display emitted light is also strongly absorbed, the
display r:1Ust be driven at a high average current to be
readily vi;ible. For dim ambient light, the fIlter may have a
high valu,~ of transmittance as the ambient light will be at

Three manufacturers
of plastic wavelength fIlters are
Panelgraphic Corporation (ChromafIlter@), SGL Homalite
and Rohm & Haas Company (plexigla~). The LED fIlters
produced by these manufacturers are use able with all LED
display and lamp devices. Table 6.9-2 lists some of the filter
manufacturers and where to go for further information.
Table 6.9-1 lists some specific wavelength fIlter products
with recommended applications.

Optical glass fIlters are typically designed with constant

density so it is the thickness of the glass that determines the
optical density. This is just the opposite of plastic filters
which are usually designed such that all material thicknesses
have the same optical density.
The primary advantage of an optical glass contrast fIlter
over a plastic fIlter is its superior performance. This is
especially true for red LED fIlters. Figure 6.3.2-1 illustrates
two thicknesses of a red optical glass fIlter for use with
standard red LED displays. The relative transmittance is
generally higher than that of a comparable plastic filter, and
the slope of the relative transmittance curve is usually much
steeper and follows more closely the shape of the radiated
spectrum of the LED. This particular filter provides
excellent contrast in a bright ambient. A reddish-orange
optical glass fIlter which is suitable for use with a
high-efficiency red display in a moderate ambient is shown
in Figure 6.3.2-2. The relative transmittance of this fIlter

follows almost exactly the leading edge of the shape of the

LED's radiated spectrum.
A leading manufacturer of optical glass filters is the Schott
Optical Glass, Inc. of Duryea, Pennsylvania and Munich,
Germany.

SYMBOL
R
rO
0
yO
Y
gY
YG
yG
G
bG
BG
gB
B
pB
bP
P
rP
RP
pR
pPK
PK
OPK
C

The application of wavelength fIlters as described in the

previous section is the most widely used method of contrast
enhancement under artificial lighting conditions. However,
they are not very effective in daylight due to the high level

0.8

COLOR NAME
RED
REDDISH-ORANGE
ORANGE
YELLOWISH-ORANGE
YELLOW
GREENISH-YELLOW
YELLOWGREEN
YELLOWISHGREEN
GREEN
BLUISHGREEN
BLUEGREEN
GREENISHBLUE
BLUE
PURPLISH-BLUE
BLUISH-PURPLE
PURPLE
REDDISHPURPLE
RED-PURPLE
PURPLISH-RED
PURPLlSHPINK
PINK
ORANGEPINK
CIE ILLUMINATED C

0.6

0.4
~ = 626; HI-EFF. RED LED

Ad = 640;

STD. RED LED

0.2

Figure 6.2-3 Dominant Wave ngths and Corresponding Colors

for LEOs. Sho","n on the CI E Chromaticity
Diagram.

1.0

ambient light. Filtering in daylight conditions is best

achieved by using louvered filters (discussed in a later
section) .

.9
.8
w

.7

lI-

.6
~
en
2

a:

.5

.4

I-

>

-I
W

.3

a:

.1

0
500

550

600
A - WAVELENGTH

650

The figures in sections 6.4.1 through 6.4.4 show the

relationship between artificial lighting and the spectra of
LED displays, both unfiltered and fIltered. Figures 6.4.1-1
through 6.4.4-1 show the relationship between the various
LED spectra and the spectra of daylight fluorescent and
incandescent light. The photometric spectrum (shaded
curve) is obtained by multiplying the LED radiated
spectrum [f(A)] by the photopic curve [y(A)]. Thus,
photometric spectrum = f(A)-y(A). Figures 6.4.1-2 through
6.4.4-2 demonstrate the effect of a wavelength filter. The
fIltered photometric spectrum is what the eye perceives
when viewing a display through a filter (shaded curve).
Thus, filtered photometric spectrum = f(A)-y(A)-T(A). The
ratio of the area under the fIltered photometric spectrum to
the area under the unfiltered photometric spectrum is the
fraction of the visible light emitted by the display which is
transmitted by the filter:

- nm

Figure 6.3-1 Variation in Relative Transmittance

vs. Thickness
for a Constant Density Filter Material.

Fraction of Available
Light from Filtered Display

= [

f(A)- (A)-T(A) -d

f f(A)-Y(A)-dA

HIGH TRANSMITTANCE
FOR MODERATE AMBIENTS
.60'; T (Ap) .; .75
w

~ .70

:!E .60
CI)

LOW TRANSMITTANCE

LOW TRANSMITTANCE
FOR
MODERATE-TO-BR IGHT
AMBIENTS .35'; T (Ap) .; .50

>
.40

~ .40
..J
W

a:

a: .30

I .30

j::

.60

I- .50
w

..J

:<

:!E
CI)

a:

FOR BRIGHT AMBIENTS

.35 .; T (Ap) .; .50

:: .50

.70

t=

lI-

lI-

>

HIGH TRANSMITTANCE
FOR
DIM-TO-MODERATE
AMBIENTS
.35 .; T (Ap) .; .70

:<
j::

.20

Figure 6 3.1-1

Typical Transmittance
Curves for Filters to be
Used with Standard Red Displays.

.20

Figure 6.3.1-2

1.00

1.0
LOW TRANSMITTANCE

w .90
u

FOR MODERATETO-BRIGHT AMBIENTS

.27";;; T 1\1 ";;;.35

.80

:ECIl

.70

lI-

a:

.60

I-

...I

.40

a: .30
I-

HIGH TRANSMITTANCE

CIl

.70

FOR DIM AMBIENTS

.40" T (Ap) " .50

.60

a:
I--

>
I-
..J

._---

.50
.40
LOW TRANSMITTANCE

a:

-----

.30

::<

-3

.80

w .50

>
~

I-I--

'HIGH TRANSMITTANCE
FOR DIM AMBIENTS
.35";;; T 1\1 ";;;.45

.90

Typical Transmittance
Curves for Filters to be
Used with High-Efficiency Red Displays.

MODERATE

FOR

AMBIENTS

.15" T (Api" .30

I--

.20

.20

.10

.10

550

600

A-

Figure 6.3.1-3

WAVELENGTH

650
Inml

Typical Transmittance
Curves for Filters to be
Used with Yellow Displays.

Figure 6.3.1-4

Typical Transmittance
Curves for Filters to be
Used with Green Displays.

In addition to attenuating a portion of the light emitted by

the display, a filter also shifts the dominant wavelength,
thus causing a shift in the perceived color. For a given
display spectrum, the color shift depends on the cut-off
wavelength
and shape
of the filter transmittance
characteristic. A choice among available filters must be
made on the basis of which filter and LED combination is
most pleasing to the eye. A designer must experiment with Ieach filter as he cannot tell by transmittance curves alone. ~
The filter spectra presented in Figures 6.3.1-1 through ::>
o
6.3.1-4 are suggested starting points. Filters with similar w
>
f=
characteristics are commercially available.
<{

u::

.9
.8

.7

RELATIVE TRANSMISSION
REDDISH-ORANGE GLASS
FILTER
THICKNESS = 2.5 mm

RADIATED SPECTRUM
HIGH-EFFICIENCY
RED
LED

.6
.5

.4

a:

I-

ir

I-

.6

::>

o
~

.5

f=
<{

a:

.4

600
RELATIVE
TRANMISSION
RED OPTICAL GLASS
FIL TER
THICKNESS = 1 mm

A - WAVELENGTH

A Reddish-Orange Optical Glass Filter for Use

with High-Efficiency
Red Displays.

6.3.1-2
and 6.4.2-2).
reddish-orange.
6.4.3

o
500

600

650

A - WAVELENGTH

Figure 6.3.2-1

A Red Optical
Red Displays.

- nm

Glass Filter for use with Standard

Filtering out reflected ambient light from red displays i

easily accomplished with a long wavelength pass filter
having a sharp cut-off in the 600 nm to 620 nm range (se
Figures 6.3.1-1 and 6.4.1-2). Under bright fluorescent light
a red filter is very effective due to the low concentration 0
red in the fluorescent spectrum. The spectrum of
incandescent light contains a large amount of red, and
therefore, it is difficult to filter red displays effectively in
bright incandescent light.
6.4.2

- nm

Filtering High-Efficiency

Red Displays lAp

635

nm)

The use of a long wavelength pass filter with a cut-off in th

570 nm to 590 nm range gives essentially the same result
as is obtained when fIltering red displays (see Figure

The

resulting

color

is a rich

Filtering Yellow Displays lAp = 583 nm)

The peak wavelength of a yellow LED display is in the

region of the photopic curve where the eye is most sensitive
(see Figure 6.4.3-1). Also, there is a high concentration of
yellow in the spectrum of fluorescent light and a lesser
amount of yellow in incandescent light. Therefore, fIlters
that are more optically dense than red fIlters at the peak
wavelength are required to filter yellow displays_ The most
effective filters are the dark yellowish-orange (or dark
amber) filters as shown in Figure 6.3.1-3. The use of a low
transmittance yellowish-orange filter, as shown in Figure
6.4.3-2 results in a similar color to that of a gas discharge
display. Pure yellow filters provide very little contrast
enhancement.
6.4.4

Filtering Green Displays lAp

565 nm)

The peak wavelength of a green LED display is only 10 nm

from the peak of the eye response curve (see Figure
6.4.4-1). Therefore, it is very difficult to effectively filter
green displays. A long wavelength pass fIlter, such as is used
for red and yellow displays, is no longer effective. An
effective fIlter is obtained by combining the dye of a short
wavelength pass filter with the dye of a long wavelength
pass filter, thus forming a bandpass yellow-green filter
which peaks at 565 nm as shown in Figure 6.3.1-4. Pure

1.00

.80

----

.90
l-

.80

...::>

:;)

.60

::>

CL
l-

.60

a
w
>

.50

...J

.40

:;)

.70

..."-

.70

i=

0
w

.50

...~

er;

.40

...J

.30

w
II:

.30

.20
.20

.10

.10

0
500

I
I
550

600

6501
AP: 655

A - WAVELENGTH

Figure 6. t 1-'

(nm)

Relative Relationship Between the Standard Red

lED Spectrum, Photopic Curve and Artificial
lighting.

1.00
.90
.80

...

.70

...::>"-

.60

::>

0
w

.50

.40

>
...J
W

II:

.30
.20
.10

I
I

550

650

Ap : 635
A - WAVELENGTH
Figure 6.4.2-'

Relative Relationship Between the HighEfficiency Red lED Spectrum, Photopic Curve
and Artificial lighting.

green fIlt\lrS peak at 520 run and drop off rapidly in the 550
nm to 570 run range and are not recommended. The best
possible tilters for green LED displays are those which are
yellow-grllen bandpass, peaking at 565 nm and dropping off
rapidly b ~tween 575 nm and 590 nm. As shown in Figure
6.4.4-2, this fIlter passes wavelengths 550 to 570 while
sharply reducing the longer wavelengths in the yellow

(nm)

Effect of a long Pass Wavelength Filter on a

High-Efficiency Red lED Display.

region.
To effectively fIlter green LED displays in
fluorescent light would require the use of a filter with a low
transmittance value at the peak wavelength. This is due to
the high concentration
of green in the fluorescent
spectrum. It is easier to filter green displays in bright
incandescent light due to the low concentration of green in
the incandescent spectrum, see Figure 6.4.4-1.

1.0
YELLOW
.90

SPECTRUM

1.00

.80

.90

.80
.70

::J
::J

0
w

>

.60

.60

0-

::l

FILTERED
PHOTOMETRIC
SPECTRUM
TlAl-

l-

::l
Q.

.70

l-

LED

PHOTOMETRIC

>
i=

.50

AMOUNT OF
DISPLAY EMITTED
LIGHT PASSING
THROUGH FILTER
=27%

<C .40
...I

.50

a:

.40

.30

\
\

.30
.20

,,

FILTER

TRANSMITTANCE

.20
.10

.10

1
500

0
400

550

600

650

700

\=583

ARelative Relationship
Spectrum,
Photopic

Between the Yellow LED

Curve and Artificial Lighting.

Figure

6.4.3-2

WAVELENGTH

Effect of a Wavelength
Display.

Filter

lnml
on a Yellow

LED

1.00
.90
1.00

.80
.90
I-

.80

l>.
.70
.60

0
w

.50

::J

>

~
<0:
..J

I-

.60

0
w

.50

::l

I-

::J
0-

I-

.70

::l

>
I
...J

.40

a:

.40

.30

II:

.30

.20

.20

.10

.10

0
500

700

550
AP = 565
A - WAVELENGTH

Figure

6.5

6.4.4-1

Relative Relationship
Spectrum,
Photopic

Between the Green LED

Curve and Artificial Lighting.

Reduction of Reflected Ambient Light as Provided

by a Contrast Filter

Incident ambient light that is reflected back to an observer

from the face of a ftltered display travels twice through the
contrast ftlter. The amount of light actually reflected from
the face of the display is dependent upon the diffused
reflectance, off the face of the display, as a function of
wavelength, R(X).

Figure

6.4.4-2

Effect
Green

of a Bandpass
LED Display.

(nm)

Wavelength

Filter

on a

Diffu sed reflectance is absorbtion dependent and is

measured with the incident light beam normal to a surface.
Those wavelengths which are not absorbed are reflected
back as color. This is not the same as specular reflection. In
specular reflection, the light is incident to a surface at some
angle less than 90 where all wavelengths are essentially
reflected
equally.
Figure 6.5-1 shows the diffused
reflectance as a function of wavelength for stretched
segment displays that have colored packages.

The amount of incident light reflected back to an observer

from a filtered display as compared to an unfiltered display
may be calculated from the ratio of the filtered reflected
light to the unfiltered reflected light.

reflected fluourescent light from the face of a yellow or

green LED display without significantly decreasing the
emitted light from the LED.

Amount of Filtered = [X(A)R(A)T2p\)V(AJdA

Reflected Light
f X(A)R(A)V(A)dA
=
=

T(A)
VIA)
6.5.1

Spectral distribution of the incident

ambient light.
Relative diffused reflectance off the
face of the display.
Filter relative transmittance.
Photopic curve.

Effectiveness of a Wavelength Filter in an Ambient

of Artificial Lighting

The contrast is very dependent upon the ambient lightinf

If most of the spectral distribution of the ambient light is
outside the radiated spectrum of the LED, it is very easy t,)
reduce the reflected ambient light to a very low level while
imposing only minimal attenuation on the emitted light
from the display. Such is the case when using a red LED
display in a fluorescent ambient or a green LED in a 1
incande3cent ambient. The opposite is also true. It is very
difficult to effectively reduce the reflected incandescen t
light from the face of a red LED display, or to reduce

Figure

6.5.1-1

S
.8

Spectral Distribution
and an Incandescent
Lighting.

for a Daylight Fluorescent

Lamp used for Artificial

XI~)
SPECTRAL DISTRIBUTION
INCANDESCENT LAMP

OF AN

(MEASUREMENTS OBTAINED WITH A DETECTOR

THAT HAS A FLAT SPECTRAL RESPONSE)
~

.8

~
Iu .7

w
..J
U.
W

a: .6

HIGHEFFICIENCY
RED DISPLAY

(I)

::>

u.
u.

.2

.5

Ci
w

>
~
..J

.4

.3

REFLECTED LIGHT
OFF FACE OF AN
UNFIL TERED
DISPLAY

Figl:re

600
A WAVELENGTH

Figure

6.5-1

Diffused
Segment

nm

Reflectance
for the Faces of Stretched
Displays that have Colored Packages.

6.5.1-2

REFLECTED LIGHT
OFF FACE OF A
FILTERED DISPLAY

Reflected
Incandescent
Light Off an Unfiltered
and Filtered High-Efficiency
Red Display with a
Colored Package, as Seen by an Observer.
The
Filtered Reflected
Light is 17% of the
Unfiltered
Reflected
Light.

