MODULE 4: HARDWARE FEATURES FOR CONTROL OVER EQUIPMENT MALFUNCTIONS

A. How the equipment can malfunction.

B. Equipment controls
C. Central processor hardware controls
D. Card reader hardware controls
E. Card punch hardware controls
F. Printer controls
G. Magnetic tape hardware controls
H. Hardware controls for direct access storage devices
I. Hardware controls for data communications
J. Other input/ output devices
K. Hardware controls and the audit

WEEK: 4

TIME ALLOTMENT: 3 Hours

OBJECTIVES/ LEARNING OUTCOME:

At the end of the topic, the students will be able to describe the general reasons for equipment malfunctions,
the general types of equipment controls and the controls associated with the major equipment items.

LEARNING CONTENT:

HARDWARE FEATURES FOR CONTROL OVER EQUIPMENT MALFUNCTIONS

Introduction
Modern computer equipment has a high degree of reliability and the auditor can usually assume that
the equipment will perform reliably and will detect machine-based errors if they occur. Except where
difficulties in processing are found to be a result of machine errors, the auditor can usually rely on the
efficacy of these controls. Some of these are operable, however, only if program instructions test the
equipment indicators. In order to understand the working of the equipment, to understand why it is reliable,
and to evaluate the effectiveness of machine checks in cases where there are problems, it is desirable for
the auditor to have a general understanding of built-in hardware control features. This chapter describes
the general reasons for equipment malfunctions, the general types of equipment controls and the controls
associated with the major equipment items.

Discussion
A. How the equipment can malfunction
A computer system consists of both electronic elements and mechanical parts. The central processing
unit, for example, consists almost entirely of electronic elements such as transistors, resistors and
diodes, whereas most input/output equipment and file storage devices contain both electronic
components and parts which move mechanically. Failure in the system, therefore, can result from the
malfunction of either a mechanical part or an electronic element.

Failure in Electronic Components

A computer system operates with electrical pulses that are created, counted, delayed, transmitted, etc.
Communication between units of the system is by a controlled transmission of electrical pulses, and the
electronic circuitry of the computer is designed to control the timing, shape, strength and frequency of
these pulses. Failure of an electronic element such as a transistor, resistor, diode, etc., may cause a
change in the timing, shape, strength or frequency of the pulses and lead to an error. Some of the
reasons for deterioration of an electronic element are extremes of heat or humidity, power disturbances,
mishandling and normal wear. Quality control in manufacture, built-in equipment checks and programs
of periodic preventive maintenance have made the electronic portion of the computer system very
reliable. Preventive maintenance procedures usually detect elements which are getting out of adjustment
or are close to failure and allow adjustments or replacements to be made. The preventive maintenance
is performed daily on complex computer systems and less frequently on simple configurations.
 Failure in Mechanical Operation
Mechanical operation is required in almost all input/output and file storage equipment. Two mechanisms
with mechanical actions are usually used:
1. A transport mechanism to move the media (input, output or file storage) past the reading or
writing mechanism
2. Mechanism to read or write
These actions occur at high speeds. For example, a card reader must transport up to 1100 cards a
minute past the read mechanisms (brushes or photoelectric cells) at a precise speed and in a precise
position. A printer that prints 1000 lines of 136 characters per minute requires as many as 136,000
individual print mechanism movements each minute. A machine error can be caused by a failure in the
timing, speed or movement of a transport mechanism or by a malfunction of the read/write units. Such
failures may result when devices get out of adjustment, are mishandled by operators, become worn,
and so on. Failures may also be traced to bad media such as warped cards, magnetic tape with surface
defects, or poor quality paper stock.

B. Equipment controls
Equipment controls are usually based on the concept of redundancy. This concept has received much
attention in communication theory but is also applicable elsewhere as a general basis for error control.
With respect to error control, redundancy involves the addition of an element to a process or to the
code for an item for the sole purpose of detecting any error which may occur. If there is no possibility
of error, the redundancy is usually eliminated. Equipment controls can be divided into five types:
redundant character check, duplicate process check, echo check, validity check and equipment check.
Each of these involves a separate operation which provides a check on the results of the main
operation. The general types of controls are described in this section; their application to specific
equipment is described in the remaining sections of the chapter.

Redundant Character Check

A redundant character is a character attached to a data item for the purpose of providing for error
detection. The redundant character is developed from the characteristics of the data item to which it is
attached. After an operation such as the moving of a data item in the system, the computation used to
obtain the redundant character is repeated to derive a second character. The two redundant characters
are compared, and, if they are the same, it is assumed that there has been no malfunction affecting
the data during the transfer or some other operation.

