INTRODUCTION TO

ANALOG COMPUTER
PROGRAMMING
oaoe l.Rummnr

A
/
INTRODUCTION TO

ANALOG COMPUTER
PROGRAMMING
INTRODUCTION TO

ANALOG COMPUTER
PROGRAMMING

DALE I. RUMMER, PH.D.

University of Kansas
Lawrence, Kansas

HOLT, RINEHART AND WINSTON, INC.

New York, Chicago, San Francisco, Atlanta, Dallas
Montreal, Toronto, London, Sydney
 To My Wife
PAT

Copyright © 1969 by Holt, Rinehart and Winston, Inc.

All rights reserved
Library of Congress Catalog Card Number 69-17660
SBN: 03-080362-4
Printed in the United States of America
1 2 3 4 5 6 7 8 9
PREFACE

Why should another book on analog com­

puters be written? In particular, why should this book have been
written? This book was produced in response to a need for an
introductory text on analog computer programming written for
second semester freshmen or first semester sophomores. Such a
student would have previously taken a five credit course in cal­
culus and analytic geometry. It is assumed that the student using
this text has proficiency with the integration and differentiation
of simple functions. Definitions and explanations are provided
for the ideas needed from mechanics, thermodynamics, and
electrical circuits.
Why teach analog computer programming at
this level? Why not wait until the students have taken a course
in differential equations? The experience that a student gets in
learning about analog computer programming helps motivate
his study both of mathematics and engineering. The student's
concepts of integration and differentiation are strengthened by
further application to real problems and by the graphic display
of the solutions. Simple differential equations are introduced
because such equations are needed to describe systems which
involve rate processes. References to the analytical solutions to
these equations are minimal because the student is not assumed
to have previously seen these solutions. The background pro­
vided by a study of analog computer programming is most
helpful to students when they later study the subject of differen­
tial equations.
The first chapter provides a perspective on
computers in general and upon analog computers in particular.
The second chapter introduces the arithmetic operations as
implemented using potentiometers and operational amplifiers.
Amplitude scaling of the problem variables to fit the limitations
of the analog computer is described. These techniques are then
applied to the solution of algebraic equations. The third chapter
introduces the implementation of the operations of integration
and differentiation using operational amplifiers. These techniques
are applied to the generation of ramp, polynomial, and sinusoidal
functions. Chapter 4 describes the various methods of recording

v
vi PREFACE

the solutions generated by the analog computer. Chapter 5 shows

how first-order differential equations naturally arise when one
wishes to describe systems involving rate processes. The method
of obtaining the analog computer block diagram from the
mathematical equations describing the system is developed. The
need for time scaling arises and a simple method of accomplishing
time scaling is described. Time scaling and amplitude scaling
are considered as separate operations which are to be done at
distinctly different times in the process of preparing a problem
for solution on an analog computer. Chapter 6 considers some
systems which require second-order differential equations to
adequately describe the physical phenomena. The physical pro­
cesses are described fully to help the student understand how the
equations for the mathematical model are produced. Time scaling
is considered again to demonstrate the more general process of
time scaling. Chapter 7 introduces some of the nonlinear opera­
tions which are possible on the analog computer. Again these
new ideas are introduced in response to the need for new tools
in solving certain classes of problems. If there is not time avail­
able to cover this material in an introductory course, the students
can read the material later as it becomes appropriate to the prob­
lems which they are interested in solving.
Because there are many different analog com­
puters in use in schools it is not practical to describe them all
in detail. The details of the operation of some particular com­
puters are placed in the appendix. The student can refer to the
description of the computer which is available to him if it is
described, or to the one which is most nearly like the one which
he is using.
I am indebted to the many persons who have
written papers and books on analog computing. This indebted­
ness is difficult to acknowledge specifically without a high
probability of doing some persons an injustice. I do wish, how­
ever, to acknowledge that the approach which I have taken to
scaling is due to the influence of a former colleague of mine,
Dr. J. Robert Ashley. * This definition of scale factors in terms
of problem units/volt seems to make both the creation of the
analog model from the equations of the physical system, and
the interpretation of the analog computer results in terms of the
physical system being studied easier to understand. I urge that
those teachers who may be accustomed to using the reciprocal
scale factors try this other approach with their students.
I am most appreciative of the comments of
students and teachers who have used these notes in mimeographed
form over the past several years. My wife has been most helpful
not only in typing the manuscript, but also in reading it critically.
Her questions have eliminated many errors and have helped to
clarify some obscure arguments.
D.I.R.
*J. R obert Ashley,
Introduction to Analog Computation (New York: John Wiley & Sons,
 TABLE OF CONTENTS

PREFACE, vii

CHAPTER 1 PERSPECTIVE, 1
1.1 Digital vs. Analog, 2 1.2 Mathematical Modeling, 3 1.3
Active vs. Passive, 4 1.4 Past, Present, and Future, 5 1.5
Exercises, 6

CHAPTER 2 ANALOGS OF ARITHMETIC OPERATIONS, 7

2.1 Operational Amplifiers, 7 2.2 Amplitude Scaling, 9 2.3
Inversion, 11 2.4 Summation, 13 2.5 Multiplication by a
Constant, 16 2.6 Solution of Algebraic Equations, 18 2.7
Solution of Simultaneous Linear Equations, 21 2.8 Exercises, 27
CHAPTER 3 ANALOGS OF CALCULUS OPERATIONS, 33
3.1 Differentiation, 34 3.2 Integration, 40 3.3 Generation
of Ramp Functions, 47 3.4 Generation of Polynomial Func­
tions, 49 3.5 Generation of Sinusoidal Functions, 53 3.6
Exercises, 54