Figure 6.5.1-1 reproduces the spectral distributions for

fluorescent and incandescent lighting. Fluorescent lighting
contains almost no red, yet contains a high level of yellow
and long wavelength green. Incandescent, is just the

.9

X1(:>..)
SPECTRALDISTRIBUTION OF
AN INCANDESCENTLAMP

I- .6

::>

Reflected Incandescent Light Off an Unfiltered

and Filtered Yellow Display with a Colored
Package, as Seen by an Observer. The Filtered
Reflected Light is 4% of the Unfiltered
Reflected Light.

.5

>
~
<l:
..J

.4

w
II:

REFLECTED LIGHT
-OFF FACE OF AN
UNFILTERED
DISPLAY

550

:>"p = 583
:>..

600

= WAVELENGTH = nm

X (:>..)
F
SPECTRAL DISTRIBUTION
OF A DAYLIGHT
FLUORESCENT LAMP

I-

ir

.6

I-

Reflected Fluorescent Light Off an Unfiltered

and Filtered Yellow Display with a Colored
Package, as Seen by an Observer. The Filtered
Reflected Light is 6% of the Unfiltered
Reflected Light.

::>

o
w
>

.5

.4

~
<l:
II:

opposite, being composed primarily of orange and red light

As can be seen from Figure 6.5.1-1 and the Figures'
Section 6.4, the contrast filter may have to reduce the sam
wavelengths that are predominant in both the LED an
ambient light in order to achieve adequate contrast. This i
especially true for green LEDs under fluorescent lighting.

Figures 6.5.1-2 through 6.5.1-5 illustrate a comparative

difference between the filtered and unfiltered diffused
reflected light off the face of a stretched segment LED
display as seen by the eye of an observer. The curve of the
unfiltered diffused reflected light is obtained by the
product of the lamp spectral distribution [X(A)], diffused

reflectance off the face of the display [R(;\)] and the

photopic curve [yQ,.)]. Since the incident ambient pass{s
througl( the filter twice, the curve of the filtered reflected
light is the unfiltered curve multiplied by the square of the
filter transmittance [T2(A)]. The filters used are the sam~
as those used in Figures 6.4.2-2, 6.4.3-2 and 6.4.4-2. Thes~
curves illustrate that a wavelength filter substantiall}
reduces the incident ambient light that is reflected off th~
face of It display. Even so, the peak of the reduced reflected
ambient light is at or near the peak wavelength of the LEI>
radiated spectrum.

XF(X)
SPECTRAL DISTRIBUTION
OF A DAYLIGHT
FLUORESCENT LAMP

I-

I-

.6

:>

6.5.2

ffectiveness
mbients

of a Wavelength

Filter in Dayligh

The purpose of the wavelength filter is to reduce th<:

luminous sterance of the background to a level that is very
much less than the luminous sterance of the display':.
illuminated segments in order to achieve a high value 0:'
contrast, Referring to Equation 6.1-1 for contrast, th{
contrast goes to zero if the background luminance steranc{
equals the luminance sterance of the illuminated segments
This situation occurs when the ambient light level i~
sufficient to effectively mask the display's emitted light.
The spectral distribution of bright sunlight is nearly a flat
curve across the complete color spectrum, as shown in
Figure 6.5.2-1. Bright sunlight that is directly incident upon
a display with a wavelength filter raises the background
luminou' sterance to a level that may actually exceed the
luminous sterance produced by the emitted light from the
illuminated segments. Therefore, the luminous sterance of
the displ<ty segments, "ON" or "OFF", equals the luminous
sterance of the background and the contrast goes to zero.
A similar situation occurs with overcast sky daylight that is
directly incident onto the face of a wavelength filtered
display. The situation is not as pronounced for standard red
LED displays as for green LED displays, since the overcast
sky filters out a considerable amount of the ambient red
light, see Figure 6.5.2-1. It is obvious, then, that a
waveleng th filter by itself is not sufficient to achieve the
necessary contrast in daylight ambient. Some shading or
blocking out of the incident daylight is also required to
achieve sufficient contrast between the display background
and the illuminated segments.
It is almost impossible to achieve a minimum acceptable
contrast if the incident daylight is parallel to the viewing
axis (the incident daylight is perpendicular to the face of
the disphy). If the incident daylight is not parallel to the
viewing I xis, other filter techniques may be employed to
obtain a .ceptable contrast. These techniques include the
use of louvered or cross hatch filters, incorporation of an
eyebrow shade and recessing the display. Also, a circular
polarizin filter may be employed with those displays that

o
w
>

.5

f=
<l:

uj
a:

.4

550 Ap = 565

600

X - WAVELENGTH

650
- nm

Figure 6.5.1-5

Reflected Fluorescent Light Off an Unfiltered

and Filtered Green Display with a Colored
Package, as Seen by an Observer.
The Filtered
Reflected Light is 10% of the Unfiltered
Reflected Light.

Figure 6.5.2-1

Spectral Distributions for Bright Sunlight

Overcast Sky Daylight.

and

have a face which is a specular reflecting surface. These

techniques are discussed in more detail in later sections.
Indirect daylight that is incident to the face of a display is
already reduced in level from direct daylight. Simple
shading of a display, which is filtered by a louvered filter,
will produce adequate contrast. It is the shading that
reduces the indirect daylight to a low enough level to allow

the fIlter to reduce the background luminous sterance to a

level sufficient for adequate contrast.
6.6

Special
Wavelength
Combination

Filters

and

Filters

in

A designer is not limited to a single color wavelength fIlter

to achieve the desired contrast and front panel appearance.
Some unique wavelength fIlters and fIlter combinations
have been successfully developed. One is the purple color
fIlter for use with red LED displays, and another is the use
of a neutral density filter in combination with a light amber
filter t achieve a dark front panel for yellow LED displays.

A most effective contrast filter is shown in Figure 6.6.1-1.

The color, purple, is not defined by a dominant wavelength,
see Figure 6.2-3. Purple is a mixture of red and blue light
which is perceived by the eye as a distinct color from red .
Psychologically, a purple contrast filter is more pleasing to
many people than a red filter. The reason for this may be
that when illuminated, the standard red display stands out
so vividly against the purple background. Therefore, it is
the color difference that enhances the contrast. This makes
the purple contrast mter extremely effective in brightly
lighted ambients .

1.0

.9
.8
.7
RELATIVE
TRANSMISSION
PURPLE Fit TEA

..

PERCEIVED
RED!PURPlE
COLOR CONTRAST PROVIDES
EXTREMELY HIGH lED
ON/OFF CONT
AST

.5

The wavelength fIlters that have been previously discussed

have a distinct color that may be identified by a dominant
wavelength. The contrast that they provide is essentially
due to the high level luminous sterance of the display
segments contrasted against the low level luminous sterance
of a background of the same color. Another approach
maintains the contrast ratio, but has as the background a
different color than that of the LEDs in the display. This is
easily accomplished by using a dark purple fJ.Iter with
standard red LED displays.

.4

.3

.2

.1

0
400

450

Figure 6.6.1-1

500

A Purple Color Wavelength

Red LED Displays.

Filter for Standard

The amount of light reaching the eye of a viewer is 24% of

the unmtered LED spectrum:

I-

::> .6
I::>
D.

.5

f=

...J

A neutral density gray mter is often used in combination

with other mters to provide a dark gray filter window as
well as increased contrast in bright ambients. A typical
example is given in Figure 6.6.2-1. The resulting fJ.Iter is the
product of the relative transmittance of the light amber,
T LA(X), and the relative transmittance of the neutral
density gray, T NcCX):

w .4

TLA(A) .TNG(A)
PRODUCT
FIL TER
CURVE

cc

t
Ap = 583

600

A . WAVELENGTH

nm

A Neutral Density Gray Filter in Combination

with a Light Amber Filter for Use with Yellow
Displays.

Fraction of
Available Light
Through a
Combination
Filter

If(X)-Y(X)-

[T LA(X)-T NG(X)]-dX

If(X) - Y(X) - dX

The advantage is a dark gray front panel window with very

low luminous sterance (zero transmission below 525 nm)
that retains its appearance in bright ambients. The trade-off
is a considerable reduction in the luminous sterance of the
display which reduces the contrast ratio. This is somewhat
offset by the distinct color difference between the
illuminated yellow segments of the display and the dark
gray background.

o
~

~
en
2

a::

60

40

I-

.035 IN.

(O,89mml

Figure 6.71 Construction Chal acteristics

Louvered Filter.

Louvered fllters are very effective in reducing the amount

of bri~t artificial light or dayligh~ reflected. fro~ th~ fac,~
of a <Usplay, without a substantlal reductIOn III dlSp1a.r
emitted light. The construction of a louvered fllter i,
diagrammed in Figure 6.7-1. Inside a plastic sheet are thiJl
parallel louvers which may be oriented at a specific angl,~
with respect to the surface normal. The zero degre,:
louvere . fllter has the louvers perpendicular to the filter
surface.

AVAILABLE

ASPECT RATIO AND

VIEWING ANGLE

Density

The operation of a louvered filter is similar to a venetian

blind as shown in Figure 6.7-2. Light from the LED display
passes between the parallel louvers to the observer. Off-axis
ambient light is blocked by the louvers and therefore is not
able to reach the face of the display to be reflected back to
the observer. This results in a very high contrast ratio with
minimal loss of display emitted light at the on-axis viewing
angle. The trade-off is a restricted viewing angle. For
example, the zero degree louvered fllter shown in Figure
6.7-1 has a horizontal viewing angle of 180; however, the
vertical viewing included angle is 60. The louver aspect

OPTIONS FOR LOUVERED FILTER:,-

ANY COMBINATION

2.75:1 = 60
2.00:1 = 90
3.50:1 = 48

of a 0 Neutral

IS POSSIBLE

LOUVER
ANGLE
OPAQUE BLACK
TRANSl UCENT GRAY
TRANSFARENTBLACK

ratio (louver depth/distance between louvers) determines

viewing angle. A list of louver option possibilities is given in
Figure 6.7-2.
Some applications require a louver orientation other than
zero degrees. For example, an 18 degree louvered filter may
be used on the sloping top surface of a point of sale
terminal. A second, is the use of a 45 degree louvered filter
on overhead instrumentation
to block out ambient light
from ceiling mounted lighting fixtures.
Louvered filters are effective fIlters for enhancing the
viewing of LED displays installed in equipment operating
under daylight ambient conditions. In bright sunlight, the
most effective fIlter is the crosshatch louvered filter. This is
essentially two zero degree neutral density louvered fIlters
oriented at 90 degrees to each other, as illustrated in Figure
6.7-3. Red, yellow and green digits may be mounted side by
side in the same display. Using only the crosshatch fIlter, all
digits will be clearly visible and easily read in bright sunlight
as long as the sunlight is not parallel to the viewing axis.
The trade-off is restricted vertical and horizontal viewing.
The effective viewing cone is an included angle of 40
degrees (for a filter aspect ratio of 2.75: 1).
Neutral density louvered filters are effective by themselves
in most bright ambient lighting conditions without the aid
of a secondary wavelength fIlter. However, colored louvered
fIlters may be used for additional wavelength fIltering at the
expense of display emitted light.
A most effective filter which provides exceptional contrast
in bright sunlight is the 45 louvered fIlter. The difference
from standard cross hatch is that the louvers are at a 45
angle with respect to the edge of the fIlter material as
shown in Figure 6.7-4. The louvers are transparent black,
with a cross light transmission of 12% to 15%. This small
amoun t of cross transmission virtually eliminates the
double image due to ghosting which was a problem in
earlier designs. The horizontal and vertical included viewing
angles are increased to about 60 (for a louver aspect ratio
of2.75:1).
A combination
tha t provide s significant contrast
improvement for standard red LED displays being used in
bright sunlight is the 45 cross hatch with a purple tint to
the fIlter plastic. The purple tint provides a color contras
for the illuminated display segments, so they may be
recognized when the sunlight rays are somewhat parallel the
viewing axis.
The combination of a 45 cross hatch filter and a fresne]
lens i tinted plastic form a special magnifying fIlter fOI
LED displays. Both the cross hatch and fresnel grooves are
made extra fme so that they are virtually invisible to the
eye. The result is a superior lens/filter combination 0

Figure 6.73 Conceptual Sketch of a Cross Hatc~ Filter" Two

Louvered Filters Bonded Together In a 90
Orientation with Respect to Each Other.

minimum thickness that produces a magnified digit which is

visible in bright daylight. Figure 6.7-5 illustrates a barrel
lens magnifier in combination with a 45 cross hatch for
use with unmagnified monolithic LED displays.
Louvered fJJ.ters are available with either a "light matte" or
"very light matte" anti-reflection surface which does not
produce a fuzziness to the display appearance. Another
option is a very hard scratch resistant surface which is
incorporated directly into the plastic.
3M Company,
Light Control Division, manufactures
louvered fIlters for LED displays. Their product trade name
is "Light Control Film", which is useable with all LED
display and lamp products.

TRANSPARENT
LOUVERS

BLACK

HAVE

12%

TO 15% TRANSMISSION
WHICH ALMOST
ELIMINATES A SECQNDAY
IMAGE DUE TO GHOSTING

Figure 6.7-4 45 Cross Hatch Filter for Use with LED Displays
in Bright Daylight Ambient.

Circular polarizing fJJ.ters are effective when used with LED

displays that have specular reflecting front surfaces.
Specular reflecting surfaces reflect light without cattering.
Displays that have polished glass or plastic facial surfaces
belong to this category.

Polaroid
Corporation manufactures
circ~~ar polarizing
fJJ.ters in the United States. In Europe, E. Kaseman of West
Germany produces high quality circuilar polarizers.
~

,~PURPLE
PLASTIC
1
FRESNEL LENS

PURPLE PLASTIC
45" CROSS HATOI

BOTH PLASTIC PIECES

BONDED TOGETHER
(GROOVES ON THE INSIDE)
TO FORM A SINGLE
FILTER/MAGNIFIER
o

Figure 6.7-5 A Fresnel Lens Magnifier/45

Cross Hatch Purple
Filter for Use with Unmagnified Standard Red
Monolithic LED Displays.

6.9
The op~ration of a circular polarizer may be described as
follows As shown in Figure 6.8-1, the fIlter consists of a
laminat~ of linear polarizer and a quarter wave plate. Th~
quarter wave plate has its optical axis parall:l to th~ flat
surface of the polarizer and is oriented at 45 to the lmear
polariz: tion axis. Non-polarized light is first line.arl:!
polarized by the linear polarizer. The linearly polanzeoi
light has x and y components with respect to the quarter
wave plate. As the light passes through the quarter wav~
0
plate, tile x and y components emerge 90 out of phasl
with e,ch other. The polarized light now has x and :r
forrnin~ a helical pattern with respect to the optical path,
and is termed circular polarized light. As this circular
polarized light is reflected by the specular reflec~in,~
surface, the circular polarization is reversed. When the hght
passes hack through the quarter wave plate it become,
linearly polarized at 90 to the linear polarizer. ThUi
reflectell ambient light is blocked. The advantage of ;\
circular polarizer is that reflected ambient light is reduced
by mor: than 95%. However, the trade-off is that display
emitted light passing through the circular polarizer i;
reduced by approximately 65% at the peak wavelength.
This thlm necessitates an increased drive current for th(~
display, more than that required for a wavelength fIlter.
Circula r polarizers
are normally colored to obtain
additional selected wavelength fIltering. One Caution
outdoor applications will require the use of an ultraviolet,
uv, fIlter in front of the circular polarizer. Prolongec
exposurll to ultraviolet light will destroy the fIlter':.
polarizir.g properties.