Duplicate Process Check

Another type of equipment control involves having the same process performed twice and the results
of the two operations compared. Any difference between the first operation and the second signals an
error. The duplicate process may be a complementary action such as reading after writing to check
what was written.

Echo Check
In an echo check the central processor sends a command to an input or output device to perform an
operation. The device returns a signal that verifies that the proper mechanisms for performing the
actions have been activated. This check verifies that the equipment was activated without testing the
actual results obtained.

Validity Check
Since, on many operations, only certain results can be considered correct, one method of checking is
to compare a result obtained against all valid results. Any result not fitting into this set of valid results
is considered incorrect.

Equipment Check
In this control, the computer checks the equipment to see if it is functioning properly instead of
checking the results from the operation. It is not a positive check, since the equipment may be
working properly while defective media or other factors are causing improper results.
C. Central processor hardware controls
The central processor has two equipment error-control problems. The first is that of ensuring that all
data elements transmitted through the internal circuitry of the central processor are transmitted
correctly and that no problem with the timing or pulses has destroyed or altered the data being moved.
The second is that of avoiding the performance of an operation that is invalid. The typical central
processor hardware controls are the parity bit, the operation validity check and interlocks. In some of
the very early computers there were duplicate circuitry and double arithmetic in order to ensure correct
functioning of the central processing unit, but these precautions are no longer considered necessary or
desirable.

Parity Bit
The computer represents data and instructions internally using devices which have
two recognizable states. A magnetic core, for example, is polarized in one of two
directions. Each of these two states is used to represent a binary digit (abbreviated
as bit) —one state represents a 1-bit and the other a 0-bit in the binary system.
Decimal numbers and alphabetic characters are converted to binary
representation for internal processing or for storage on magnetic media such as
magnetic tape and are converted back to decimal numbers and alphabetic
characters when printed on reports. There are various different approaches to
coding of data for internal representation in binary form. In all of these
approaches, the computer works with groups of storage devices, each of which
can hold either a 1-bit or a 0-bit.

A redundant character, called a parity bit or check bit, is associated with each
group of bits separately identified and moved as a separate unit through the
computer. This unit may be a large, fixed set of bits called a computer word, a
smaller set which can encode a single alphanumeric character, or a separately
identified part, called a byte, of a larger fixed word. The parity of the binary word,
character set or byte is made even or odd when the data is first converted to
binary form from input items or when data is formed from a computation or other
manipulation. The number of bits is summed and a 1-bit or a 0-bit is placed in the
parity bit position to make the total number of bits odd if an odd parity check is
used or even if an even parity check is used. Figure 4-1 (this page) illustrates the
parity bit in a 6-bit binary-coded character bit set. Each time this basic group of
bits is moved in the computer, the parity of the group is checked. If the parity bit
as newly computed is different than the parity bit as previously computed, a bit
has been destroyed. This test is not infallible since the destruction of two bits will
leave the parity bit unchanged, but the probability of this happening is very low.

There are a few computers (usually equipment with a scientific processing orientation) which do not use
a built-in parity bit control. Other reliability features including programming checks usually compensate
for this omission. Parity errors in the central processor are seldom encountered and many of the errors
which are registered are due to the checking circuitry itself rather than to an underlying error. The
auditor should be aware of the parity check, but the lack of this check does not usually affect his
assessment of the system or the extent of the audit tests.

Validity Check
In the central processor there are only certain operation codes which are valid and which the computer
can execute and there is only a certain range of numbers which the computer can access as memory
addresses. Before attempting to execute an operation code or access a memory location, the computer
usually performs a validity check to determine that it is a valid operation code or a valid address. Validity
checks are used in some computers with character coding to check the movement of data. Each bit set
can encode one character, but not all combinations are valid. An invalid combination indicates an error.

Interlocks
The computer system has automatic controls to prevent the equipment from attempting certain
operations at the wrong time. For example, there is an input/output interlock which prevents the
computer from signalling an input or output device to perform an action while it is already performing
another operation. Thus, a card reader cannot be signalled to read a card while it is already performing
 a card-reading operation. A storage protection interlock is a hardware control used with fairly advanced
computer systems that process several programs concurrently. It prevents the computer from using a
block of memory locations that are not assigned to the particular program.