CHAPTER 4 THE RECORDING AND DISPLAY OF SOLUTIONS, 61

4.1 Strip Chart Recorder, 62 4.2 X-Y Plotters, 62 4.3
Cathode Ray Oscilloscope, 64 4.4 Exercises, 65
CHAPTER 5 FIRST ORDER SYSTEMS, 70
5.1 Creation of Analog Models of Dynamic Systems, 71 5.2
Hydraulic System, 73 5.3 Time Scaling, 82 5.4 Freely
Falling Body, 84 5.5 Falling Body with Viscous Drag, 88
5.6 Thermal Behavior of a Semiconductor Device, 92 5.7 Hy­
draulic System of Three Tanks, 98 5.8 Exercises, 104
CHAPTER 6 SECOND ORDER SYSTEMS, 109
6.1 Spring-Mass-Dashpot System, 110 6.2 Non-zero Forcing
Function, 117 6.3 The Swinging Door System, 120 6.4
Simple Pendulum, 128 6.5 Exercises, 134

vii
Viii C O N TEN TS

CHAPTER 7 NON LINEAR OPERATIONS, 137

7.1 Diode Limiters, 137 7.2 Diode Function Generators, 144
7.3 Multipliers, 145 7.4 Sine Cosine Generator, 148 7.5
Comparators, 152 7.6 Exercises, 154

BIBLIOGRAPHY, 156

APPENDICES

I-A BASIC OPERATION OF THE ELECTRONIC ASSOCIATES INC.

TR-20 Analog Computer, 158 l-A-1 General Description, 158
I-A-2 Multiplication by a Constant, 159 I-A-3 Inversion and
Summation, 160 I-A-4 Integration, 162 I-A-5 Normal
Operation of TR-20, 163 I-A-6 Repetitive Mode of Operation
of TR-20, 163 I-A-7 References, 164
I-B BASIC OPERATION OF THE BURR-BROWN MODEL 600
ANALOG COMPUTER, 165
I-B-l General Description, 165 l-B-2 Multiplication by a Con­
stant, 166 I-B-3 Inversion and Summation, 168 I-B-4 Inte­
gration, 169 I-B-5 Multiplier, 170 I-B-6 General Operation,
172 I-B-7 References, 172
I-C ELECTRONIC ASSOCIATES INC. TR-48 ANALOG COMPUTER, 173
I-C-1 General Description, 173 I-C-2 Multiplication by a
Constant, 174 l-C-3 Inversion and Summation, 175 l-C-4
Integration, 176 I-C-5 Non Linear Modules, 176 I-C-6
General Operation, 177
II DERIVATION OF ANALOG COMPUTER GAINS, 180
II-A. Inversion, 180 II-B. Summation, 183 II-C Integra­
tion, 185
III ELECTRICAL CIRCUITS, 188
III-A. Pot Set, 188 III-B. Initial Condition, 190
IV STABILITY OF SOLUTIONS FOR ALGEBRAIC EQUATIONS, 192

INDEX, 197
INTRODUCTION TO

ANALOG COMPUTER
PROGRAMMING
CHAPTER PERSPECTIVE

The purpose of this chapter is to place the contents of

this book — an introduction to the programming of electronic differential
analyzers, or EDA for short — into proper perspective in the larger realm
of computers and the use of computers. A proper perspective is important
in order to appreciate the advantages and the limitations of the EDA as
compared to other kinds of computers. For many years there was a
running battle between the proponents of analog computers and the
advocates of digital computers. Much heat was generated, but very little
light was shed on the subject because few people had adequate com­
petence in both areas to make really valid comparisons. About 1960 some
people realized that for some problems a combination of an analog com­
puter and a digital computer would be advantageous. Such a combination
is called a "hybrid computer," and much work is being done developing
hybrid computation techniques.

1
2 PERSPECTIVE

1-1 Digital vs. Analog

Numbers are represented in a basically different way in these two types

of computers. The digital computer represents a number by one of a set of
discrete conditions. The analog computer, on the other hand, represents
numbers in terms of a continuously variable condition. Consider a dial
which has ten positions numbered 0 to 9. Mechanical detents prevent this
dial being set at any position other than the numbered positions. This dial
permits a digital representation of the numbers 0 through 9. Consider, on
the other hand, a dial that has numbers in the same positions as before
but that does not have the mechanical detents. This dial could be set half
way between 7 and 8 to represent 7.5. Any device that represents numbers
in terms of a continuous variable is analog in nature. Some examples of
digital devices are as follows: the abacus, an adding machine, a desk
calculator, a traffic counter. Some examples of analog devices are as
follows: an ordinary ruler for measuring, a slide rule, a voltmeter, a
micrometer, a mercury thermometer, a pressure gauge. The precision of
the representation of numbers in analog computers seldom exceeds 0.05
per cent and is limited by the precision and stability of the components
available. The critical components of analog computers are sometimes
placed in temperature controlled ovens to improve the stability of the
values of the components. The precision of the representation of numbers
in digital computers is not determined by the accuracy or stability of the
components, but rather depends on the quantity of components used to
represent the number. The accuracy of representation of numbers in a
digital computer is usually described in terms of the number of binary
bits used to represent a number. Ten binary bits give an accuracy of one
part in (210 — 1) or one part in 1023. This is the order of 0.1 per cent which
is the precision of good quality analog computers. Most digital computers
use at least 16 binary bits to represent a number and many use as many as
32 bits or more. It is evident that the digital computer can represent
numbers much more accurately than can the analog computer. Digital
computers also may use what is called "floating point" representation of
numbers to obtain an even greater range from very, very small numbers
to very, very large numbers. For many problems the extreme accuracy of
the digital computer is not needed, however, nor even appropriate, and
the accuracy of the analog computer is adequate. By means of suitable
scaling, the analog computer can represent numbers of any desired size.
The digital computer solves a problem by performing a suitable number
 MATHEMATIC AL MODELING 3