Anti-Reflection
Suggestions

Filters, Mounting Bezels and Other

Anti-reflection filters: A fIltered display still may not be

readable by an observer if glare is present on the fIlter
surface. Glare can be reduced by the addition of an
anti-reflection surface as part of the fIlter. Both sections of
the display shown in Figure 6.9-1 are fIltered. The left hand
fIlter has an anti-reflection surface while the right hand
fIlter does not.
An anti-reflection surface is a mat, or textured, fmish or
coating which diffuses incident light. The trade-off is that
both incident ambient and display emitted light are
diffused. It is therefore desirable to mount the fIlter as
close to the display as possible to prevent the display image
from appearing fuzzy.
Panel graphic ChromafJJ.ter~
come standard with an
anti-reflection coating. SGL Homalite offers two grades of a
molded anti-reflection surface. 3M Company and Polaroid
also offer anti-reflection surface options. Optical coating
companies will apply anti-reflection coating for specialized
applications, though this is usually an expensive process.
Mounting bezels: It is wise to take into account the added
appearance of a front panel that has the display set-off by a
bezel. A bezel of black plastic, satin chrome or brushed
aluminum, as examples, will accent the display and attract
the eye of the observer. The best effect can be achieved by
a custom bezel. Commercial black plastic bezels for digits
up to 7.62mm (.3 inch) tall are available.

Other
suggestions:
When designing the mounting
configuration of a display, consider recessing the display
and ftlter 6.35mm (0.25 inch) to 12.7mm (0.5 inch) to add
some shading effect. If a double sided printed circuit board
is used, keep traces away from the normal viewing area or

Figure 6.9-1 The Effect of an Anti-Reflection Surfaceon an

Optical Filter.

cover the top surface traces with a dark coating so they can
not be seen. Mount the display panel in such a manner as to
be easily removed if service should become necessary. If
possible, mount current limiting resistors on a. separate
board to reduce the ambient temperature in the vicinity of
the displays.

Ambient
Lighting
Panelgraphic Chromafiltelwith

Anti-Reflection

Ruby Red 60
Dark Red 63
Purple 90
Scarlet Red 65
Yellow 27

Standard Red

Green 48

Green

High-Efficiency
Yellow

Red

Moderate
Bright
Bright
Moderate
Moderate
to Bright
Moderate

SGL Homalite, Grade 100

H 100-1605
Standard Red
H10o-1804 (Purple)
H100-1670
H10o-1726
H10o-1720
H10o-1440
H10o-1425

High-Efficiency
Yellow

Red

Moderate
Dim
Moderate
Dim
Moderate

Anti-Reflection
LR-72; 0.5 inch (12.70mm) Mounting Distance From Display
LR-92; Up to 3.0 inch (76.20mm) Mounting Distance
From Display
Rohm & Haas
Plexiglas 2423
Oroglas 2444
3M-Company - Louvered Filters
R~6~
~an~~R~
Violet
Red 625
High-Efficiency
Amber 590
Yellow
Green 565
Green
Neutral Density
All

Red

Bright
Bright
Bright
Moderate
to Bright

Schott Optical Glass

RG-645
RG-630
RG-610

Standard Red
High-Efficiency Red

Bright
Moderate
Moderate

E. Kasemann GmbH
D 8203 Oberaudorf
Panelgr Iphic Corporation
10 Hen Jerson Drive
West CcIdwell, New Jersey 07006
Phone: (201) 227-1500

Chromafilter'ID - Wave
length filters with
anti-reflective coating;
Red, Yellow, Green

Thorn/I 'anelgraphic
Great Cambridge Road
Enfield, Middlesex
ENGLfl ND

SGL Homalite
Comtro lic GMBH
D8000 IJlunich 90
Theodo inden Str 4
GERM,IlNY

Wavelength filters; two

optional anti-reflectivtl
surfaces; three plastic
grades; Red, Yellow,
Green

Phone: (089) 643011

3M - Co npany
Visual P oducts Division
3M Cen'.er, Bldg. 220-10W
St. Paul, Minnesota 55101
Phon!: (612) 733-0128

3M - Brand
Light control film;
louvered filters

Plexiglas; sheet and

molding powder; wave
length filters

Roehm,'3mbH
Chemisclle Fabrik

Plexiglas; wavelength
filters

Polaroid :orporation
549 Technology Square
Cambrid~ e, Mass. 02139
Phone (617) 864-6000

Schott Optical Glass

Physikalische Optik
Jenaer Glaswerk Schott Gen.
Hattenberg Stra. 10
Mainz, W. GERMANY
Phone: (06131) 6061
Norbex Division
Griffith Plastics Corporation
1027 California Drive
Burlingame, California 94010
Phone: (415) 344-7691

DIGIBEZEL~ Plastic
bezels for LED displays

Industrial Electronic Engineers,

Plastic bezels for .30

inch (7,62mm) tall
LED displays

7720-40 Lemona Avenue

Van Nuys, California 91405
Phone: (213) 787-0311
Rochester Digital Displays, Inc.
120 North Main Street
Fairport, New York 14450
Phone: (716) 223-6855

Rohm and Hass

Independence Mall West
Philadelphia, Penn. 19105
Phone (215) 592-3000

Polarizer Division

Phone: (08033) 342

Inc.

3M Eure pe S.A.
53/54 A tenue Des Arts
1040 Bmxelles, Belgium
Phonn: 12-39-00

6100 Dal mstadt

Kirschen.tllee
WEST GI:RMANY
Phone: (06161) 8061

filters

WEST GERMANY

Duryea, Pennsylvania 13642

Phone: (717) 457-7485

Phore: 01-366-1291
SGL He malite
11 Broe kside Drive
Wilmington, Delaware 19804
Phone: (302) 652-3686

Circular Polarizing

Complete mounting kits

for H.P. 5082-7300,
-7700 and -7600
displays.

Photometry deals with flux (in lumens) at wavelengths that

are visi Ie, so the unit symbols have the subscript 'v' and
the unit names have the prefIx "luminous".
Radiometry deals with flux (in watts) at all wavelengths of
radiant energy, so the unit synbols have the subscript 'e'
and the unit names have the prefIx "radiant".
Except for the difference in units of flux, radiometric and
photometric
units are identical in their geometrical
concepts. Luminous flux is related to radiant flux by means
of the "luminosity function" , VA.' also known as the
"standard observer curve" or the "CIE curve". At the peak
wavelength, 555 nm, of the luminosity function, the
conversion factor is 680 lumens per watt. The luminosity
function is given in Figure 7.1-1 on a log scale, so the
accuracy is the same at all wavelengths. Values of VA.taken
from Figure 7.1-1 are accurate enough for all but the most
exacting calculations.
Figure 7.2-1

1.0
.8
.6

.2

a:
w
l>.
I

.10
.08

>
!::

.06

.04

>
i=
.
..J
w

a:
I
,<

>

.01
.008

'\

.006

.004

.002

.02

11

...J

'\

=>

"'- '\

en

"'"

!::
z
=>

..........

/'"

.4

Generic Terms and Symbols a

Relationships.

-I

1\

/
00

450

500

550
A -WAVE

600

650

ENGTH - nm

Figure 7.1-1 CIE Relative Ll minosity Function.

1 watt = 680 lu Tlens.

.. 1

At Peak,

700

75

For a spectrum of radiant flux, (dlPe/dA), the total radiant

flux i, obtained by spectral integration:

Luminous flux integration requires that each spect ral

element be weighted by the value of the luminos ty
functi on. It is therefore found as:
IP (Qn)
v

= 680 (Qm)
W

00

dIP

(__ e)

VA dA

dX

Conve.sion of photometric units to radiometric units is

conveniently done by the luminous efficacy, 17v' which is
just the ratio of the luminous flux to the radiant flux:
Qnl
lPy (Qm)
17 (_.)=--y W
lP (w)

Adding
subscripts
and prefixes quantifies
radiometric or photometric units, e.g.:

Iv - LUMINOUS INTENSITY (lm/sr) or candelas (cd). cd = Im/sr

lumens

these

as

per steradian

As a practical matter, flux is used mainly in relating to the

other terms. Few applications actually utilize all the flux
available from a source. The same is true for exitance. The
only situation in which "exitance"
and "flux" have
practical significance is with a receptor so tightly coupled
to a source that virtually all the flux leaving the source
enters
the receptor,
such as in "sandwich"
type
optoisolators.

Luminous efficacy can be evaluated from any relative

spectmm,
i.e., one which is normalized,
since the
normdization
factors cancel in the numerator ard
denoninator
of equation (7.1-3). Typical values of
luminous efficacy for LEDs are:

"STAI'\ DARD RED"

HIGH-I:FFICIENCY
RED
YELLOW
GREEr,1

ApEAK(nm)

17y(Qm/w)

655
635
585
565

60
135
540

640

Generic. Illy, there are only five units (and symbols) f01
radiant .mergy, as illustrated in Figure 7.2-1, and quantified
in Table 7.2-1 and Table 7.2-2.
- FLUX, describes the rate at which energy is passing to,
from, or through a surface or other geometrical entity.
E - INCIDANCE, describes the flux per unit area normally
(perpencjcularly) incident upon a surface.
M - EXIT ANCE, describes the flux per unit area leaving
(diverging) from a source of finite area.
I - INTENSITY, describes the flux per unit solid angle
radiating (diverging) from a source of finite area.
L - STI!.RANCE, describes the intensity per unit area of a
source.
w - SOLID ANGLE; a solid angle, w, with its apex at the
center 0:' a sphere of radius, r, sub tends on the surface of
that sphere an area, A, so that w = A/r2 in steradians (sr).

Incidance has the same units as excitance but is of a vastly

different nature. Exitance ignores the direction taken by
the exiting flux. Incidance takes account of the flux
component which is normal to a surface. If the flux
direction at a surface is not normal, then the incidance is
just the normal component of the angularly incident flux
density.
Incidance
is most
useful
in describing
photodetector properties.
Intensity
is an extremely useful concept in both
photometry
and radiome try, and the candela unit of
luminous
intensity
is the only universally utilized
photometric
unit. (Other photometric
quantities are
plagued by a profusion of English, metric, MKS, and CGS
units of length.) Since flux passing through space is usually
divergent, it is usually possible to define an equivalent point
from which it diverges in terms of a solid angle, the flux
therein, and hence the intensity of the equivalent point. As
seen in Section 7.3, intensity is the most easily measured
and most uniformly repeatable quantity. Therefore,
although sterance is more fundamentally significant in
many applications, the actual parameter to be measured for
performance verification should be the intensity.
Sterance is most significant because of its constance in an
optical (lens) system. This is discussed in much detail in the
National Bureau of Standards publication NBS Technical
Note 910-1 issued in March 1976. While magnification and
image position varies, sterance does not.
Luminous sterance (or luminance) is the luminosity basis
for visibility (i.e., distinguishing an object from its
background).
As such, luminous sterance is a useful
measure for contrast and photographic exposure. However,
luminous sterance is not the only basis for visibility -- color
is another. For example, spots of brown gravy on a green
tie are visible, even if the spots have the same luminous

Term, Definition

Symbol

FLUX, rate of flow of energy, Q

Q - radiant, Q - luminous,
e
v
Q - photon [see note 1]
q

<P

INCIDANCE, flux per unit area

on a reception surface

EXITANCE, flux per unit area

from an emitting surface

I Y, nux per Unit

solid angle from a remote source

STERANCE, flux per unit solid

angle per unit area of emitting
surface at angle 0 with respect
to surface normal.

New Term
(Old Term)

Defining
Equation
dQ
dt

cJt
dA

~
dA

c&
dw
dl
dA cos

dW\dA cos

Symbol

SI Units, Abbr.

New Term
(Old Term)

Symbol

SI Units, Abbr.

qJv

lumens,

Equivalent

1m

Radiant Flux
(Radiant Power)

<Pe
(P)

watts, W

Luminous Flux
(Luminous Flux)

Radiant Incidance
(lrradiance)

Ee
(H)

watts per sq.

meter, W/m2

Luminous Incidance
(Illuminance or Illumination)

E
v
(E)

lux, Ix
[see note 3]

lumens per s~.

meter,lm/m

Radiant Exitance
(Emittance)

Me
(W)

watts per sq.

meter, W/m2

Luminous

lumens per s~.

meter,lm/m

[see note 2]

Radiant Intensity
(Radiant Intensity)

Ie

watts per
steradian, W/sr

Luminous Intensity
(Lu minous Intensity)

Iv
(I)

candelas,

(J)

lumens per
steradian, Im/sr

Radiant Sterance
(Radiance)

Le
(N)

watts per
steradian per
sq. meter,
W/sr/m2

(Luminous Sterance
(Luminance)

candelas per sq.

2
meter, cd/m
[see note 3]

Exitance

(B)

0)

Note 1. Quantametric
terms use the prefix word "photon"
and their symbols have a subscript,
Note 2. Lux and other units of luminous incidance DO NOT APPLY to luminous exitance.
Note 3. Other units in Table 7.2-2.

"q".

cd

lu mens per steradian per sq.

meter,lm/sr/m2
nit, nt

In general, confusion is averted by use of self-explanatory units, such as: lumens per square meter for luminous incidance; and, candelas per
square meter for luminous sterance. At times, for brevity, where the risk of confusion is negligible, other units used are:
Luminous Sterance
Equivalent
Unit, Abbr.

Footlambert, fL

1/rr cd/cm2
1/rr cd/ft2

Apostilb, asb

1/rr cd/m2

*Stilb, sb

cd/cm2

Candelas per sq. foot

cd/ft2

**Candelas/sq. meter

cd/m2

Lambert, L

=
=
=

Luminous Incidance
Unit, Abbr.
Equivalent

=
=

Lambert
L

Footlambert
fL

Apostlib
asb

*stilb
sb

cd/ft2

**cd/m2

nit

929

10,000

.3183

295.7

3183

3183

.001076

10.76

.0003426

.3183

3.426

3.426

.0001

.0929

.00003183

.02957

.3183

.3183

3.1416

2919

31,416

929

10,000

10,000

.003382

3.1416

33.82

.001076

10.76

10.76

.00031416

.2919

3.1416

.0001

.0929

*Phot
ph

*Footcandle
fc

**, *Lux
Ix

929

10,000

* Footcandle, fc

Im/cm2
Im/ft2

.001076

10.76

**, *Lux, Ix

Im/m2

.0001

.0929

*Phot, ph

*CIE Unit
**Recommended SI Unit
Many of the other units
are in common usage, but
efforts are being made to
standardize on the SI units.

USE OF TABLE: In any row, the quantities are all equal; for example, in the table for luminous sterance, cd/m2
= 3.1416 asb = .0001 sb = .0929 cd/ft2 = 1 nt; in the table for luminous incidance,lx = .0001 ph = .0929 fc.