D. Card reader hardware controls

The card transport mechanism in a reader must pull a card from the input hopper and move it past the
read stations (reading brushes or photoelectric cells) at a precise speed and in a precise position. (See
Figure 4-2, this page). Malfunction of the card reader may therefore occur because the cards move past
the reading stations at incorrect intervals of time, because the card is positioned incorrectly, or because
the reading mechanism fails to sense properly. Card reading speeds vary from 200 to over 1,000 cards
per minute, so that a slight delay or slight skewing can result in an incorrect sensing. The types of
controls commonly used in card readers are dual-read and hole-count controls (both duplicate process),
validity checks and photocell (circuit) checks. Another check, the double-punch blank-column check, has
been used in some older equipment. These checks are summarized in Table 4.1 (page 43).

When an error is detected by the card reader, an

internal switch is set. The typical error procedure is
for the card which was misread to be routed to a
different output bin (stacker) from those without
errors. Depending on the application, the computer
may halt processing for the error to be handled or it
may continue processing with the error noted for later
attention. If it halts, the reader will usually reject all
cards then in motion. A corrected card and the other
cards in process when the error was detected are re-
inserted. Since the more a card is handled, the
greater the chance of error, many installations prefer
to complete the run with errors noted but without
card re-entry during the run.

E. Card punch hardware controls

The problems of card punching are similar to the problems of card reading but there are some differences
in controls. These are summarized in Table 4.2 (below). In one approach, the results of the punching
operation are checked. There is a separate read station following the punch station. Cards are read after
they have been punched in order to develop information for either a hole count or a full comparison. A
second approach to punch checking is an echo check in which a signal is sent from the punch dies
verifying that they have been activated. Both approaches are widely used.
F. Printer controls
The first step in printing a line is to assemble the characters in storage. These characters are then loaded
into a print buffer where they are decoded into signals which will select the characters to be printed at
each print position. There are usually from 100 to 160 print positions on a line, the most common
numbers being 120 and 132. A hammer located behind the paper presses the paper and a ribbon in front
of the paper against the type face at the exact moment the selected character is in position (Figure 4-
3, page 45). After a line is printed the paper and the print ribbon are advanced and the hammer recoils.
There are two basic types of printers—impact and non-impact. Almost all are of the impact type which
uses a mechanically driven type face pressed against the paper and ribbon. Non-impact printers form
an image by electrical charges. This discussion deals only with impact printers. The major methods of
impact printing are type bar, drum, chain and type slugs.

The hardware error controls for printing, as detailed in Table 4.3 (below) are echo, print synchronization
and validity. Note that these controls test in various ways the mechanisms performing the printing but
none of them is a positive test of what was actually printed. In practice, however, these controls have
proved to be satisfactory.

G. Magnetic tape hardware controls

A magnetic tape drive consists of a feed reel, a takeup reel, read/ write heads, a mechanism to drive
the tape past the heads and a mechanism to dampen the effect of rapid stopping and starting. The
device used to drive the tape past the read/write heads is a pinch roller, a vacuum capstan or a clutch
operated capstan. Vacuum columns, mechanical storage arms or storage bins are used alone or in
combination to provide several feet of slack tape for immediate movement before the reels start their
independent motion. The most widely used tape units use pinch-roller drives and vacuum column buffers
(Figure 4-4, this page). Data is recorded by magnetizing small areas on the tape. A set of bits, encoding
either one character or part of a computer word, is arranged as a vertical row of six or eight bits on the
tape. A parity bit position is added to each row making the total number of bit positions equal to seven
or nine (Figure 4-5, page 47).

The positions are called channels. At the end of each record there is a blank space (usually ¾ inch)
called an interrecord gap (or, more precisely, an interblock gap) to separate the blocks on the tape, each
of which contains one or more records.

Problems with magnetic tape may result from difficulties with the magnetic tape itself (the primary
reason), from malfunction of the read/write heads, or from malfunctions in the tape drive mechanism.
The typical magnetic computer tape consists of a half-inch wide strip of polyester (plastic) film 2,400
feet long with a thin iron oxide coating. As the tape passes over the read / write head, any imperfection
in the coating may lift the tape from the head and cause a signal drop. These defects can stem from
manufacturing defects or from dust particles, but the most frequent source of difficulty comes from
imbedded particles of oxide. Flakes of oxide break away from the coating and are redeposited on the
tape surface by the tape heads and guides. These can be removed by a special cleaning process. Over
a period of time, parts of a tape become worn. These worn sections must be removed or skipped over
and not used for recording.

The proper operation of the read/write mechanism in the tape unit depends
upon correct signals from the central processor and correct speeds and
positioning as the tape moves past the read/ write heads. If the tape moves
at an improper speed the recording or reading mechanism will not operate
properly. A malfunction of the write head may cause a recording to be made
at less than an acceptable signal level. The drive mechanism has a device,
usually a photoelectric sensing device, to sense the beginning and end of the
tape. If this is not operating properly, the read/write heads may attempt to
write or read on the leader at the beginning or end of the tape. Incorrect
processing or loss of data will result. The hardware controls for magnetic tape
consist of a parity check for reading of data and a read-after-write comparison
for writing on tape.