of computations one after another in serial fashion. The analog computer,

on the other hand, performs many operations at the same time, that is, in
parallel fashion. Typically the analog computer can solve systems of
differential equations much more rapidly than can the digital computer.
The complexity of the problem that can be handled by an analog com­
puter is limited by how much computing equipment is available. The
complexity of the problem that can be solved on the digital computer is
limited by the rate at which the digital computer operates and by how
much computer time is available. In both cases the ultimate limitation on
the size of the problem is really one of money. If a problem can be solved
by both the digital computer and the analog computer and if both kinds
of comp uters are available, the decision should be made on the basis of
which will do the job most economically and conveniently. This book is
concerned with solving problems using the EDA, which is frequently
called by the shorter name, analog computer.

1-2 Mathematical Modeling

Before a problem can be solved by any kind of computer, or indeed

before it can be solved even by hand calculation, the problem must be
described in quantitative terms. Frequently the actual relationships may
be very co mplicated, or in some cases not precisely known. Simplifying
assumptions are needed many times to permit solving the problem with or
without the aid of computers. One of the advantages of using computers
as an aid to solving problems is that fewer simplifying assumptions may
be needed. The equations that define the relationships between the vari­
ables in a problem form the mathematical model for the problem. An­
other advantage of using computers is that several different models may
be tried for one problem in order to find the one that is the best model.
Before rel ying on a particular model for a given problem, the engineer
must check the results in some way to find how good a model he has
created. For example suppose that one is interested in how fast water will
empty from a tank. As a first assumption one might consider that the rate
at which the water runs out is proportional to the depth of t he water in
the tank, that is, the pressure at the bottom of the tank. A very simple
experiment will show that this model may be grossly inaccurate. See
Exercise 2. Once the accuracy of the assumed relationships has been
verified, the engineer can proceed with confidence that the solutions which
he produces based on the assumed models will be correct to the accuracy
verified.
4 PER SPEC TIVE

1-3 Active vs. Passive

If two devices have mathematical models of the same form, then one
of these may be used as an analog of the other. The mathematical equa­
tion that relates angular velocity, a>, iner tia, J, and mechanical damping,
D, for a simple rotational mechanical system will later be shown to be

(1-1)

The mathematical equation that relates current, i, inductance, L, and re­

sistance, R, for a simple series electrical circuit will be shown to be

(1-2)

Thus the electrical circuit could be considered to be an analog of the

mechanical system by considering inductance, L, as analogous to inertia,
J, current, i, as analogous to angular velocity, co, a nd resistance, R, as
analogous to mechanical damping, D. Results obtained by studying the
electrical circuit can be interpreted in terms of performance expected
from the mechanical system. This kind of use of a nalogies is very helpful
in extending our understanding of, for example, electrical circuits to an
understanding of the dynamics of mechanical systems, or conversely, if we
are better acquainted with mechanical systems. We find as we study in
different fields of engineering and science that the same mathematical
equations occur repeatedly with only the names and/or the symbols
changed.
If actual experiments were performed on an electrical circuit in order to
predict the performance of a mechanical system as suggested above, then
the electrical circuit could be called an analog computer. In order to
provide a distinction between this kind of computer and the others which
we will be studying, the former are called "passive analog computers."
Passive analog computers have been built for modeling electric power
distribution systems, water distribution systems, petroleum reservoir
dynamics, ground water flow, and many other systems. Those models
based on systems of simultaneous algebraic equations, such as models for
electric power systems, can now be economically handled by large scale
digital computers. Those models based on systems of simultaneous partial
differential equations, such as models for ground water dynamics, can be
 PAST, PRE SE NT , AN D FU TU RE 5

solved on large digital computers, but a very large amount of time is

required. Passive electrical analogs using networks of resistors and ca­
pacitors are still very useful for some studies in these areas. Passive elec­
trical ana logs using resistance paper are useful for solving problems in
such diverse fields as electrostatics and the growth of cities. Although the
passive analog computers have been very important and in some activities
will continue to make useful contributions, the major applications of
analog computing require the active type of analog computer, which is
properly called the electronic differential analyzer, or EDA. Passive
analog computers using resistance, inductance, and capacitance are
limited in accuracy because physically realizable inductors may have
appreciable power loss. The active analog computer using only resistors,
capacitors, and operational amplifiers can model any linear physical
system. For example, the EDA using resistance and capacitance elements
can model a n electrical circuit that has inductance present in addition to
resistance and capacitance.