.00031416 L = .2919 fL

DEVICE

STERANCE
INTENSITY

~PPARENT
MITTING
AREA
(mm2)

DESCRIPTION

cd/m2 per mcd

, Lilv

fL per mcd

LEDs

DOT MATRIX

2.0
18.0

DISPLAYS

LARGE 7-SEGMENT

97
580
240
970
730
150
16

11,000

3,200

230
590

66
170

0.18
0.24

5,600
4,200

1,600
1,200

0.17
0.15
0.18
0.27

5,900
6,700

1,700
1,900

5,600
3,700

1,600
1,100

24,000
12,000
9,100
7,700
16,000

7,100
3,400

0.09

DISPLAYS
4.4

10.9 mm (.43 in.) TALL

7.6 mm (.3 in.) TALL
MONOLITHIC

330
2,000
830
3,300
2,500
500
56

3.0
0.5
1.2
0.3
0.4

T-1 3/4 SIZE DIFFUSED

UNDIFFUSED
T-1 SIZE DIFFUSED
UNDIFFUSED
SUBMINIATURE
HERMETIC
RECTANGULAR

1.7

7-SEGMENT

LARGE CHIP (5082-74XX)

SMALL CHIP (5082-743X)
CALCULATOR

DIGITS

8, 9 CLUSTER (5082-7440 SERI ES)

12, 14CLUSTER
(-7442/4/517)
8,9 CLUSTER (5082-7240 SERIES)
5,15 CLUSTER (-7265/75/85/95)
CHIPS
1.35
1.42
1.91
2.34
0.38

x
x
x
x
x

1.50
2.24
2.72
3.25
0.38

mm
mm
mm
mm
mm

(-7811/21)
(-7832/42)
(-7851/61/52/62)
(-7871/31)
(-7890/93)

0.041
0.085
0.11
0.13
0.063

2,700
2,200
4,600

Lv (cd/m2)
Iv (mcd)

2
_Lv_(_f_L)_
= _Lv_(_fL_)_( 1T Lv (cd/ft )
Iv (mcd)

Iv (mcd)
= 1T

I (~(mcd)

Lv (fL)

03 I (cd)
v

) = ..!I..- __ 1 _ ( _92_9_0_3_A_(_ft_2)
103

A (ft2)

92.9

A (mm2)

TABLE

7.2-3

Intensity-to-Ste

L
ance Conversion (_v
Iv

1)
A

A(mm2)

To understand the relationship between sterance and

intensity, consider a uniform source whose area is one
square meter and having an intensity of 1000 candelas. Its
sterance is then 1000 cd/m2. If this area were subdivided
into 106 pieces, each having an area of one square
millimeter, the intensity of each piece would be 10-3
candelas or one millicandela, but the sterance of each piece
would still be 1000 cd/m2.

sterance as the tie. On a brown tie of the same color as the

gravy, a spot would be visible only if its luminous sternce
is different from that of the tie by a factor of two or mOre.
While luminous sterance is the basis for visibility, lumin,ms
inten:;ity is the popularly preferred performance parameter
for LEDs and LED displays. The basic reasons are:
RELEVANCE - Most applications utilize the light radiated
by all portions of the emitting area, but sterance i:; a
measure of the intensity of only a small incremer tal
portion. A large area device could have a higher intemity
than 1 device of higher sterance but smaller area. Luminous
intem.ity, being the product of sterance times area, ranks
devices properly according to their effectiveness.

Sterance can be an important consideration in product

selection for a given application but is a poor parameter for
specification.
The figures in Table 7.2-3 give the
sterance-to-intensity ratios for a number of product groups.
They are simply the inverse of emitting area, with
appropriate adjustment of units:

VIEU'lNG ANGLE (RADIATION PATTERN) - Sterance

of an LED remains nearly constant at all viewing angles hut
the projected light-emitting area does not. Luminous
intem,ity, the product of sterance and area gives pro:)er
radiation pattern information.
EASE OF MEASUREMENT
- Sterance measurem~nt
requi~es an optical system to image and field-stop the
emitting surface; intensity measurement requires only a
calibrated detector and a scale to find the LED-to-detector
distance.

\~

CIE PHOTOPIC
RESPONSE

\~ I! '\
I

I
UJ

VI

2~
C)

60

0..
VI
UI

a:
UJ

>
~: 40

RED LED OUTPUT

_.i
UI

a:

PHOTOMETRIC
DETECTOR
RESPONSE

o
400

600
A "':WAVELENGTH
Figure 7.3.1-1

- nm

Spectral Response of a Photometer

Spectrum Radiation.

For Broad-

In radiometry, spectral effects are not much of a problem

because radiometers can usually be made to have very
nearly flat spectral response over the spectrum of interest.
The flatest and broadest response is obtained from
thermal-type
radiometers
(e.g.
thermopile
or
thermocouple).
Quantum-type
(e.g. silicon diode)
radiometers are much more sensitive than thermal-type, and
have a higher speed of response; while the spectral flatness,
is not as good, it is adequate for most LED measurements.
In photometric measurement of LEDs, it is important that
the photometer be calibrated at the wavelength of the LED
spectrum. This is necessary because most photometers do
not have a spectral response that fits the luminosity
function precisely at all wavelengths. To achieve the best
overall fit, the response is adjusted (by colored mters) so
that, over the spectrum, negative and positive deviations are
balanced. This provides precise calibration for photometry
of incandescent and other broad-spectrum sources, but for
the narrow-spectrum LED sources, the entire spectrum
might lie within a wavelength region where the photometer
deviation is all positive or all negative, as seen in Figure
7.3.1-1.
In addition, for red LEDs, the spectral slope of the
photometer must be very close to the -.25 dB per nm slope
of the luminosity function. This is because LED spectra
change with temperature and drive conditions, shifting as
much as 10 nm. If a photometer has -.20 dB/nm (instead
of the correct -.25 dB/nm), the incremental error would be
0.5 dB (+12%).
Suitable
photometers
Photo-Research Corp.,
Technology.

are available
from EG&G,
Tektronix, and United Detector

If the source to be measured has a plane of emission with

respect to which intensity can be defmed, then all that is
necessary for intensity measurement is a photometer to
read the incidance, Ev at a distance, d, from the source. The
governing relationship is illustrated in Figure 7.3.2-1a. The
flux, 'Pv (lm) in the cone may be referred to the source as
the product of the intensity, Iv (lm/sr) times the solid
angle, w (sr) of the cone, or it may be referred to the sensor
as the product of the incidance, Ev (lm/m2) times the area
A (m2). Since both relationships describe the same flux,
they can be equated:

If the distance, d, is large enough, then the value of w can

be expressed as:

The fact that the left side is in lumens per steradian while
the right side is in lumens may seem incorrect, but note
that the steradian is a dimensionless unit.
The relationship, Iv = Ev d2, is valid only if d is lar~e
enough that the solid angle can be described as w = A/d .
For the error in this assumption to be less than 1%, it is
necessary that the distance, d, be at least ten times the
diameter of the photodetector or ten times the diameter of
the source, whichever is greater. This is called the TEN
DIAMETERS RULE. Its purpose is to make the solid angle
small enough that the flat photometer receptor can be
regarded as a portion of a spherical surface.
To be absolutely correct, intensity should be measured with
an incrementally small solid angle. The TEN DIAMETERS
RULE defines a solid angle of 7.84 msr which is slightly
larger than the 4.0 msr recommended by experienced
observers.
When the distance, d, cannot be defined, intensity
measurement can be made by placing the photometer at
two different distances from the source, as in Figure 7.3.1-1
(b and c). With the "approximate" method, making e small
improves the accuracy, as far as equation (7.3.2-2) and
(7.3.2-3) are concerned, but if e is too small there may be
other problems. One other problem is getting enough flux
for an adequate signal-to-noise ratio. Another is the possible
variation of the ratio 'P/w; undiffused LEDs, especially with
very narrow radiation pattern, can cause the 'P/w ratio to
vary substantially.
The "precise" two-point method,
reduces potential radiation pattern problems by keeping the
solid angle nearly the same for both observations. Note in
the expression for Iv by the "precise" method, that
provision is made for 'P2 'P1; however, if optical bench
positioning can be adjusted so that 'P2 = 'PI' the accuracy
will be limited mainly by the accuracy obtained in
measuring Al and A2- These areas can be found as a ratio
to the photometer area, AO' as follows:

1.

Place the photometer at large distance (>lm) from

a small source
lOmm)
of high intensity.
(Wavelength and spectral effects are unimportant
here.) Note the photometric reading as EO'
Place Al over the photometer
as E1 Then Al = AO(E1/EO)'

and note the reading

PHOTOMETER
SENSOR

rfiv=lvw=EvA
IF d

(a)

D, THEN A "" wd2 AND Iv "" Evd2

RELATIONSHIP
BETWEEN LUMINOUS INTENSITY.
Iv. OF AN LED
AND LUMINOUS INCICANCE. Ev AT THE rHOTOMETER

w,

READS:.~~

READSE.

A/
2
~

= A

= ~

v2

v1

AREA OF PHOTOMETER
0

(e)

PRECISE TWO-POINT IVIETHOD USES TWO APERTURES. A1 AND A2

SO THE SOLID ANGLf: IS NEARLY THE SAME AT EACH POINT.
THIS REDUCES ERROR DUE TO RADIATION
PATTERN PECULIARITIES.

3.

Place A2 over the photometer

as E2. Thus A2 = AO(E2/EO).

and note the reading

Several observations of Al and A2 at different

from the source will improve the accuracy.

distances

Sterance, being intensity per unit area, requires apparatus

capable of resolving an incrementally small area. It is not
actually necessary for this area to be precisely known
because the measurement is usually done by comparing the
unknown sterance to a sample of known sterance.
Figure 7.3.3-1 shows the arrangements that can be used for
sterance measurement. Apparatus utilizing the beam splitter
and aperture field stop is available commercially from
Photo-Research Corp. and is called a brightness spot meter.
Apparatus using the fiber-optic at the coaxial image plane is
employed in the photometric microscope from Gamma
Scientific Co. In both of these systems, a viewer can
directly observe the precise portion of the LED for which
the sterance is being measured. The area increment is the
area of the fiber-optic core (or of the aperture field stop)
divided by the square of the magnification of the objective
lens. Spot diameters as small as .02 mm (.0008") can be
resolved. The smaller the spot used, the larger will be the
sterance variation observed across the surface of the source.
This is the reason why sterance is a difficult specification to
describe or use.

PLANE GLASS AT IMAGE

SUPPORTS END OF FIBER-OPTI(
OR BEARS RETICLE FOR
AIMING APERTURE FIELD STOf

Total flux from a source is the integral of the flux radiated

in all directions. If a source is placed at the center of a
sphere, the total flux can be described as the integral of the
incidance over the entire inside surface of the sphere. The
completely general expression would require integration
over both the azimuth angle (0 to 21T) and the polar angle
(0 to 1T). However, most sources have radiation patterns
that are very nearly a surface of revolution about the
optical axis, so the integral is simplified as seen in Figure
7.3.4-1. That is, for a particular value of polar angle, e,
(also called the off-axis angle) the intensity is constant all
the way around the polar axis. The differential flux can
therefore be described as:

'P(e)

'P =

I (0) 2

1T

sine de

1T

I (e) 2

1T

sine de

FIBER-OPTIC
FIELD STOP
EYEPIECE
ASSEMBLY

__ --

-------,.

--------==- __

~~
OBJECTIVE LENS

---

---

--

I
J

-,-Y

BEAM SPLITTER

L_.
APERTURE ~
FIELD STOP

dl{)= I (0) x dw
= 10 Ir (0) x dw

dw

= 21T

sin

0 d0

Ir (0)

21T

sin

1T
I{)

Figure 7.3.4-1

fir

(0) 2 1T sin 0 d 8

00

RADIATION
PATTERN

Ir (0)

POINT SOURCE

1.00

NONE

21T

LAMBEFlTIAN

*oos 0

1I = 60
3

1T

COM MOl \I
APPROXIMATION
FOR LEl>s

*cosn 0

005-1 ('51/n)

I{)/Io

I{)(O)/Io

0%

(1-005 0 )

41T

sin2 0

21T

(1_oosn+1
n+1

1T

0)

21T

n+1

Flux-to-lntE nsity Ratio; Analytic Integral of

Radiation P~ttern.

*bponent
n is derived from the LED radiation pGttern and usually is a valid approximation out to 0y", where
1(6) =.510. The value for n may be determined as follows: Assume cos" (Ox) = y, then n = Log(y)/Log cos Ox.

I{)(O) and I{) can be related to the axial intensity since the
intensi ty at any angle, 0 can be given as the product of the
axial intensity, 10, a constant, and the relative intensity,
Ir(O), for which Ir(O) = 1.00 as in customary radiaticn
patten s. The axial intensity, of course, would be eithilI
licd) Jr le(wjsr).
I{)(O)

= '0

S Ir (0)
o

21T

sinO dO

(7.3.4-5)

The term [2 1T sinO t:, 0] is called a zonal constant, CZ' in

which the incremental angle, t:,0, is in radians. Of course,
the magnitude of C varies as sinO, and depends on the size
of the t:,0 increment. A little experience will show that
adequate accuracy (3%) is obtained by selecting M equal to
one fourth of the angle at which I/O) = O.s that is
M = (0!h)j4. Thus, if there are to be N equal parts over
0
0
180, choose N= 180 (4j0!h) = 720 j0!h rounding off to
the next larger integer that yields convenient values for

Figure 7.3.4-1 gives the results of integrating radiatioh

patterns with analytic functions. Also given is the value c f
0!h which is the angle at which Ir(O) = !h.
If the adiation pattern is not analytic but is a surface of
revolutIOn about the polar axis, then the integration is
performed by converting equation 7.3.4-5 to a summation

0= 180 jN.

Figure 7.3.4-2 shows how 0 and N are used in zonal

integration. The customary method appears acceptable in
Figure 7.3.42, but if 0 had been S, then it would have
required using the values 2.5, 7.s, 12.5, etc. for I/O)
and Cz(O). An accuracy check, using the analytic cosnO
radiation pattern, shows that with the customary method
I{)(O) is slightly above the analytic (correct) value, while the
modified method yields I{)(O) slightly below the correct
value. Both methods have adequate accuracy.

e-p
10

10

LAR ANGLE
30

20

4fJ0

- DEGREES
60

50

90

80

70

~
Z

::l
0
3.0
Ir (e) [2 1rsin

eJ

~
~
2.0

'i;j
l:::
N

1.0

~
90

..

0
3

--.\
M
cp(Mo) = L
I (m 0 - o
- ) Cz (m 0 - -0 )
2
2
m=1 r

(a) CUSTOMARY

METHOD

f"e

o = 180
N

~M
2

- MAY R QUIRE AWKWARD

(e) = 21r

sin e

VALUES

FOR (mo -0/2)

CI.I

I-

::l
3.0
Ir (e) [2 1r sin

eJ

0
Z
I

~
2.0

';;;
l:::

N
~

1.0

~M

cp(Mo) = ~

M-1
Ir (Mo) Cz (Mo) + ;=1

Ir (mo) Cz (mo)

Zonal-Cq,nstant
N ;;;;'720 1Bv.-

Flux Integration;

z (e)

3% Accuracy with

= 2 1r sin e

..

LED optoelectronic devices are solid state semiconductor

p-n junctions. They are fabricated and packaged utilizing
many of the technologies
which are used in the
manufacture
of silicon and germanium semiconductor
products. It is reasonable to expect, therefore, that LED
products will exhibit reliability aspects similar to those
observed in these other semiconductor technologies. An
LED offers many reliability advantages with respect to
other available light sources. Long life-time, wide operating
temperature
range, and reliable operation in adverse
environmental conditions are only a few of the important
attributes of LEDs when compared to other light emitting
technologies. These attributes can, however, only be
realized when a device is properly designed, fabricated,
packaged, and operated. Optoelectronic device data sheets
define the limits within which devices should be operated in
order to achieve the expected reliability performance. It is
important that the optoelectronic system designer be aware
of how variations in these limits will effect the long term
performance of a device.
The reliability
of an optoelectronic
device can be
considered to be dependent upon two separate variable
factors. These are the reliability characteristics of the LED
semiconductor
chip and the reliability characteristics
inherent to the package into which that chip is assembled.
The variables are not necessarily independent because, for
instance, package failure which is not in itself catastrophic
can trigger
eventual failure or degradation of the
semiconductor device.