The basic parity check for magnetic tape is a row check, often called a lateral
or frame check, in which each character (coded one character per frame) is
given a parity bit when the data is put on the tape. When the data is read
from the tape the parity bit is checked. An improved parity check is the
addition of longitudinal or track parity bits to give a two-dimensional check.
Each record encoded on the tape is given a track parity bit in addition to the
row parity bit associated with each frame. This serves as an added check over
the single dimensional row parity. In addition, the intersection of a missing
row bit and a missing longitudinal bit will, if only one bit is in error, define the exact bit position causing
the error and will allow for automatic error recovery (Figure 4-6, this page).

A cyclic or diagonal check is used in a few magnetic tape units. This is a check character formed by
taking a parity check diagonally instead of down or across. This provides additional capabilities for
pinpointing any bit positions that happen to be in error.
Read after Write Check
The read/write heads on a magnetic tape drive may take two forms: the single-gap head which acts as
both a read mechanism and a write mechanism (only one operation being performed at a time) and a
two-gap head which has both a read head and a write head. The two-gap head (Figure 4-7, page 49)
allows for a read immediately after a write comparison in which case the data just written is read and
compared. The data is tested to see if it was recorded at a proper signal level and parity is checked.
The automatic write/read comparison depends upon having the two-gap head. It is preferable from an
error control standpoint because recording errors are detected and corrected when they occur rather
than when the file is next processed. In the latter case, reconstruction is usually more costly and it is
also time consuming.

Other Tape Controls

A control used with some magnetic tape is that of counting the number of characters recorded and
writing this count on the tape. When the tape is read, the character count is again computed and
compared. An infrequently used method involves a dual recording of the information so that there is a
back-up recording of information in case of any problem with the major recording.

Handling of Tape Errors

Since most tape errors are caused by surface defects on the magnetic tape, an error in reading or writing
is usually handled by backspacing the tape one record and repeating the operation. If the error persists,
the operation is repeated again. Dust or oxide particles will usually be dislodged in this way. If the error
is still uncorrected, the record is noted on the error listing for subsequent correction and processing is
continued. A bad spot on the tape can be skipped by marking the beginning and end of the tape portion
which is not to be used.
 H. Hardware controls for direct access storage devices
A direct access storage device (drum, disk or strip file) consists of (1) a
transport mechanism to move the recording media, (2) read/write heads and
(3) recording material. The direct access devices can read or write directly on
any section of the recording medium and for this reason they are also referred
to as random access devices. Either the recording medium or the read/write
head, or both, move until they are in correct position.

A magnetic drum storage involves a revolving drum with a magnetizable

surface. The data is arranged in tracks and read/ write heads record or read
data as the surface revolves beneath them. Magnetic disk storage consists of
metal disks mounted on a shaft. As the disk revolves, data which is recorded
on concentric tracks passes beneath a read/write head mounted on an arm.
There is often a separate access arm for each disk, but the arm must usually
move in and out in order to be positioned above the track to be used. The
read/write head must be positioned and held a few ten-thousandths of an inch above the surface.

A common approach is to have the arm float on a layer of air. In the event of power failure, a safety
device retracts the heads to prevent them from damaging the disk. Strip file storage consists of
cartridges loaded with plastic strips coated with a recording material. For a read/write operation, the
selected strip is dropped from the cartridge onto a positioning drum which moves it past a read/write
head. Errors may occur in this type of equipment because of incorrect positioning or because of improper
reading or writing by the read/write mechanism. The major error control methods are the parity check,
check character and read-after-write check (Table 4.4, page 52).

A major error can occur in a disk storage device if information which is sent to be recorded at one
location is recorded instead at another. The recording at the incorrect location not only puts the
information where it should not be, but also destroys information at the location improperly used. In
order to avoid the possibility of such an error’s happening through an improper movement of the
read/write head, it is considered good practice, either through hardware controls or through
programming, to compare the address of the location at which data is to be written with the address
called for by the instructions.

I. Hardware controls for data communications

Errors occur in data communications primarily as a result of (1) the inherent characteristics of the
communication links and (2) equipment malfunctions. The most common causes of data transmission
errors are noise, fading, amplitude and phase distortion, and multipath distortion. A full discussion of
these sources of error is beyond the scope of this review. In general, a switched or dial-up connection
will have a higher error rate than a leased line and the error rate tends to increase with the distance
over which the data is transmitted and the transmission speed.