1-4 Past, Present, and Future

The electronic differential analyzer, EDA, as implied by the name,

makes use of electronics. The theoretical groundwork was laid in the late
1920's by such men as Black at Bell Telephone Laboratories. The reduc­
tion of the theory to practical form did not occur until the late 1930's
with improved vacuum tubes and with improved circuits for direct
coupled amplifiers. One of the major problems with the first analog com­
puters was adjusting the balance so that the output voltage would be zero
when the input voltage was zero. Most present day computers have auto­
matic balancing by a process called "chopper stabilization." Thus the user
of such ana log computers need not concern himself with the problem of
balancing the amplifiers. A check on the balance is made periodically as
part of preventive maintenance activity. The development of the transistor
as a substitute for the vacuum tube has permitted the physical size and the
electric power consumption of analog computers to be substantially re­
duced. A vacuum tube analog computer that had a volume of 7 5 cubic
feet and consumed 3500 watts of electric power has been reduced to 15
cubic feet and consumes 500 watts. The accuracy of these two computers
is comparable. The development of solid state multipliers and function
generators as replacements for the electromechanical devices used on
earlier analog computers permits the operation of the computer at higher
frequencies. Thus 500 to 1000 solutions may be generated in the time
previously required for the generation of one solution.
 6 PERSP ECT IVE

Digital computer programs exist today for taking the statement of a

problem in equation form and producing the specifications for connecting
up the analog computer model of the problem. The technology exists,
though not at present in economical form, to have the digital computer
connect up the analog computer components, to set all the coefficients,
to operate the analog computer, and to record the results. Although such
a system may not be appropriate for solving problems on analog com­
puters, it is most probable that such a system will evolve in hybrid com­
puting that makes use of both a digital computer and an analog computer
to solve problems cooperatively more economically than would be
possible using either computer separately. For example, launch windows,
those time intervals during which it is possible to launch a rocket for a
particular target in outer space, can be computed more economically on a
hybrid system than on a purely digital system, and one would not con­
sider doing it on a purely analog system. The development of tran sistor
operational amplifiers makes it worthwhile to construct special purpose
analog computers as part of a real-time data processing system, or as a
part of a control system.

1-5 Exercises

1. Some devices and processes are basically analog and others are basically
digital. For each device or process listed below, state whether it is digital or
analog in nature and state your reasons. Some of these devices or processes
may be either depending on certain variations of the device or the process.
a. a stop watch
b. an automobile speedometer
c. the odometer on an automobile
d. the tuning of a radio receiver
e. the computation of the cost of gasoline delivered by a gasoline pump
t. the playing of a piano
g. the playing of a violin
2. You are to create a mathematical model that describes the relationship
between the rate at which water flows from a hole in the side of a tank and
e height of the water above the hole. Devise and describe a procedure
using only materials normally found in a kitchen to obtain data on this
process.
3. If a a COnvenient to do so> carry out the experiment devised in Exercise 2
f
and determine an appropriate mathematical model from the data obtained.
Hint, semilog graph paper may be useful. If it is not convenient to carry
on thee^Penme^dutermme an apPr°Priate mathematical model based
on the data provided below.
Height-in. 23579
Flow rate-oz/sec .284 .347 .448 .529 .600
 2 ANALOGS OF
ARITHMETIC OPERATIONS

The purpose of this chapter is to show how the basic

arithmetic operations of addition, subtraction, and multiplication can be
performed on an analog computer. These methods will then be applied to
solve some algebraic equations. Since most of the operations performed
on an analog computer make use of operational amplifiers, the properties
of opera tional amplifiers will be considered first. The problems must be
amplitude scaled because the operational amplifiers are capable of operat­
ing only over a limited range.

2-1 Operational Amplifiers

An operational amplifier is an electronic device that amplifies or multi­

plies electrical voltages. Amplifiers are used in many familiar situations
such as radios, phonographs, tape recorders, and hi-fi music systems. The

7
8 ANAL OGS O F A RIT HMETIC OP ERATIONS

strength of an electrical voltage is measured in volts by an instrument

called a voltmeter. Voltage is always measured between two points and so
may be referred to as potential difference. One of these points, particularly
in the case of analog computers, is frequently a reference point called
ground. Thus when we speak of the output voltage of Amplifier 1 on an
analog computer, we mean the potential difference between the output
terminal of Amplifier 1 and computer ground. The potential difference
between the terminals of an ordinary flashlight cell (a zinc-carbon cell) is
1.5 volts. Most automobiles today have a battery consisting of six lead-
acid cells connected in series. The potential difference for each cell is
about 2.2 volts and thus the battery has a terminal voltage of 13 volts.
Since the voltage under operating conditions drops to about 12 volts, the
system is described as a 12-volt system. The voltage between the terminals
of a phonograph pickup is a few millivolts. A millivolt is 0.001 volt. A
phonograph amplifier amplifies or increases this potential difference to a
few volts applied to the loudspeaker. Thus the phonograph amplifier
multiplies the input voltage by a factor of approximately 1000. The ratio
of the output voltage to the input voltage for an amplifier is called voltage
gain. The phonograph amplifier described has a voltage gain of approxi­
mately 1000. The operational amplifiers used in analog computers
typically have voltage gains of 50,000 to 50 million. This very high voltage
gain is maintained as long as the output voltage is within prescribed limits.
These limits for transistorized amplifiers are usually ±10 volts, and the
limits for vacuum tube type amplifiers are usually ±100 volts. Some
transistorized amplifiers have a ± 100-volt range. These limits impose the
need for amplitude scaling that is discussed in the next section. Another
property of operational amplifiers is that the input resistance is very high.
This means that the input current is very small, since by Ohm's law the
input current is the ratio of the input voltage to the input resistance.
Consider the following example:

EXAMPLE 2-1

An operational amplifier has a voltage gain of 100,000 and an input

resistance of 5 megohms. Compute the input current when the output
voltage is 75 volts.