From a theoretical standpoint, an LED like any p-n

junction should have a nearly infinite lifetime. This
operating life would be limited only by the infinitely slow
natural diffusion of the dopants. In practice, an LED device
when operated under reasonable stress conditions is
normally expected to exhibit an operating lifetime of about
lOOK hours. In some cases, the end of operating life is
defined as the point at which a device light output has been
reduced to 50% of the initial value. This could occur slowly
over an extended period or catastrophically as a result of a
short or an open. The value of 50% has been chosen as a
ratio at which an observer would be expected to be capable
of detecting a visible change in the light output. There are
no known failure modes in LED semiconductors which
should result in occurrence of catastrophic failures after
extended periods of operation (such as notching in a
tungsten mament incandescent lamp which will result
eventually in a catastrophic failure of the mament).
Conversely, the slow reduction of light output with time is
a well known (though not well understood) characteristic

of an LED device. This degradation of light output exhibits

a direct correlation to the current density in the junction.
Figures 8.1-1 and 8.1-2 depict the normalized light output
vs. time for direct gap and indirect gap plastic package
lamps when operated at various junction current densities.
As can be seen, increased current density resulted in a more
rapid degradation and greater total degradation for stress
times up to about 5000 hours. The maximum operating
levels for the devices of Figure 8.1-1 represent a stress level
of 200% of maximum device rating and 350% of maximum
device rating in Figure 8.1-2. Figure 8.1-3 represents long
term operating life data for hermetic package lamps
operated at the maximum allowable device stress level for
stress times up to 40,000 hours. This data represents the
potential for LED devices to be successfully used for
periods in excess of lOOkhours.
Close inspection of Figures 8.1-1 to 8.1-3 will reveal that
the rate of light output degradation decreases as a function
of increasing stress time. Figure 8.1-4 depicts the range of
the expected rate of degradation (in % per 100 hours) vs.
stress time. This plot represents a composite of several
different device types from many different production lots
when operated at 20 mA dc. It is important to note that
with increasing stress time, even those device lots which
start out with a very high rate of degradation converge
quickly to a very low degradation rate. This implies that
devices
showing a high initial degradation will not
necessarily continue to degrade to failure.
The degradation of LED light output demonstrates some
dependency on junction temperature. As can be seen in
Figures 8.1-1 and 8.1-3, however, this dependency is so
small as to be inconclusive.
The data sheet maximum dc operating level for an LED is
generally established somewhat below the level at which
unacceptable degradation would be encountered within 10k
hours of operation. To achieve the lowest possible LED
degradation consistent with useful operating levels, it is best
to select an operating point at 40% to 50% of the maximum
data sheet rating.
8.2

Reliability Aspects of LED Packaging

The package into which a semiconductor

device is
assembled should provide environmental protection for the
device as well as providing electrical contacts and a
mounting technique. For an LED, the package is often also
required to provide an optical system to enhance coupling
of the emitted light to the detector. An ideally designed
and utilized package would perform these functions very
reliably. There are, however, many variables which can
affect this performance. To evaluate a package, tests such as
solderability, temperature cycling, thermal shock, moisture
resistance, mechanical shock, acceleration and terminal

1.1

100C STORAGE

>
Ien

1.0

z
w

I-

en

::> 0.9
0

:2:

::>
..J
Cl

0.8

N
..J

O.L. @ 100mA, 8SoC

:2:

(129 A/cm2)

a:
0

0.7

Figure 8.1-1

>

I-

enz

Direct Gap LED Lamp Normalized Luminous Intensity

ys. Operating !.ife and High Temperature Storage
T-l 3/4 Plastic Packages.

1.0

IZ

o
z
:2:

0.9
O.L. @20mA, 2SoC
(80 Alcm2)

::>
..J

0.8

N
..J

:2:
o
07
Z
a:

O.L. @30mA, 25'C

(120 A/cm2)
O.L. @7SmA, 2SoC
(300 A/cm2)

Figure 8.1-2

Indirect Gap L ED Lamp Normalized Luminous

Intensity ys. Operating Life and High Temperature
Storage T-l PI.lstic Package.

water vapor, and long term exposure to high humidity will

eventually subject the active elements to high humidity.
The time required for moisture to saturate most plastic
packages is in the order of 100 hours. High humidity can
lead to failure from corrosion of the metallization or from
increased surface leakage currents. The humdity associated
with most normal
applications
does not harm
plastic-packaged devices, but accelerated humidity testing
should be considered a destructive test. Most units can be
expected
to withstand
10 days of non-operating
environmental
testing in high humidity and elevated
temperature (e.g. 95% RH at 65C).

Figure 8.1-3

Direct Gap LED Lamp Normalized Luminous

Intensity vs. Operating Life TO-18 Hermetic Package.

Another package-related limitation of plastic products lies

in the area of thermal fatigue life. Because the different
materials (lead frame, plastic package, junction coating, die,
bond wires, etc.) are in physical contact, and because
coefficients of thermal expansion cannot be perfectly
matched, temperature changes are accompanied by physical
strain. Extended temperature cycling or thermal shock can
lead to fatigue failure of the bond wires, or to die attach
failures. It is not unusual to see 5% failures after 50 cycles
temperature cycling _55C to +125C or thermal shock OC
to +100C.
Thermo-mechanical life could be a limiting factor for
plastic products in markets where reliability is required
through many temperature excursions, such as automotive
applications. A sizeable effort of product improvement is
currently underway to develop the materials that will
realize the reliability objectives of these new markets.

z
w4

a:

z
o

i=3

a:

&12
o
X

::E

1000

100
TIME IN HOURS
Figure 8.1-4

Expected Rate of Degradation vs. Time for T-1 3/4

Plastic Package.

strength are performed at various stress levels. These tests

can be used to establish the worst case environmental
stresses which should be applied to the package. Figure
8.2-1 is a copy of a typical plastic lamp reliability data
sheet. Generally, such a data sheet will provide a reasonably
good analysis of the performance a user can expect from a
device. Deviation from this performance can, however, be
caused by either variations in the manufacture of the
package or by exceeding the defined environmental
conditions in the end use product. For plastic packages,
several environmental aspects are very important.

There
are theoretical
limitations to the maximum
temperature at which the device materials are stable and
compatible.
The gold wires bonded to aluminum
metallization form intermetallic compounds at a rate which
depends
very strongly on temperature.
At ordinary
temperatures the reaction is so slow that it is not a
significant contributor to failure rate. At +150C bond life
may be a few months. At +200C, bond life may be a few
days.
Trace impurities and bonding conditions can
significantly affect bond life. Stress from surroundings
package
materials can break a bond weakened by
intermetallics.
The thermal expansion coefficients of epoxy package
ma terials are relatively constant over normal service
temperature, but at some high temperature the expansion
coefficient increases significantly. This temperature is called
the "glass transition temperature", or Tg. Tg of most of the
presently utilized epoxy casting compounds is specified at
+120C, but it varies by several degrees, depending on
batch history and assembly conditions. Also, the glass
transition effect occurs over a range of several degrees, so
that any exposure over a +115C is potentially harmful.
The epoxy and dye materials, being organic, are also subject
to chemical degradation if overheated.

If pro er care is not taken in assembly, plastic packages are

potentially
susceptible to the occurrence of thermal
intermittents.
The solid encapsulation around the bOlkd
wires keeps adjacent parts of a broken wire together, :;0
that a defective bond, or a bond which failed for some
reason, may not be a consistent failure. A unit with tl:is
phenomenon will often be good at room temperature and
open when hot. A continuity test monitored ov~r
temperature ranging from room ambient to the maximu
storage temperature is quite effective at detecting therm I
intermittents.
A high temperature function test is Ie s
effective, but still quite good.
Failur resulting from improper soldering operations ha'ie
occurred frequently in LED applications. This is general y
the re ult of soldering at too high a temperature and/or for
too I ng a time period. As pointed out above, exce,s
temperature above Tg, can cause substantial changes in the

The following cumulative test results have

in accordance with the latest revisions of N
Ml L-STD-202 and Ml L-STD-750. Because
intended for commercial and industrial ma
requiring military high reliability performa
MIL-STD-750
REFERENCE

TEST

mechanical properties of the epoxies commonly available

for LED prq.ducts. Since most devices are relatively small,
there is very little thermal capacitance available to
distribute heat conducted into the package from the leads.
Exposure of the epoxy to extreme excess temperature
conditions can result in softening of the epoxy around the
leads where mechanical stress could cause loosening of the
die attach or lead bond posts or the formation of voids in
the epoxy in the vicinity of the leads. Either condition may
result in catastrophic failure of the device from broken
wires or wire bond separation.
From a manufacturing standpoint, there has historically
been a high degree of parametic variability in the epoxy
materials used for LED encapsulation. Incoming inspection
criteria such as gell time, hardness, darty, and viscosity can
be utilized by the LED manufacturer to improve product
uniformity.

leen obtained from testing performed at OED Division,

ilitary Semiconductor Specifications Ml L-S-19500,
this device has a non-hermetic, cast epoxy enclosure
kets, we do not recommend its use in applications
Ice.
UNITS
TESTED

TESr CONDITIONS

FAILED

Physical
Dimensions

2066

Device pr Jfile at 20X

50

Solderability

2026

SN 60, Pl 40, solder at 230 C

50

Temperature
Cycling

1051.1

5 cycles f om -65C to +100C,

.5 hrs. at 'lxtremes, 5 min. transfer

50

Thermal
Shock

1056.1

5 cycles f, om OC to +100C,
3 sec. trar sfer

50

Moisture
Resistance

1021.1

10 days, ~098% RH, 10 to +65C,

non opera ling

50

Shock

2016.1

5 blows e, ch Xl, Yl, Y2,

1500 G, 0.5 msec. pulse

Vibration
Fatigue

2046

32 8 hours each X, Y, Z,
96 hr. total, 60 Hz, 20 G min.

50

Vibration
Variable
Frequency

2056

4, 4 minwe cycles each X, Y, Z,

at 20 G min., 100 to 200 Hz

50

Constant
Acceleration

2006

1 minute ( ach Xl, Y 1, Y2, at

20,000 G.

50

Thermal
Strength

2036.3

Condition A, 1 lb. for 10 sec.

50

Lead Fatigue

2036.3

Condition F, Method A, 8 oz.,

10 sec.

50

Figure 8.2-1

Typical Mechani,:al Reliability

T-1 3/4 Plastic Llmp.

Data Sheet for a

MECHANICAL
HANDLING
FOR LED DEVICES

CONSIDERATIONS

The fir;t
stage in designing a circuit utilizing an
optoelectronic device is selecting the proper device for the
application. The second step is to establish the electrical
operating conditions and design the circuit. The third step
is to install the optoelectronic device into the physical
assembly, be it a printed circuit board, front panel
mounting or some other mounting arrangement. The
moun ting considerations
are primarily mechanical in
nature, requiring attention to such items as the similarity of
LED packages, the bending of leads, silver plated lead
frames, soldering and post solder cleaning, socket mounting
and heat sinking if required. Reliable operation of the LED
device is more positively assured when all of these
mechanical
considerations
have been given careful
attention.

TO-18 header. The devices will withstand considerable

mechanical and temperature stress without any effect upon
performance.
BALL WIRE BOND
BONDING

WIRE

TO TOP CONTACT

WEDGE BOND TO
ANODE LEAD

LED, DIE ATTACHED

TO CATHODE LEAD

L _ __

I
DEVICE PACKAGE
~FORMEDBY
ENCAPSULANT

Most plastic encapsulated LED devices are assembled using

the lead frame technology. The exceptions are some
stretched segment display devices which are assembled on
substrates. Independent of the lead frame or package
design, a lead frame device has the LED die attached
directly to one lead and wire bonded across to another lead
as shown III Figure 9.1-1.
'
Figure 9.1-2

The primary thermal path to the LED is the cathode lead.

Any mechanical and thermal stress applied to the leads is
transmitted directly to the LED, die attach and wire bonds.
The plastic encapsulant forms the device package and is the
only supporting element for the lead frame. Therefore, the
integrity of the encapsulation must be maintained to insure
reliable operation of the LED device throughout its
expected operating life.
Devices that do not use an encapsulating epoxy are most
likely assembled on a ceramic substrate. Thick film
metallization is usually applied to the face of the substrate.
The LED is die attached to one metallization pad and wire
bonded a~ross to another pad, as shown in Figure 9.1-2.
The circUlt metallization is the primary thermal path from
the LED 1:0 the leads, and the substrate acts as a secondary
thermal p:lth to the external ambient. The substrate isolates
the LED~; and wire bonds from mechanically applied
stresses; b.e amount of isolation dependent upon the type
of substrate and package configuration. Devices of this
construction are mechanically more rugged than a plastic
encapsulated device and are more tolerant of temperature
extremes.
One other type of device is assembled on a TO header such
as the high-reliability LED lamp which is packaged' on a

Basic Construction
LED Device.

of an Unencapsulated,

Substrate

In many LED lamp applications, it is necessary to bend the

leads in order to mount the lamp at some angle other than
90 to the surface of a PC board. The leads of a lamp may
be easily bent without introducing mechanical stresses
inside the plastic package. The proper procedure for
bending leads is illustrated in Figure 9.2-1. Bend the leads
prior to soldering. Firmly grip the leads at the base of the
lamp package with a pair of needle nose pliers. The pliers
form a mechanical ground to absorb the stresses when
bending. Bend the leads, one at a time, to the angle desired.

Since the price of gold has increased several times during

the past few years, the cost of a gold plated lead frame has
increased substantially, necessitating the search for an
alternative. The impact of this cost increase has been
industry wide. Many plating material alternatives were
examined, and silver plating offered most of the desired
properties of gold, while remaining price competitive when
compared to other materials.

Since silver is porous with respect to oxygen, the copper

strike acts as an oxygen barrier for the lead frame base
metal. Thus, oxide compounds of the base metal are
prevented from forming underneath the silver plating.
Copper is miscible and readily diffuses into silver to form a
solution that has a low eutectic point. This inter-diffusion
between the copper strike and the silver overplate improves
the solderability of the overall platillg system. If basic
soldering time and temperature limits are not exceeded, a
lead frame base metal-copper-silver-solder
metallurgical
system will be obtained.

Silver resists attack by most dry and moist atmospheres,

such as carbon monoxide or high temperature steam.
Halogen gases do attack silver, however, once the initial
mm layer is formed the process does not continue.

By US'lg silver plating, no additional manufacturing proce's

steps
are required.
Silver has excellent electricll
condudivity. LED die attach and wire bonding to a silwr
lead fume is accomplished with the same reliability as wi1h
a gold lead frame. Also, soldering to a silver lead frame
provides a reliable electrical and mechanical solder joint.
Soldenng silver plated lead frame LED devices into a
printec circuit board is no more complicated than solderirg
LED d ~vices with gold plated lead frames.

The silver plating process is performed as follows: the lead

frame Jase metal is cleaned and then plated with a coppm
strike, nominally 50 microinches (0.00127 mm) thiclc
Then a 150 microinch (0.00381 mm) thick plating of silvm
is added. A "brightener" is usually added to the silvtlf
platillg bath to insure an optimum surface texture to the
silver lating. The term "brightener"
comes from the
medium bright surface reflectance of the silver plate. FigUJe
9.3.1-1 illustrates the metallographic cross-section of the
silver plating system as it would appear with a 1200)(
magnifcation.

The best technique for insuring good solderability of a

silver plated lead frame device is to prevent the formation
of tarnish. This is easily accomplished by preventing the
leads from being exposed to sulfur and sulfur compounds.
The two primary sources of sulfur are free air and most
paper products, such as paper sacks and cardboard
containers. The best defense against the formation of
tarnish is to keep silver lead frame devices in protective
packaging until just prior to the solderillg operation. One
way to accomplish this is to store the LED devices
unwrapped in their original packaging. For example,
Hewlett-Packard ships its seven segment display products in
plastic tubes which are sealed air tight in polyethylene. It is
best to leave the polyethylene intact during storage and
open just prior to soldering.
listed below are a few suggestions for storing silver lead
frame devices.

150 "inch
MINIMUM
SilVER
PLATING

Figure 9.31

Silver reacts chemically with sulfur to form tarnish, silver

sulfide (Ag2S). The build-up of tarnish is the primary
reason for poor solderability. However, the density of the
tarnish and the killd of solder flux used actually determine
the solderability. As the density of the tarnish increases, the
more active the flux must be to penetrate and remove the
tarnish layer.

1.

Metallographic Cross-Section Through a Silver Plated

Lead Frame, 1200X Magnification.

Store the devices in the original wrapping unopened

until just prior to soldering.
If only a portion of the devices from a single tube
are to be used, tightly re-wrap the plastic tube
containing the unused devices in the original or a
new polyethylene sheet to keep out free air.