Methods of Error Control

Errors will occur in every data communications system. Therefore, the facts to be noted with respect to
errors are these: degree of freedom from errors, probability of detecting errors which occur and efficiency
in handling these errors. The methods of error control are code validity and data redundancy. The
redundancy controls involve either a constant ratio code or parity bits. These controls are summarized
in Table 4.5 (page 53). Where data is transmitted directly into a computer, the program accepting the
data may perform additional error control procedures.

Handling of Hardware-Based Communication Errors

It is more difficult to correct errors than to detect them. There are codes which are designed for efficient
correction of data transmission errors but they require a relatively large number of redundant checking
bits and are expensive to implement. The most common approach to error handling is retransmission of
erroneously received messages or segments of messages. The retransmission may require operator
intervention or it may be handled automatically. In one method, the receiving terminal checks each
segment (character, block or message) as received and transmits an instruction back to the sending
terminal either to repeat the transmission because there is an error or to continue with the next segment.
J. Other input/ output devices
Magnetic Ink Character Readers
Readers for magnetic ink character recognition (MICR), used extensively in banking, have validity and
timing checks built into the equipment. The timing check tests that the documents are moving past the
scanning heads at the proper speed. As with other validity checks, the patterns read by the scanning
heads are compared with the set of valid symbols. In addition, the circuitry may perform other checks.
It may count the characters in the field to determine that the size of the field being read is correct or
automatically compute a check digit and compare it with a check digit in the field being read.

Optical Scanning Equipment

An optical scanner reads characters or marks by scanning with a beam of light. A character is read by
recognizing a particular pattern of light and dark areas. There are several different methods for
performing this task. There are two rates to consider in optical reading—the reject rate and the error
rate. The reject rate is the percentage of documents rejected because the equipment is unable to
recognize the character. At present, reject rates range from 2-20%. The error rate is the percentage of
documents which were read but which contained one or more characters incorrectly identified. The error
rate typically ranges from less than 1% of documents up to 2%.

The reject rate is significant in terms of handling time and reprocessing. The seriousness of errors
undetected because of misreading depends on the type of application. A 1 % error rate may be quite
acceptable for one application but totally unacceptable for another. Programmed tests discussed in
Chapter 6 can be used to detect many of the errors that are not found at the time of reading.

K. Hardware controls and the audit

There will be errors in data processing equipment. There are, however, satisfactory methods for
detecting such errors and in some cases automatically correcting them. The auditor can usually put
considerable reliance on the equipment and the hardware controls for detecting errors. The auditor is
interested, for background purposes, in knowing the major types of equipment controls and in having
an assurance that they are operating properly. This information is helpful for understanding the error
procedures established by the client. The lack of a particular type of equipment control should probably
not alter the scope of the audit unless the system does not seem to be operating within a tolerable rate
of errors. In some cases, the hardware controls are not automatic but must be tested by programmed
instructions. The fact that a client does not implement hardware controls requiring programmed tests
may indicate a lack of administrative controls.

The auditor should be able to obtain general information on the reliability of the equipment from
operating reports and from computer logs which record periods of machine breakdown (downtime) and
the reasons for it.
 In most cases the hardware error detection methods are satisfactory and do not require special audit
attention. However, procedures for handling detected errors are themselves sources of possible errors
that may go undetected. For example, a card reader with a validity check will reject a card with invalid
punch combinations or a card which was sensed incorrectly so that it appeared to be invalid. If the
procedures for handling this error are not followed properly, the result may be a double reading of the
card or a skipping of the card. In reviewing processing procedures, therefore, the auditor should usually
devote more attention to the procedures for handling errors than to the hardware controls which detect
them.

Summary:
Malfunction in the equipment can be prevented through the following control:
 Equipment controls
 Central processor hardware controls
 Card reader hardware controls
 Card punch hardware controls
 Printer controls
 Magnetic tape hardware controls
 Hardware controls for direct access storage devices
 Hardware controls for data communications
 Other input/ output devices
 Hardware controls and the audit

Through the use of the above control, the internal control of the business will increase and can assure the
reliability and relevance of the information that will be generate in the system.

REFERENCE:
1. Davis, G. CPA, PhD. Auditing and EDP. American Institute of Certified Public Accountants, Inc. 666
Fifth Avenue, New York, N.Y.

Congratulations for finishing the Module 4! Keep up the good work.

Prepared by: Review/Approved:

AIZA P. RUMAUAC, CPA LILY MAE M. KIMAYONG, MBA

Instructor Vice President for Academic Affairs