Via ~ gain = 100,000 = 75 X 10_5 volt

J kin 75X10~5 1c ^n
" ~ J?i„ " 5 X 10« = 15 x 10 amP
 AMPL ITUDE SCA LIN G 9

2-2 Amplitude Scaling

Amplitude scaling is required in order to place the range of the problem

variables withi n the range of voltage possible on the analog computer.
This is similar to the process used in making a scale model of a ship.
Assume that the real ship is 200 ft long and the block of wood available
for the model is 10 in. long. The scaling used must be such that the maxi­
mum dimension of the ship when reduced by a scaling factor will produce
a length which is not greater than the length of the block, that is, 10 in. If
a larger scale factor is used, one end of the model ship will be beyond the
end of the block of wood, and hence the model will not be complete and
accurate. The scale factor needed to convert from distance on the real
ship to distance on the model ship can be computed as follows:

u _ (max |L| of model) 10 in. A nc /f!t

kd ~ (max |/| of ship) = 200ft = °-°5 m'/ft

To convert any distance (length, width, or height) on the real ship to the
corresponding distance on the model ship the distance in feet is multiplied
by the scale factor, kd. Assume that a cabin on the real ship has the
following dimensions:
length, / = 40 ft; width, w = 20 ft; height, h = 10 ft

The dimensions of the cabin on the model ship are computed as

length, L = kd * I = (0.05 in./ft) * (40 feet) = 2.0 in.

width, W = kd * w = (0.05 in./ft) * (20 feet) =1.0 in.
height, H = kd* h = (0.05 in./ft) * (10 feet) = 0.5 in.

Notice how the units of distance, feet and foot, cancel when the product is
taken to yield the unit of distance, the inch, for the model ship dimensions.
A conversion factor to convert from distance on the model ship to dis­
tance on the real ship is sometimes desired. This factor can be computed
as:

_ (max |/[ of sh ip) = (200 ft) = Qf ,.

d (max \L| of model) (10 in.)

A distance, L, on the model ship is multiplied by the scale factor, Kd , to

give the corresponding distance, /, on the real ship. Assume that the mast
10 A NALOGS OF ARITH ME TIC OPE RA TION S

on the model is 3.0 in. high. The height of the mast on the real ship is
as follows:
h = Kd * H = (20 ft/in.) * (3.0 in.) = 60.0 ft

The maximum voltage that an operational amplifier is rated to deliver

corresponds to the length of the block of wood used to make a model of a
real ship. If the scaling is incorrect such that the model is required to
produce a larger voltage than the rated maximum value, the model will be
distorted and inaccurate. A convenient symbol for this maximum voltage
is Cm, which stands for "computer maximum." Assume as before that the
maximum distance on the real device to be modeled is 200 ft and the
rated maximum output voltage from the operational amplifiers is 10 volts.
The scale factors, referred to as amplitude scale factors to distinguish
them from time scale factors to be discussed in Chapter 5, are computed
as follows:

{Cm of c omputer) _ (10 volts) _ 0 os vn1t/ft

d (mas \d\ of device) (200 ft)
(max \d\ of device) _ (200 ft) _ f , ,
Kd ~ (Cm of computer) (10 volts) 7

A distance of 50 ft would be represented on the computer by

D = kd* d = (0.05 volt/ft) * (50 ft) = 2.5 volts

A voltage of 5 volts on the computer would represent an actual distance of

d = Kd* D = (20 ft/volt) * (5.0 volts) = 100 ft

Throughout this book lower-case symbols will be used for variables in the
real device or system, and the corresponding upper-case letters will be
used for the related variables in the analog computer model. Note also
that the lower-case letter is used for kd, which multiplies actual system
variables (lower-case) to obtain the computer model variables. Likewise
the upper-case letter is used for Kd, which multiplies computer model
variables (upper-case) to obtain system variables. The word system has
been introduced as a means of referring to the thing being modeled. When
a specific case is considered, there may be appropriate adjectives to
further identify the system of interest, that is, hydraulic system, vertical
position control system, speed regulating system, suspension system.
After the amplitude scaling has been established, the next step in creating
 INV ER SION 11

a model on an analog computer is to model the mathematical operations

required by t he problem. These will be considered next.

2-3 Inversion

Inversion is the name used in analog computing to identify the process

of multiplying a variable by — 1.0. Thus if a voltage of 5.0 volts is inverted,
the result is —5.0 volts . If a voltage of — 8.0 volts is inverted, the result is
+8.0 volts. If a variable, X, is inverted, the result is —X. The interconnec­
tion of analog computing equipment to create the model of the real system
is described by block diagrams. The analog computer block diagram
serves a function very similar to that of the flow chart for a digital com­
puter pro gram. The block diagram symbol for the process of i nversion,
an inverter, is shown in Fig. 2-1. The triangle indicates that an operational

input 1

FIG. 2-1

Block Diagram Symbol for an Inverter.

amplifier ha s been connected as an inverter. The line from the vertex on

the right side is the output of the operational amplifier. The line connect­
ing to the left side of the triangle is the input to the inverter. The number
"1" shown beside this line means that the input is to be multiplied by
-1.0. If the number "1" is omitted, the implication is that the factor is
still —1 .0. In some cases a different multiplying factor such as 0.1,2.0, 5.0,
or 10.0 may be desired. The presence of such a number instructs the
person c onnecting up the analog computer to connect the operational
amplifier to give the indicated multiplying factor, or gain, as it may be
called. The connection of an operational amplifier as an inverter is shown
in Fig. 2-2. Th e pie-shaped symbol represents the operational amplifier.
The right vertex is the output terminal and G and B are the input terminals.
The number 01 identifies which operational amplifier is used. For a de­
tailed ex planation of how this circuit functions see Appendix 2-A. The
following is a simplified explanation. The potential difference between the
input termin als of the operational amplifier, B and G, is very small be­
cause the voltage gain of the operational amplifier is very high, for example
100,000. When the output voltage of the amplifier is 100 volts, the in­
put voltage would be 0.001 volt. For output voltages less than 100 volts
12 A NALOGS OF ARITHME TIC OP ER A TION S