Loose devices may be stored in zip-lock or tightly

sealed polyethylene bags.
For long term storage of parts, place one or two
petroleum naphthalene mothballs inside the plastic
package containing the devices. The evaporating
naphthalene creates a vapor pressure inside the
plastic package which keeps out free air.
Any silver lead frame device may be wrapped in
"Silver Saver" paper for positive protection against
the formation
of tarnish. "Silver Saver" is
manufactured by:
The Orchard Corporation
1154 Reco Avenue
St. Louis, Missouri 63126
314/822-3880
To reduce shelf storage time, it is worthwhile to use
inventory control to insure that the devices fust
received will be the first devices to be used.
One caution: The adhesives used on pressure
sensitive tapes such as cellophane, electrical and
masking tape can soak through silver protecting
papers and may leave an adhesive fIlm on the leads.
This fIlm reduces solderability and should be
removed with Freon T-P35, Freon T-E35, or
equivalent, prior to soldering.
Pe troleum naphthalene in the form of mothballs or
chips may be obtained from the Frank Curran
Company, 8101 South Lemont Road, Downers
Grove, illinois 60515, 312/969-2200.

Optoelectronic devices are usually soldered into a printed

circuit board along with other components. It is necessary
to achieve reliable soldering in order to obtain reliable
circuit operation. A basic understanding of the types of
solder, fluxing agents and surface conditioners that are
typically used in the electronic industry will aid the
process engineer in establishing a reliable soldering process.

The solder most widely used for wave soldering electronic

components into printed circuit boards is Sn60 (60% tin
and 40% lead) per Federal Standard QQ-S-57l. Two
alternatives are the eutectic composition Sn63 and the
solder SN62 which contains 2% silver. Table 9.4.1-1 lists
the composition, maximum acceptable contaminant levels
and temperatures for these solders.

As the device leads pass through the solder wave of a flow

solder process, the tin in the solder scavenges silver from
the silver plating and forms one of two silver-tin
intermetallics (Ag6Sn or Ag3Sn). This silver in the molten
solder should not be considered a contaminant. As the
silver content increases, the rate of scavenging decreases and
the probability
of obtaining
the desired
base
metal-copper-silver-solder metallurgical system is improved.
The result is that the silver content in solder, which reaches
a maximum of 2~% in Sn60 at 230C, aids in producing
reliable solder joints on silver plated lead frames. Periodic
replenishing of the solder as it becomes contaminated, with
fresh solder, helps assure reliable soldering on a continual
basis.
For hand solder operations, a high quality resin core wire
solder is recommended. The solder may be obtained in wire
diameters from .254mm (.010 inch) to 3.l75mm (.125
inch) as common sizes. The core is typically an RMA flux
in the amount of 2.2% or 3.3% by weight.

Solder flux classification per Federal Standard QQ-S-57l,

listed in order of increasing strength, are as follows:
Type
Type
Type
Type

R: Non-Activated Rosin Flux

RMA: Mildly Activated Rosin Flux
RA: Activated Rosin Flux
AC: Organic Acid Flux, Water Soluable

The Type R flux is a pure water white (WW) rosin without

any additives. Rosin is a complex natural product obtained
from the gum of live trees. It is mixed with a suitable
solvent to form a homogeneous solution. Fluxes are graded
as to the percentage by weight of rosin solids in the
solution. The flux and its residue are non-corrosive and
non-conductive.
The RMA flux is a homogeneous mixture of WW rosin in a
blended alcohol vehicle into which a small amount of
activating agent has been added. The flux and its residue are
non-corrosive and non-conductive.
The RA flux is the same as RMA flux except that a greater
amount of activating agents have been added. The flux and
its residue
are non-corrosive,
and the residue is
non-conductive only if all the solvent is volatilized and the
residue remains dry.
An AC flux is considered full active and has a greater
fluxing ability than a rosin flux. An AC flux may contain
acids, organic or inorganic chlorides and therefore is
corrosive. Due to their organic nature, the AC flux residues

BisComposition

Sn60

SN62

Approximate
melting range

max

Solidus

Liquidus

Tin

Lead

Silver

Antimony

muth
max

Copper,
max

Iron,
max

Zinc,
max

num,
max

nie,
max

59.5
to
61.5

Remainder

----

0.20
to
0.50

0.25

0.08

0.02

0.005

0.005

0.03

----

0.08

183

191

61.5

Remainder

1.75
to
2.25

0.20
to
0.50

0.25

0.08

0.02

O.OO!'>

o om,

Q.!!3

----

c.ce

~...,n

~.
n

Remainder

----

0.20
to
0.50

0.25

0.08

0.02

0.005

0.005

0.03

---.

0.08

183

62.5
Sn63

62.5
to
63.5

Arse-

Total
of all
others,

Cadmium,
max

"to

Alumi-

01

,J

I::J

183

decompc se at soldering temperatures, thereby eliminating ~

large ammnt of the corrosive residue. All residues must b
removed to prevent the formation of conductive ionic path
and corrosion caused by residual cWoride salts.

optoelectronic
devices. Complete immersion of an
optoelectronic device into a surface conditioner solution is
NOT recommended. For best results, immerse only the
tarnished leads.

Suggested applications of these flux types with respect t

various t.lrnish levels are as follows:
Silver pllted lead frames that are clean, contaminant an
tarnish free may be soldered using a Type R flux such a
Alpha lCO.
Minor
Tarnish: Since some minor tarnish or other
contaminant may be present on the leads, a type RMA flu
such as Plpha 611 or 611 Foam, Kester 197 or equivalent i
recommf nded. Minor tarnish may be identified by reduce
reflectance of the ordinarily medium bright surface of th
silver plcting. Type RMA fluxes which meet MIL-F-1425
are used in the construction of telephone communication
military md aerospace equipment.
Mild Tarnish: For a mild tarnish, a type RA flux such a
Alpha 7]1-35, Alpha 809 foam, Kester 1544, Kester 158
or equh alent should be used. A mild tarnish may b
identifiell by a light yellow tint to the surface of the silve
plating.
Moderatt Tarnish: A type AC water soluable flux such a
Alpha 8::0, Alpha 842, Kester 1429 or 1429 foam, Lonc
3355 or equivalent will give acceptable results on surface
conditiollS up to a moderate tarnish. A moderate tarnish
may be identified by a light yellow-tan color on the surface
of the siller plating.

If a mor ~ severe tarnish is present, such as a heavy tarnish

identified by a dark tan to black color, a cleaner/surface
conditioner must be used. Some possible cleaner/surface
conditlOr ers are Alpha 140, Alpha 174, Kester 5560 and
Lonco TLrl. The immersion time for each cleaner/surface
conditior er will be just a few seconds and each is used at
room temperature. For example, Alpha 140 will remove
severe
tarnish
almost upon contact; therefore, the
immersi:m
time need not exceed 2 seconds. These
de an e r I su rf ace conditioners
are acidic formulations.
Therefon:, immediately, thorougWy wash all devices which
have been cleaned with a cleaner/surface conditioner in
cold wat ~r. A hot water wash will cause undue etching 0
the surface of the silver plating. A post rinse in deionized
water is advisable.
CAUTIO "I: These cleaner/surface conditioners may etch
exposed glass surfaces and may have a detrimental effect
upon
t Ie glass filled encapsulating epoxies used in

Alpha Metals, Inc.

56 G Water Street
Jersey City, New Jersey 07304
302/434-6778
Kester Solder Company
4201 G Wrightwood Avenue
Chicago, Illinois 60639
312/235-1600
London Chemical Company (Lon co)
240 G Foster
Bensenville, Illinois 60106
312/287-9477

Before the actual soldering begins, the printed circuit

boards and components to be soldered should be free of
dirt, oil, grease, finger prints and other contaminants.
Fluorinated cleaners such as Genesolve DI-15 or DE-IS may
be used to preclean both the printed circuit boards and
LED devices. Operators may wear cotton gloves to prevent
fmger prints when loading components into the printed
circuit boards.
If the silver lead frames have acquired an unacceptable layer
of tarnish, remove this tarnish layer with a cleaner/surface
conditioner just prior to soldering. Since a cleaner/surface
conditioner does slightly etch the surface of the silver
plating, the silver leads are now more susceptible to tarnish
formation. Therefore, use a cleaner/surface conditioner
only on those silver lead frame devices which will be
soldered within a four hour time period.

The temperature of a solder wave should be at least 38C

(lOOF) above the melting temperature (solidus) of the
solder. For Sn60 solder, this 221C (430F) minimum.
Most wave solder operations maintain a solder wave
temperature between noc (446F) and 260C (500F).
At 230C, Sn60 solder dissolves silver at the rate of 60
microinches per second, and at 260C it dissolves silver at
the rate of 80 microinches per second. Therefore, with an
initial silver plating thickness of 150 micro inches, a dwell
time of less than 2 seconds in a 230C solder wave will
provide the desired lead base metal-copper-silver-solder

P.C. BOARD
METALLIZATION

10to 20
IJinches of SILVER

SOLDER HAS WETTED LEAD AND

PLATED THROUGH HOLE AND HAS
FLOWED TO TOP OF P.C. BOARD.
EXPOSED LEAD SHOWS 95% SOLDER
COVERAGE, INDICATING PROPER
WETTING.

LEAD
BASE

METAL

Figure 9.!i.1-1

The Desired Metallurgical System After a Solder

Operation.

metallu gical system, as illustrated in Figure 9.5.1-1.

Tho
copper strike is intact. At 4 seconds dwell time in a 230C
solder wave, all of the silver has been dissolved and thn
solder I'1Ust adhere to the copper strike; at 5 seconds, it i:;
possible to have completely dissolved the copper and now
the sol<er must adhere to the lead base metal. For plastin
LED devices, the 230C solder temperature is preferable to
260C.

Contrarr to popular belief, hand soldering can be mon

injurioU> to an LED device than wave soldering, the reasor
being that it is difficult to control dwell time and
temperature. To effectively hand solder, it is best that the
operato~' realize that "the hotter and longer the better" i~
not true. It is best that a 15 watt iron (25 watt maximum)
be used. Maintain the iron in contact with the lead for only
that time which is sufficient to allow the solder to flow
evenly around the joint. Heat sinking the leads may be
worthw:Jile, though this should not be necessary if
sufficient care is obseI'Ved. Always keep the tip of the iron
clean and well tinned.

The best rule to follow is to cut the leads of an LED device

after soldering, not before. This accomplishes two things
First, the additional lead length provides heat sinkin~
during the soldering operation. Secondly, the soldered join1
provid's
a mechanical
ground which prevents the
mechalLcal stresses due to cutting from being transmitted
into the device package. If it is necessary to cut the lead~
prior to soldering, support the device by the leads with 2
pair of long nose pliers during the cutting operation as i~
illustrated in Figure 9.2-1.

Printed circuit boards, either single sided, double sided or

multi-Ia~rer, may be manufactured with plated through
holes with a metal trace pad surrounding the hole on both
sides of the printed circuit board. The plated through hole
is'desira Ie to provide a sufficient surface for the solder to
wet, all( thereby be pulled up by capillary attraction along

Figure 9.5.4-1

LACK OF SOLDER COATING

ON LEAD INDICATES INSUFFICIENT WETTING

Comparison Between an Ideally Soldered

and an Undesirably Soldered Lead.

Lead

the lead through the hole to the top of the printed circuit
board. This provides the best possible solder connection
between the printed' circuit board and the leads of the LED
device. Figure 9.5.4-IA illustrates an ideally soldered lead.
The amount of solder which has flowed to the top of the
printed circuit board is not critical. A sound electrical and
mechanical joint is formed. Figure 9.5.4-1B illustrates a
soldered lead which is undesirable.

Unlike most semiconductor devices, LED devices for visual

applications are limited in the kinds of materials that may
be used for encapsulation or hermetic sealing. In the case of
encapsulating materials, the optical properties required
limit the selection to only a few highly specialized epoxies.
Exposure to a solvent used in a cleaning operation must not
alter these optical properties in any way. For this and other
reasons, only certain cleaning solvents may be used to post
solder clean LED devices.

It is important to remove both the rosin and ionic residues

after soldering to insure reliable operation of the complete
circuit. This is best accomplished by using an azeotrope of
fluorocarbon and alcohol. The fluorocarbon is used to
dissolve the residual rosin and the ionic contaminants are
removed by the alcohol.
The type of fluorocarbon is very important. Tests have
demonstrated that the only fluorocarbon that is compatible

CLEANII\lG AGENT

COMPOSITION

CLEA NING OPERATION

60C 140F) Wash to remove an AC flux residue.

WATER
ETHA OL,
ISOPROPANOL

Alcohol

Gener al cleaning after hand soldering.

FREO TF
GENESOLV 0
ARKLDrJE P

100%
Fluorocarbon
(F113)

Gener al cleaning agent for removing grease, oils, etc.

FREO TE
ARKLOrJE A

"'='4%Ethanol

Improved general cleaning agent.

GENESOLV DE-15
01-15
BLACO-' RON DE-15
01-15

"'='15% Ethanol
"'='15%Isopropanol
"'='15%Ethanol
",=,15%Isopropanol

Vapo cleaning at boiling - up to 2 minutes in the vapors - best for

post s plder cleaning.

ARKLmlE

"'='25%Isopropanol

Vapor cleaning at boiling - up to 2 minutes in vapors. Best for post

solder cleaning.

"'='35%Ethanol
"'='35%Isopropanol

Room temperature

FREON -roE 35
T-P 35

post solder cleaning.

TABLE 9.6.1-1 Table of Suggested Post Solder CI aning Agents that are Compatible with LED Devices.
with pla:;tic LED devices is trichloro-tri-fluoroethane
(FIB),
sold under tradenames as Freon, Genesolv D and
Arklone. Some suggested cleaning products are listed in
Table 9.6.1-1.
Cleaning solvent mixtures based on the fluorocarbon
tetrachlow-di-fluoroethane
(F 112) are not recommended
for cleaning plastic LED parts. Also, such cleaning agents
from the keystone family (acetone, methyl ethyl ketone,
etc.) and from the chlorinated hydrocarbon
family
(methylen,~ chloride,
trichloroethylene,
carbon tetrachloride, etc.) are not recommended for cleaning LED
parts. All of these various solvents attack or dissolve the
encapsulaHng epoxies used to form the packages of plastic
LED devices.
9.6.2.

Bulk Cleaning Processes

Post cleaning of soldered assemblies when a type RMA or

Type RA flux has been used may be accomplished via a
vapor cleaning process in a degreasing tank, using an
azeotrope of fluorocarbon and 15% to 25% alcohol as the
cleaning agent. A recommended method is a 15 second
suspension in vapors, a 15 to 30 second spray wash in liquid
cleaner, a:ld fmally a one minute suspension in the vapors.
When a water soluable Type AC flux such as Alpha 830 or
Kester lL-29/1429F is used, the following post cleaning
process
is suggested:
thoroughly wash with water,
neutralize using Alpha 2441 or Kester 5761 foaming, then
thoroughly wash with water and air dry.

Allied Chemical Corp. (Genesolvf

Speciality Chemicals Division
P.O. Box 1087R
Morristown, New Jefsey 07960
201/455-5083
E.!. DuPonte de Nemours & Company
Freon Products Division
Wilmington, Delaware 19898
302/774-8341
Baron-Blakeslee (Blaco-Tron)
1620 S. Laramie Avenue
Chicago, lllinois 60650
312/656-7300
Imperial Chemical Industries, Ltd. (Arklone)
Imperial Chemical House, Millbank
London SW1P3JF, England
9.6.3

Special Cleaning
Board Displays

Instructions

for Monolithic

PC

The monolithic printed circuit board display devices use a

lens made from a special plastic chosen for its superior
optical properties. This plastic does not lend itself to vapor
cleaning processes. Also, the lens does not environmentally

protec the LED chips. Therefore,

cleanir g procedures are suggested.

the following specil1

For cleaning after a solder operation, the following proce,s

is reconunended: wash display in clean liquid Freon T-P 35
or Freon T-E 35 solvent for a time period up to 2 minuks
maximum. Air dry for a sufficient length of time to allo.,,,
solven1 to evaporate from beneath display lens. Maintain
solve t temperature
below 30C (86F). Methanol,
isopropanol, or ethanol may be used for hand cleaning at
room 1emperature. Water may be used for hand cleaning If
it is not permitted to collect under display lens.
Solven vapor cleaning at elevated temperatures is nc t
recommended as such processes will damage display len:,.
Ketones, esters, aromatic and chlorinated hydrocarbon
solvents will also damage display lens. Alcohol base active
rosin flux mixtures should be prevented from coming i
contac with display lens.