SJ R/
x> wv A/W
i2

0
G
ground

}' -

FIG. 2-2
Connection of an Operational Amplifier as an Inverter.

the input voltage would be even smaller. Since the input resistance of
the operational amplifier is very high, the current into the B t erminal of
the amplifier is very small. The potential difference, V, across a resistor,
the current, / , throu gh the re sistor, and the va lue of the res istor, R ,
are related by Ohm's law as follows:

(2-3a)

The potential difference is measured in volts, the current is measured in

amperes, and the resistance is measured in ohms. Ohm's law can be
solved for current, I, to show that the electrical current flowing in a re­
sistor is equal to the ratio of the potential difference across that resistor
and the value of that resistor. The potential difference across the input
resistor, Ri} in Fig. 2-2 is essentially the voltage, X, because the potential
of th e point SJ wi th respect to ground, G, is very nearly zero. Therefore
the current, I\, is as follows:

(2-3b)

By Ohm's law, the potential difference across a resistor is equal to the

product of the current flowing through that resistor and the value of
resistance of that resistor. Thus the potential difference across the feed­
back resistor, Rf, is h * Rf with the right end of Rf being at a lower
potential than the point SJ. Since the potential of SJ is nearly zero, the
potential of the output terminal with respect to ground is as follows:

Y = -I2* Rf (2-3c)
 SUMMATION 13

Since the current into the summing junction is very, very small, the cur­
rents h and I2 are essentially equal, and the substitution of Eq. (2-3b) into
Eq. (2-3c) yields t he general relationship:

7 »-3d>

2-4 Summation

The name summation describes the analog computer process of adding

up a set of v ariables. Subtraction is not considered separately, but rather
is accomplished by summing the negative of the quantity to be subtracted.
This need for the negative of a quantity points to an application of the in­
verter of th e previous section. The block diagram symbol for the process
of summation, called a summer, is shown in Fig. 2-3. The triangle indi­
cates that the operational Amplifier 1 has been connected as a summer.

5= - ( A +2.0 5+ 5.0C)

FIG. 2 -3

Block Diagram of a Summer.

The line from the vertex on the right side is the output of the operational
amplifier and is identified by the symbol, S. The three inputs to the sum­
mer are identified as A, B, and C. The number by each of the inputs
shows the factor by which that input is multiplied before being summed.
The output from the summer is therefore as follows:

5 = - ( A + 2.0 * B + 5.0 * C) (2-4a)

The negative sign on the process arises for the same reason that the sign
change takes place in the inverter. In fact, if the variables B a nd C are
made equal to zero, the summer shown in Fig. 2-3 reduces to the inverter
shown in Fig. 2-1. Fig. 2-4 shows the connection of an operational ampli­
fier as a summer. The gain for each input is computed as:
 ANAL OGS O F AR ITHMETIC O PER AT IONS
14

(2-4b)
-m

Rf
1 meg
AAAr-
Ri
1 meg
SJ
A > AAAr 01 ys
Ri
0.5 meg
r
ground
yy AA/V-

Rt
0.2 meg
cy yy AW

= [ A+ (2.0)5 + (5.0)C
]

FIG. 2-4

Connections of an Operational Amplifier as a Summer.

The possible gains are determined by the available values of resistance

for the feedback and input resistors. On some computers the operator
must plug in these resistors and therefore has a wide choice of ratios and
hence a wide choice of gains. On other computers these resistors are
permanently connected, the gains are marked for each input terminal,
and the operator selects from those which are available. If the positive
sum of a set of variables is needed, the result can be obtained as shown in
*y
By
cy

s = - ( - S ) = [- - ( A +2.05 + 5.0C)]
S = A+ 2 . 0 B + 5 . 0 C
FIG. 2 -5

Connection of Summer and Inverter to Obtain a Positive Sum.

 SUMMA TION 15

Fig. 2-5 by combining an inverter with the circuit used previously. The
same result could have been obtained by inverting A, B, and C before
summing, but this procedure would have required the use of four opera­
tional amplifiers (three inverters, and one summer) rather than the use of
one inverter and one summer. Such a reduction in the amount of com­
puting equipment required to perform a particular job can frequently be
made by proper choice of the sequence of operations. If Eq. (2-4c) was
to be mod eled:

5 = A + 2.0 * B - 5.0 * C (2-4c)

the desired result could be obtained by inverting C as shown in the upper

portion of Fig. 2-6. The block diagram in the lower portion of Fig. 2-6
accomplishes the same result with two summers rather than one summer
and two inverters.

c>
s = A + 2.05-5.0C

FIG. 2-6
Connections of a Summer and Inverters to Implement Subtraction.

In t he process of summing and inverting the variables can be multi­

plied by constants such as 0.1, 1.0, 2.0, 5.0, and 10.0. For some purposes
multiplication by a more general constant is required. Such multiplication
is considered in the next section.
16 A N AL OGS O F ARIT HMETIC OPE RA TIO NS

2-5 Multiplication by a Constant

The volume control on a radio or phonograph provides an adjustment

in the intensity of the sound that is produced. A device that is identical in
electrical function is used on analog computers to provide a continuously
adjustable multiplying factor. These devices are properly called coefficient
setting potentiometers but are frequently referred to as potentiometers,
or pots. The name potentiometer is derived from the words "potential
meter" which describes the use of such a device to measure voltage or
potentials. The use considered here is more adequately described as a
voltage divider. The block diagram symbol and the actual electrical cir­
cuit for a potentiometer are shown in Fig. 2-7. The circle represents the

HI

<> k R

ground

FIG. 2-7

Block Diagram Symbol and Electrical Connections for a Two-Terminal Poten­

tiometer.

potentiometer, the number inside the circle identifies which potenti­

ometer is used, and the constant k shows what ratio is to be set on the
potentiometer. The symbol A identifies the potential difference with
respect to ground G, the voltage which is applied to the upper or HI
terminal of the potentiometer. The symbol B identifies the voltage at the
output or ARM terminal of the potentiometer with respect to ground. As
the adjustable tap on the potentiometer is moved from the bottom to the
top, the ratio k changes from zero to 1.0. This ratio is shown on a cali­
brated dial if t he potentiometer has such a dial. In actual operation the
ratio may be less than indicated on the dial because some electrical cur­
rent flows from the ARM of the potentiometer to the input resistor on the
 MULTIP LIC ATION B Y A C ONS TAN T 17

operational amplifier. Most analog computers are equipped with a POT

BAL or POT SET mode of operation. After actuating a switch associated
with the potentiometer, the ratio on the potentiometer can be set with
reference to the ratio set on a master potentiometer or can be set by
reading the ratio directly on a digital voltmeter. See Appendix 3-A for
details. These procedures automatically compensate for any current
flowing from the ARM of the potentiometer and hence give greater
accuracy than if the potentiometers are set according to the indication of
a calibrated dial on the potentiometer. This procedure also permits the
use of less expensive potentiometers because a linear relationship between
the resistance an d mechanical rotation is no longer essential.
The potentiometers discussed thus far have two terminals, HI and
ARM, available for making connections. The ARM terminal should
never be connected to the output of an operational amplifier or to any
other source of voltage. If the ARM terminal is connected to a voltage
source and the potentiometer set for a low ratio, the potentiometer may
be overheated or damaged. Some potentiometers have the lower terminal,
marked LO, available for patching. For normal operation this terminal is
connected to ground potential. The block diagram symbol and the elec­
trical circuit for a three-terminal potentiometer is shown in Fig. 2-8. The

^ ARM

ground

FIG. 2 -8
Block Diagram Symbol and Electrical Connections for a Three-Terminal
Potentiometer.

electrical circuit on the right shows the LO terminal grounded. This con­
nection is electrically equivalent to the two-terminal potentiometer shown
in Fig. 2.7.
The use of potentiometers permits the realization of more general co­
efficients in a nalog computer modeling. Fig. 2-9 shows how the potenti­
ometers a re used to obtain ratios that are not available on the basic
summer. The resulting gain is the product of the ratio set on the potenti­
ometer a nd the gain of the input channel to which the pot is connected.
18 ANALO GS O F AR ITHM ETIC OP E RAT IONS

0.25

A>

0.75

B>

10
0.30

C>

5 = —(0.25 A + 0.755 + 3.0C)

FIG. 2-9
Block Diagram Showing Potentiometers and a Summer Used for More General
Summation.

The variables A and B are connected to inputs with a gain of 1.0 on the
summer and hence appear at the output multiplied by the negative of the
ratio set on potentiometers 01 and 02. Since the variable C feeds through
an input on the summer with a gain of 10.0, the variable C appears at the
output multiplied by —(0.3 * 10.0 ) or —3.0. P otentiometers, summers,
and inverters can be used to create analog computer models for some
problems involving algebraic equations.

2-6 Solution of Algebraic Equations

The simplest kind of algebraic equation is a variable equal to the sum of

several constants. This equation is so simple to evaluate analytically that
the analog computer is not an appropriate tool for solving it. However,
the solution of this equation provides a simple example with an obvious
answer to demonstrate how the analog computer can be used to model
algebraic processes.

EXAMPLE 2-2

Consider Eq. (2-6a):

Y = 5.0 - 3.0 (2-6a)

 SO LUT ION O F AL GE BRAIC E QUAT IONS 19

Assume that the operational amplifiers to be used have an output voltage

rating of ± 10 volts. Then an amplitude scaling of one unit per volt would
be appropriate. A voltage of 5 .0 volts is needed to represent the number
5.0 and a voltage of —3.0 vol ts is needed to represent the number —3.0.
These voltages can be obtained by multiplying the reference voltages
available on the computer by a suitable ratio using a potentiometer.
These reference voltages are available at terminals on the computer
marked + REF and — REF. Sometimes these terminals are marked with
the value of t he voltage. If the operational amplifiers are rated for ±10
volts, the reference voltages might be marked +10 and —10. Sin ce the
summer introduces an inversion, the equation should be rewritten as
Eq. (2-6b).

Y = -(-5.0 + 3.0) (2-6b)

This equation shows that a —5.0 vo lts should be summed with a +3.0
volts to obtain the variable Y. The block diagram for implementing this
equation is show n in Fig. 2-10.

-10 volts

+ 10 volts )>•

Y = -(- 5.0 + 3.0) = 5.0-3.0

FIG. 2-10

Block Diagram for a Variable Equal to the Difference of Two Constants.

EXAMPLE 2-3

Consider now Eq. (2-6c) with the independe n t variable X .