LED devices may be socket mounted in the same fashion as

other ~emiconductor devices. The selection of a suitabl~
socket is generally based on a quality vs. cost trade-off,
with cost usually the primary factor.
When s~lecting a socket to mount silver lead frame device~,
it is the performance of the socket that becomes th~
import,mt factor. The socket contacts must have sufficient
force <Uldcontact area to form an air tight mechanical seal
at the ead/contact interface to prevent the formation of
tarnish at this interface. Tarnish (silver sulfide) is a
semiconductor,
thus would introduce an undesirable
resistance if allowed to form in the contact area. Figum
9.7-1 i ustrates the desired condition. A good quality

Figure 9.7.1-1

SILVER
PLATED
LEAD
SOCKET
SIDEBEARING
CONTACT
(CONTACT
ISMADE
ON FLAT
SIDE OF
LEADI

Figure 9.7-1

Connector Contacts That Have Sufficient Insertion

Force and Contact Area to Form an Air Tight
Mechanical Seal with the Silver Plated Lead to Prevent
Tarnish Build Up in the Contact Area.

socket will require considerable insertion force and provide

good wiping action upon insertion to scrape away any
initial tarnish. The socket should be capable of securely
holding the device without relative movement of the device
lead with respect to the socket contact in the presence of
mechanical shock or vibration. Sockets with side-bearing
contacts that grip the flat sides of a lead are preferred for
the mounting of DIP LED devices that have silver plated
leads. Sockets that have edge-bearing contacts may be used
to mount devices that have square silver plated leads, such
as plastic LED lamps, providing the insertion force is
sufficient to provide the required air tight mechanical seal
along the edge of the lead.

Optional SJcket Mounting Configurations

HP 5082-7300 Series of OBIC Displays.

RT ANGLE SOCKET
(AUGAT 508-AG8D)

for the

PINS

APPLICATIONS

508-AG8D
514-AG21D
514-AG25D
514-AG26D
314-AG5D-; R
324-AG6D
336-AG6D
340-AG6D
314-AG39D Low Profile
324-AG39D Low Profile

8
14

(1) HP 5082-7300; RT Angle Mtg

90
45 (1) Stretched 7-Segment
60

14
24
36
40
14
24

(1) Stretched 7-Segment

(3)

340-AG39D Low Profile

325-AG1D

40
25

BURNDY
Richards Avenue
Norwalk. Conn. 06856
203/838-4444

HBRB2S-1
DILB14P-1
DILB24P-1
DILB40P-1

8
8
8

2
14

CAMBION
445 Concord Avenue
Cambridge. Mass. 02138
617/4915400

703-3777-0
703-5151-0
703-5153-0

-12
-04-16
-04-16

CI RCUIT ASSEMBLY
3169 Redhill Avenue
Costa Mesa, Calif. 92626
714/540-5490

MANUFACTURER

PART NUM

AUGAT
33 Perry Avenue
Attleboro. Mass. 02703
617/222-2203

ER

(4) HP 5082-7300
(5)
(1) Stretched 7-Segment
(3) HP 5082-7300
(5)
Strip Socket
(1) LED Lamp;T-1

orT-13/4

24
40

(1) Stretched 7-Segment

(3) HP 5082-7300
(5) HP 5082-7300

14

(1) Stretched 7-Segment

24

(3) HP 5082-7300

CA-14S-10SD
CA-24S-10SD
CA-40S-10SD
CA-25STL- OSD
CA-S36 SP1 0

14
24
40
25
36

(1) Stretched 7-Segment

Strip Socket
Stripline Plug (for pc board Monolithic

ROBINSON-NUGENT

IC-143-S1

800 E. 8th Street

New Albany, Indiana 47150
812/945-0211

IC-163-S1
IC-246-S1
IC-406-S2
ICN-143-S3 Low Profile
ICN-406-S1
SB-25

14
16

(1) Stretched 7-Segment

(2)

24
40
14
40
25

(3) HP 5082-7300 Displays

(5)
(1) Stretched 7-Segment
(1) HP 5082-7300 Display
Strip Socket

14

( 1) Stretched 7-Segment

24
40
12
24
36
3 to
22

(3) HP 5082-7300
(5)
(1)

JERMYN
712 Montgomery Street
San Francisco. CA 94111
415/362-7431
AMP
449 Eisenhower Blvd.
Harrisburg, Penn. 17105
717/564-0100

A1237
A1252AM I ow Profile
A23-2023 L w Profile
A23-2030
583640 Ser s
Low Profile
583773 Sen s

LED display devices may be mounted in DIP sockets rather

than being directly soldered into a PC board. Stretched
seven segment displays may be mounted in standard 14 pin
dip sockets. OBIC displays may require the use of strip
sockets, LSI sockets or special sockets as illustrated in
Figure 9.7.1-1. A small alphanumeric OBIC display may be
mounted in standard DIP sockets that have been machined
down in length to accept the 12 pin devices so as to form
an end stacked display string.
Front panel mounting kits are available for LED displays.
Rochester Digital Displays, Inc., 120 North Main Street,
Fairport, New York 14450,716/223-6855,
offer two such
kits. One kit accepts the lIP 5082-7300 series display and
the other accepts stretched seven segment displays.
PC board monolithic display products may be mounted
using anyone
of several techniques. The most straight
forward is the use of standard PC board edge connectors. A

(3) HP 5082-7300
(5)
displays)

(2) HP HDSP-2000
(3)
Strip Connectors

more cost effective approach can be implemented through

the use of standard (or custom) stamped or etched metal
mounting clips such as those available from Burndy,
Richards
Avenue,
Norwalk,
Connecticut
06850,
203/838-4444, (series LED-B) or l.A.V. Manufacturing,
Inc., 20 Lucon Drive, Deer Park, New York 11729, (series
1255). Circuit Assembly Corporation,
3169 Red Hill
Avenue, Costa Mesa, California 92626, 714/540-5490 and
Burndy each manufacture low cost connectors especially
designed for mounting the display board at a given angle as
illustrated in Figure 9.7.1-2. A third approach would be to
use a series of etched clips which are first soldered to the PC
mother board. The display board is then pressed into place,
with each clip being inserted into one of the plated through
holes at the edge of display board.
Front panel mounting hardware for use with T-l 3/4 plastic
LED lamps is manufactured by the Eldema Division of the
Genisco Technology Corporation,
18435 Susana Road,
Compton, California 90221, 213/537-4750. Two of their
devices are illustrated in Figure 9.7.1-3.

MOUNTING LUGS
ACCEPT No.8 OR
No. 10 FH SCREWS
CONNECTOR
SECURELY HOLD:;
DISPLAY IN PLACE

junction temperature is 110C. A hermetic device will

tolerate a much higher IC junction temperature, but the
maximum LED junction temperature should not exceed
125C. Reliable operation is obtained when the junction
temperatures are maintained below these maximum values.
o

The thermal resistance to ambient of 3S C/W for the

display mounting structure assumes that the mounting
surface for the display is an isothermal plane. If only one
aBIC display is operated on this isothermal plane at a
power level of 1.7 watts maximum, as an example, the
o
temperature rise above ambient would be 42.S C:

CONNECTOR PINS
PROTF UDE DOWN FOR
SOLDERING INTO A
PRINT:D CIRCUIT BOARD

Figure ~.7.1-2

Mounting Connectors for Monolithic

Board Display Devices.

LED PC

Most ':-ED devices may be operated in elevated ambient

tempe -atures without heat sinking by utilizing input powllr
deratir g. Devices which usually fall into this category are
those devices that do not contain on-board integrated
circuit;. Heat sinking may be required for those on-board
integr~ted circuit (aBle) devices that have considerable
power dissipation occuring within the integrated circuit.

If a second display is placed on this same thermal plane,

with no increase in thermal dissipation capability, the
temperature
rise would double to 85C, reaching
catastrophic levels very quickly. Hence, for each aBIC
device added to the display string, there must be an
appropriate increment of heat dissipating capability added
to the mountillg plane. For short display strings (4 to 6
o
aBIC devices), the 3S C/W per device may be achieved by
utilzing a printed circuit board design which maximizes the
amount of metal surface area remaining on the board. For
longer display strings, or if sufficient metal surface area
cannot be achieved, an external heat sink must be
o
employed to obtain the 3S C/W per device case-to-ambient
thermal resistance.
In practice, heat sink design involves the optimization of
techniques used to dissipate heat through the device leads.
The heat transfer is from the device leads to the PC board
metallization to the heat sink and is dissipated in the
surrounding ambient. A heat sink of approximately 52
square centimeters (8 square inches) per aBIC device will

The specific devices which may require heat sinking are the
aBIC LED numeric and alphanumeric displays. An aBle
alphanJmeric display, for example, combines a significar t
amount of integrated circuit logic and LED display
capability in a very small package. As such, on board POW{ r
dissipa -ion is relatively high and the thermal design of the
mount ng assembly becomes an important consideration.
The di play is designed to operate over a wide temperature
range with full power dissipation. This capability may b
utilizec only if the mounting assembly has a thermcl
o
resistarce to ambient of 3S C/W per device, or less.
The pr mary thermal path for power dissipation is through
the de'ice leads. The thermal resistance junction-to-Iead is
o
typicaJ..:y 15 to 2S C/W, depending upon the packag~
configt ration. For a plastic device, the maximum LED

I
Figure 9.7.1-3

Front Panel Mounting Hardware for T-1 3/4 LED

Lamps.

METAL FRONT PANEL,

EQUIPMENT CHASSIS

BEZEL AND CONTRAST

FILTER ASSEMBLY

THERMAL CONDUCTIVE SILICONE

RUBBER SHEET
(ELECTRICALLY
INSULATING)
HEAT SINK WITH FINS;
PRIMARY HEAT DISSIPATION
PROVIDED BY CONVECTION
IN CONTINUOUS FREE FLOW
OF AIR THROUGH FINS

,
~

'"
o

"

'~

OBIC LED
DISPLAY

HEAT SINK
WITHOUT FINS;
PRIMARY HEAT
DISSIPATION
PROVIDED BY
CONDUCTION TO
EQUIPMENT METAL
FROiH PAi~EL Ai~D
CHASSIS WHICH
ACT AS THE
THERMAL RADIATOR.
MOUNTING FLANGES
PROVIDE MAXIMUM
SURFACE CONTACT
WITH FRONT PANEL. USE
THERMAL COMPOUND TO
ACHIEVE MINIMUM THERMAL
RESISTANCE HEAT SINKTO-FRONT PANEL.

OUTLINED CIRCUIT
CONDUCTOR TRACE

TWO HEAT
SINK DESIG
ALTERNATIVE

PC BOARD r~ADE WITH r1AXIt1Url

SURFACE METALIZATIOr~ AND
WIDE COrmUCTIVE TRACES. t~ETALIZATION ETCHED AWAY TO
OUTLINE COi~DUCTOR TRACES AND PADS. PLATED THROUGH HOLES.
Figure 9.8.1-1

Heat Sink and Pa el Mounting Configuration

an OBIC Alphan merie LED Display.

typically permit operation at a power dissipation

watt/device in an ambient temperature of 70C.

of 1

As an alternative to soldering, OBIC displays may be

mounte in DIP or stripline sockets which are themselves
soldered into a PC board with maximum surface area
metallization.
These sockets will allow enough space
between the PC board and the OBIC displays to permit a
heat sink bar to be inserted to conduct heat to an external
sink. These sockets add a thermal resistance of about
2C/W between the device leads and the PC board.

Heat sinking
an OBIC display is most effectively
accomplished
by constructing a direct thermal path
between the device leads and the heat sink. A direct
thermal path may be achieved by implementing the design
concept illustrated in Figure 9.8.11. The printed circuit
board is designed with a maximum amount of metallization
remaining on the board. Thin lines are etched in the

for

metallization to outline the circuit conductors. The PC

board metallization is used as part of the thermal path
between the device leads and the heat sink.
The heat sink is designed to fit flush against the PC board,
with grooves cut into it to provide clearance for the device
leads. A sheet of thin thermally conductive silicone rubber
is inserted between the heat sink and the printed circuit
board. This silicone rubber completes the thermal path and
at the same time, electrically insulates the heat sink from
the PC board metallization.
Heat transfer from the heat sink to the ambient air may be
accomplished in one of two ways. One method is to
mechanically fasten the heat sink to the metal chassis or
front panel. This extends the display heat sink by utilizing
the complete chassis as a thermal radiator. If this is not
possible, then fillS may be made as part of the display heat
sink to increase its surface area. Heat is now transferred to
the local ambient air and carried off by convection. This
process is assisted by insuring a continuous free flow of air
passing through the fillS of the heat sink.

Thermal conductive silicone rubber is sold under the trade

name "Sil-Pad", by the Bergquist Company, Inc., 4350
West 78th Street, Minneapolis, Minnesota 55435. The shtlet
is a nominal 0.305mm (.012 inch) thick, has a thermal
resistmce of approximately .08C/W per square centimeter
and a Shore A hardness of 76. Anticipated useful life at a
temperature of 140C is 300 thousand hours, in a n.m
condensing atmosphere.
The heat sink may be made from a length of aluminum
extrusion or purchased from anyone of a number of heat
sink manufacturers. In either case, the heat sink will be a
custom part and should be designed so as to (1) have
maximum surface area contact with the PC board, (2) have
as mu ch surface contact area with the supporting chassis or
(3) have the maximum possible surface area if the lo( al
ambie ot air must carry off the heat by convection.

An eI lerging technology for the efficiency transfer of heat

from me point to another is the heat pipe. When it is n:>t
possible to incorporate an on-board heat sink, these devices
are a cost effective means of efficiently transferring the
heat from the display PC board to a remote heat sink or~o
the equipment chassis. The heat pipe by itself does n,)t
functiOn as the heat sink. The input to output thermal
resista:3.ce of a 305mm (12 inch) long heat pipe is typical y
.1OC/W when transporting 10 watts. The contact thermal
resistance at either end of the heat pipe can be maintaim d
to ab>Ut .062C/W per square centimeter (AoC/W p'~r
square inch).

The foUowing list of manufacturers are only suggestions, ,s

those not listed may well be equally qualified:

Aham
968 W. Foothill Blvd.
Azusa, California 91702
Aham
2 Gill Street, Bldg. 5
Woburn, Massachusetts 01810
International Electronic Reserach Corp. (IERC)
135 W. Magnolia Blvd.
Burbank, California 91502
Industrial Heat Sink Corporation
5338 Alhambra Avenue
Los Angeles, California 90032

Noren Products, Inc.

846 Blandford Blvd.
Redwood City, California 94062
Tecknit
129 Dermody Street
Cranford, New Jersey 07016
Jermyn
712 Montgomery Street
San Francisco, California 94111

Active di1ferential driver, 3.49-3.51

Alphanu~,eric display, 5.43
Ambient light, ranges for, 6.4
Amplifier, photocurrent
.linear, 4.8-4.9
.Iogari":hmic,4.8
Analog amplifier, 3.4
Analog ci 'cuits, optoisolators
.ac COt pled, 3.35
.A/D converts, 3.36
.diffemntiallinearizer,3.34
.pulse /Vidthmodulation, 3.37
.servo inearizer, 3.32
.voltage-frequency conversion, 3.39
Analog/digital converter, 3.36
Angle, half intensity point, 2.11
Anti-lock loop, 3.64-3.65-3.66
Anti-parallel LED, 3.40
Anti-refle:tion surface, 6.16
Apparent emitting area, 7.5
Array, bar graph, 2.30-2.32-2.35
Arrays, la IIp, 2.24
Array, lamp spacing, 2.27
Array, x-., addressable, 2.24-2.28
Array, 5x"7,5.43-5.44
Array, 8-lamp position indicator, 2.35
Array, 64 LED lamp, 2.35
ASA film speed, 2.55
band gap, 1.2
Bandwidth, for noise, 4.5
Bar graph displays, 2.30
.op amll decoder, 2.33
.digital decoder, 2.36-2.37
Bar code pickup, 2.46-2.48
Base bypa IS resistor, 3.2-3.3-3.49-3.52
Beam splitter, 4.9-4.10
Boltzmann's constant, 4.5
Breakdowil voltage of PIN photodiode, 4.1
Busing, isolated, 3.62-3.65
Bypass capacitors, optoisolators, 3.17
Candela, 7.2-7.3
Character Ilarallel, 5.4
Character :;erial, 5.4
Chatter sUllpression, 3.5-3.17-3.30
Chip, LED monolithic, 5.35
CIE chromaticity diagram, 6.5
CIE iIIu iant C, 6.2
CIE photoJic curve, 6.2-6.3-7.1
CMOS interface, 5.40-5.41
CMRR, CMR, 3.5-3.9

coaxial, mirrors, 2.51-2.52

Color matching of LEOs, 2.10-2.11
Color purity (6.2-6.3)
Common-collector, 3.11
Common mode voltage, 3.6-3.49-3.53
Compatibility, 6 and 8 pin isolators, 3.15
Confocal optics, 2.46-2.47
Constant current drive, 2.19-2.20
Contrast, ratio
.in backlighting, 2.38
.enhancement, 6.1
Corona, 3.9
Critical angle, efficiency, 2.2
CTR,3.20-3.14
CTR degradation
.model, 3.20
symptoms, 3.20
worst-case design, 3.24
Current limiting, graphical solution, 2.19
Current looping, 3.66-3.67
Current transfer ratio, 3.10-3.14
Czochralski, 1.6
0*, detectivity, 4.6
Dark current, 4.5-4.7-4.8
Darlington amplifier, 3.4
DC derating, stretched segment display, 5.18
Degradation
.light output, 8.1
.rate of, 8.3
Densitometer, 4.9-4.10
Depletion region, 4.1
Detectivity, D*, 4.6
Differential linearizer, 3.34
Diffusance quotient, 2.39
Diffuser, 2.39-2.40-2.43
Digital operation of optoisolator, 3.39
Direct dc driver, 5.21
direct gap, 1.3
Dominant wavelength, Ad, 2.11
.definition, 6.1-6.3
.for LEDs, 6.2-6.5
Doping, 1.1
Drive current vs. ambient light, 5.32-5.33
Duplex, 3.62-3.63
Duty factor, 2.25-2.27
.row strobing, 5.47
Dynamic resistance, 5.16
.LED lamp, 2.12

Electr luminescence, 1.1

EMI, .49
Enable, optoisolator output, 3.5
Energt gap, 1.1
Epita, y, 1.6
Epox f, 8.3-8.4
Equi lent circuit, LEO, 5.16
Exitar ce, 7.2-7.3

Heat pipes, 9.12

Heat sinking
.To pc board, 9.12
.OBIC displays, 9.11
.Manufacturers, 9.12
Hermetic, optoisolators, 3.15
Humidity, effects of, 8.3
Hysteresis, in optical comparator, 2.48-2.49

f/nul1'~er, 2.45-2.46-2.52
Fall ti e of PIN photodiode, 4.3-4.4
Fatig ,thermal, 8.3
Fiber ptics, 2.51-2.54
Fiel stop, 7.9
Film speed, 2.55
Filters
.co-nbination, 6.13
.crosshatch, 6.15-6.16
.louvered, 6.14-6.15
.polarizing, 6.16
.pu pie, 6.13
Filters, glass, 6.4-6.5
Filters, plastic, 6.4
.For Std red LEOs, 6.6-6.7-6.8
.For Hi-Eff red LEOs, 6.6-6.7-6.8
.For yellow LEOs, 6.6-6.7-6.9
.For green LEOs, 6.6-6.7-6.9
Filters, for stretched segment displays, 5.33
Filter products
.Table of, 6.17
.Mal ufacturers, 6.18
Flip-fl P
.sele tive, 3.49-3.53
.ex lusive-OR, 3.49-3.53-3.54
Flux
.defi ition, 7.2-7.3
.measurement, 7.9-7.11
Fluxes, soldering, 9.3-9.5
Font, .2-5.3
Footcal die, 7.4
Footlambert, 7.4
Focal Ie19th, 2.45-2.5
Forwar current, average, 2.21
Forwar current, graphical solution, 2.16
Forward current, maximum, 2.24
Forward voltage, temperature coefficient, 1.8
Forward voltage, 2.18-5.16

Index of refraction, 2.1

Incidance
.definition of, 7.2-7.4
.in backlighting, 2.39-2.42
.use in optical measurements, 7.7-7.8
Incidance response, 4.3
Indirect gap, 1.3
Infrared,4.6
Insulation, 3.1-3.9
Intensity
.definition, 7.2-7.3
.measurement, 7.7-7.8
Intensity control, hexidecimal OBIC display, 5.9-5.10
Interface devices, table of, 5.57-5.58
Iso-electronic center, 1.2
Isolation, 3.5-3.9

GaAsP, 3.1
Gain, b ndwidth, 3.1-3.4
Glass tr nsition temperature Tg, 8.3
Ground loop, 3.14

Junction temperature, LEO lamp, 2.12

Junction temperature, OBIC
.display (plastic), 5.9
.display (hermetic), 9.10
Junction temperature, stretched segment display, 5.15
Lambert, 7.4
Lambertian, 7.10
Lambertian, pattern
.in backlighting, 2.39-2.41
.reflectance, 2.52
Lambertian radiator, 2.3
Lamps
.relative efficiency, 2.16
:forward voltage, 2.18
.resistive current limiting, 2.19
.constant current limiting, 2.19
.logic interface, 2.20
.arrays, 2.24
.mechanical spacing, 2.27
.microprocessor interface, 2.28
.transistor drivers, 2.29
.bar graph displays, 2.29
Lamp, diffused, 2.6
Lamp, hermetic, 2.9-2.10
Lamp, rectangular, 2.8-2.9

Lamp, wi h other components, 2.9-2.10

Lamp, po:;ition indicator displays, 2.30
Lamp, sul miniature, 2.8
Lamp, T-1, 2.8
Lamp, T-1 3/4,2.7
Lamp, T-1 3/4 low profile, 2.8
Lamp, undiffused, 2.6
Layout considerations, optoisolators, 3.17
Leads
.bending, 9.1
.cutting, 9.6
.silver Jlated, 9.1
.sol ered, 9.6
Lead frame, lamp, 2.7
Lead fran e, stretched segment display, 5.14
LED disp ay, types, 5.2
Legend, a'ea, 2.40-2.41-2.43
Lens, aspheric, 2.6
Lens, extl rnal, 5.35
Lens, immersion, 2.5
Lens, ma~nification, 5.35-5.36-5.37
Linearity, of optoisolator, 3.2-3.3
Linearizer, optoisolator
.differt ntial linearizaer, 3.34
.servo Iinearizer, 3.32
Linearity, PIN photodiodes, 4.6-4.8
Line recei/er (termination)
.one-re;istor, 3.40
.two-resistor, 3.40-3.45
.three-resistor, 3.45
.curren t-clamping, 3.45-3.47
.voltagl'-clamping, 3.45-3.48
Loop gair, (with optoelectronic feedback), 4.9
Logic inteface, lamps, 2.20
Luminositv, 7.1
Luminous efficacy, 2.3-2.44-7.2
Luminous flux, 2.5
Luminous intensity, 2.5-5.35-5.36
Luminous intensity, relative, 2.17
Luminous intensity, ratio, 5.20
Luminous intensity, vs. temperature, 5.20
Luminous intensity, time average, 2.14-5.25-5.38-5.40
Lux, 7.3-'7.4
Lux, ambit!nt, 2.39-6.4
Magnificat on, lens, 2.5-2.45
Mechanica spacing, lamps, 2.27
Metallurgi<al system after soldering, 9.6
Meter cane Ie seconds, 2.55
Microprocl ssor interface
.alpham meric HDSP-2000, 5.51-5.56
.analog ':0 digital conversion, 3.38

.Iamp array, 2.28-2.31

.monolithic display, 5.40-5.43
.OBIC nu meric display, 5.9-5.13
.optoisolator: BUS-I/O, 3.68-3.70
.optoisolator: BUS-BUS, 3.73-3.74
.seven-segment displays, 5.26-5.30
Miller effect, 3.2-3.11
Millicandela, 7.6
Mirror optics, 2.45-2.51-2.52
Moire patterns in tachometry, 2.47-2.51
Monolithic display, lead frame, 5.38
Monolithic display, pc board, 5.38
Multiplex, 3.62-3.64-3.67
N,doping, 1.3
NEP, noise equivalent power, 4.6
Neutralization, 3.49-3.50
Non-linear, source, load, 3.41-3.42
Noise formulas, 4.5
NRZ data rate, 3.12, 3.13
Numerical aperture
.of lens optics, 2.44-2.46
.of fiber optics, 2.54
OBIC LED display, 5.5
.construction, 5.6
.on board IC, 5.6-5.7
.timing, 5.8
.truth table, 5.6
.thermal resistance, 5.9
Obscuration alarm, 4.9
On/off-hook detector, 3.67-3.69
Operational amplifiers, driving LED lamps, 2.33
Operational curves, LED lamp, 2.21-2.13
.stretched segment displays, 5.17
Operating life, 2.14-2.15
Optoisolator
.PIN compatability, 3.15
.packaging-hermetic, 3.15
.packaging-plastic DIP, 3.15
.bypass capacitors, 3.17
.pc layout, 3.17
.CTR degradation, 3.20
.analog circuits, 3.30
.servo linearizer, 3.32
.differential linearizer, 3.34
.ac coupled isolated amplifier, 3.35
.digital techniques for analog applications, 3.36
Optical coupling, 3.2
Optical efficiency, 2.3

P-N ju ction, 1.1-1.2-1.6-8.1

Packa es, LED devices, 9.1
Paralh I to serial conversion, 5.5
Peakir g, 3.57-3.59
Peak \ ravelength, 2.11-6.2
Petrol.um napthalene, 9.3
Photo urrent, Ip' 4.3-4.8
Photo iode
.integrated, 3.2-3.4
.P-rJ,4.1
.PIl ,4.1-4.12
Phot, . .4
Photometers, 7.6-7.7
Photo. etric terms, 7.3
Positio indicator displays, 2.30
.op mp decoder, 2.33
.dig tal decoder, 2.35
Post so der cleaning, 9.6-9.7-9.8
.cle. ners, table of, 9.7
.cle. ner manufacturers, 9.7
Power, verage, LED lamp, 2.12
Power ferating, LED lamp, 2.12
Power issipation, average, 2.21
Pseudo random code, 3.13
Propa , tion delay, 3.11-3.12
Pulse w dth, 5.23-5.24
Pulse w dth modulation, 3.37-5.24-5.9
Punch-1 rough voltage, 4.1
Ouiescl nt, 3.41
Ouantu efficiency,
Ouantu m efficiency,
Ouantu efficiency,
Ouantu efficiency,

external, 2.3
internal, 2.3
LEOs, 2.42
photodiodes, 4.2

Radia power per steradian, 2.3

Radiate power, total (a calculation example, 2.4
RadiatiJn pattern, 2.10-2.4-2.5-7.9-7.11
Radiorr etric terms, 7.3
Reflect. nce in signalling, 2.51-2.52
Reflectl d light, off display face, 6.9-6.10-6.11-6.12
Reflecti :>n,transmission line, 3.40-3.42
Reflecti n coefficiency, R, 6.3
Reflecti In loss, 3.2
Refresh eriod, 2.28-5.23
Relative efficiency, 2.16-2.18-5.20-5.38-5.39
Resistivi current limiting, 2.19-5.22-5.25
Resistiv ty
.Isolators, 3.4
.PIN lhotodiode, 4.1
Resistor shunt, 5.22
Respons vity, 4.3

Reverse coupling, 3.14

Ring detector, 3.66-3.67
Rise-time, PIN photodiode, 4.3-4.4
Row strobing, 16 character display, 5.47
RS-232C, RS-422, RS-423, 3.61
RZ data rate, 3.12-3.14
Schottky clamp, 3.5
Scrambler, 5.14
Self test circuit, 7-segment display, 5.30-5.32
Servo-controlled intensity, 4.9
Servo linearizer, 3.32
Shift register, serial-in-parallel-out (SIPO), 5.47
Siemens, UAA 180 and UAA 170,2.34
Signal-to-noise ratio, 4.6
Silver
.in solder, 9.3
.plating, 9.2
.tarnish,9.2-9.5
Silver saver paper, 9.3
Simplex, 3.61
Smoke detectors
.reflection-type, 2.46-2.47
.obscuration-type, 4.9
Snell's law, 1.6-2.2-2.51
Sockets
.contacts, 9.8
.special, 9.8-9.9
.table of, 9.9
Solder, 9.3
.table of, 9.4
Soldering, 8.4
.hand, 9.3-9.6
.supplies, manufacturers, 9.5
.wave, 9.5
Solid angle, 7.1-7.2
Spectral distribution
.daylight, 6.12
.fluorescent, 6.10
.incandescent, 6.10
.LEOs, 6.8-6.9
Spectral integration, 2.44-2.55
Spectral response, 4.1-4.2-4.3-4.5-4.6
Speed of response
.optoisolator, 3.10
.PIN photodiodes, 4.3-4.4
Split-darlington amplifier, 3.4
Spot photometers for sterance, 7.9
Steradian, 7.2
Sterance
.backlighting, 2.38-2.39
.definition, 7.1-7.2-7.3

.measu 'ement, 7.9

Sterance/intensity ratio for LEOs, displays, 7.5
Storage, silver lead devices, 9.2
StretchEld segment display, 5.14
Strobed o!wration, 5.22
Strobing, I:olumn, 2.25-2.30-5.44-5.45
Strobin!), lOW, 2.25-2.30-5.44-5.46
Strobing rate, 5.45
Strobing rate, minimum, 5.9
Substrate, opaque, 1.6-1.7
Substrate, transparent, 1.6-1.7
Subsystem, display, 5.2-5.4
Surface conditioners, 9.5
Tachometly, 2.46-2.51
Temperatllre coefficient, forward voltage, 5.17
Tempercltu re, glass transition, 2.12
Ten diameters rule, 7.7
Thermal re sistance, LED lamp, 2.12-2.13
Thermal R ~sistance, OBle numeric display, 5.9
Theory, L1:0, 1.1
Threshold :pre-bias), 3.57-3.58
Transmission, constant density filter, TA' 6.3
Transmi!;si4In line
.refle(:ta nce, 3.40-3.42
.charalct ~ristic impedance, 3.42
Transmitta lce, relative, T(A), 6.3
Transistor drivers, lamps, 2.29
TTL interfClce, 2.20
Two-point method for intensity measurement, 7.7-7.8
U L reco!)nition, 3.10
Ultra-viole". response, 4.2-4.8-4.6
Voltage-frequency conversion, 3.39
Viewing angle, 7.6
Wavelen!Jtt, emitting, 1.2-1.4
.temperature coefficient of, 1.8
Worst case design
.lamps, ~.21-2.24
.opto~so lators, 3.24