Y = 5.0 - 3.0 * X (2-6c)

The assump tion now is that a voltage analogous to the variable + is

available. The equation is modeled by the block diagram shown in Fig.
2-11. Note the change in the gain to which the ARM of POT 02 is con­
nected. Thi s change was made on the assumption that the amplitude
scaling of the variable X is one unit per volt, that is, the same amplitude
20
ANALOGS OF ARITHMETIC OPERATIONS

0.5

Y = - (- 5. 0 +3.01) = 5.0- 3.01

FIG. 2-11
Block Diagram for a Variable Equal to the Difference of a Constant and Another
Variable.

scaling as on the variable Y . For purposes of experimenting with this

circuit, a voltage is needed to represent the variable X. The first thought
might be to obtain it by means of a potentiometer from the +10 volt
reference as shown in Fig. 2-12. The ratio set on POT 03 is equal to 0. IT.

0.5

X = 10 At
Y = -(-5.0 +3.01) -5 . 0 - 3 . 01

FIG. 2-12

Unsatisfactory Connection to Supply Variable X For Equation 2-6c.

If this circuit is tried, the voltage, X , will be found to be too small because
the current flowing in the ARM of POT 03 is much too large. This diffi­
culty can be overcome by using an inverter between the two potenti­
ometers as shown in Fig. 2-13. Note the change from +10 to —10 on the
HI terminal of POT 03. This change was needed to compensate for the
sign change introduced by the inverter 02. These ideas can be applied to
more complicated circumstances. The only difficulty that may arise is a
limitation on the number of inputs to a summer. These ideas are more
useful as applied to solving a set of simultaneous linear algebraic equa-
 SOLUTION O F SIMU LTA N EOU S L INEAR E Q UATIONS 21

0.5

-10 volts )

k
-X
- i o voits y —

FIG. 2-13

Satisfactory C onnection to Supply the Variable X For Equation 2-6c.

tions than for solving single equations. The method of creating an analog
model of simul taneous equations will be considered next.

2-7 Solution of Simultaneous Linear Equations

In the period from 1940 to 1950 there was considerable interest in using
analog computers for solving sets of simultaneous equations because the
only alternative method of solution was hand computation. Today the
interest has declined because it is possible to obtain solutions more easily
and economically by means of the digital computer. This discussion of the
analog solution of s imultaneous algebraic equations is included because
the basic method is the same as that to be considered later for solving sets
of simultaneous differential equations. The process also provides an
opportunity to practice on the analog computer the techniques learned
thus far.

EXAMPLE 2-4

Consider the set of Eqs. (2-7a) and (2-7b).

5.0 * X — Y = 3.0 (2-7a)

-X+ 3.0 * Y = 5.0 (2-7 b)

The first step is much like that learned in algebra class, that is, solve
Eq. (2-7a) for the variable X .

X= 3,° + Y = 0.6 + 0.2 *Y

5 (2-7c)
22 ANAL OGS O F ARIT HME T IC OPE R AT IO N S

The next step is different because rather than substitute into Eq. (2-7b),
we solve Eq. (2-7b) for the variable Y .

Y _ 5-° + x = 1.66 + 0.333 *X (2-7d)

1 ~ 3.0

The next step is to model each of these equations on the analog computer
as shown in the block diagram of Fig. 2-14. The circuit shown in the

0.060

0.166

FIG. 2-14

Block Diagram of Analog Model for Equation 2-7c and 2-7d.

upper part of the figure would produce the variable X provided that the
negative of t he variable Y was available. The circuit shown in the lower
part of the figure would produce the variable Y if the negative of the
variable X was available. Although it may seem like "black magic," the
model can be completed by connecting these two circuits together as
shown in Fig. 2-15 with the output of AMP 01 providing the variable Jto
the input of the circuit which computes the variable Y, and conversely
the output of AMP 03 providing the variable Y needed by the circuit
which computes the variable X. An amplitude scaling of o ne unit per volt
has been assumed. After this circuit is connected on the analog computer
and the potentiometers adjusted to the values indicated on the block dia­
gram, the output of AMP 01 will be found to be 1.0 volt and the output of
 SOL UTION OF SIM ULTAN EOU S LINEAR EQUATIO NS 23

0.060

FIG. 2 -15

Completed Block Diagram for Analog Model for Equation 2-7c and 2-7d.

AMP 03 will be 2.0 volts. These values will check if substituted in the
original equations. Although these results also check if substituted into
the derived Eqs. (2-4c) and (2-4d), a check here is a necessary but sufficient
check on the correctness of the solution. If a mistake had been made in
obtaining the second set of equations, the analog computer results would
check th e second set of equations but not the original set. The process
takes on even more the aura of "black magic" if the equations are inter­
changed and the solution repeated. The block diagram for this case is
shown in Fig. 2-16. When this circuit is connected on the analog com­
puter, the output voltages of the amplifiers take on one of the two possible
extreme values and the overload signal is actuated. For operational ampli­
fiers rated for ± 10-volt outputs, the limiting values will have a magnitude
of 1 2 to 14 volts. The form of t he block diagram is the same for the two
cases, and both cases represent the same set of algebraic equations.
The reason for this seemingly strange behavior is not immediately
obvious, but some insight into this problem can be gained by considering
an an alytical solution by successive approximations. Assume that X is
zero and then solve Eq. (2-7d) for Y. Substitute this value of Y, 1.667,
into Eq . (2-7c) and solve for X. Substitute this value of X, 0.9334, into
Eq. (2-7d) and solve again for Y. This result, Y = 1.9711, when sub­
stituted into Eq. (2-7c), yields an improved value for X of 0.99422. Thus
 ANALOG S OF A RIT HMETIC OPE RAT IO NS
24

0.5

Y = 5.0^ -3.0 = -( -5.01+3.0)

1= 3.07-5.0 = - (- 3.0 y+5.0)

FIG. 2-16
Block Diagram for an Unstable Analog Model for Equation 2-7c and 2-7d.

the successive values of X and Y are seen to converge to the values of

(1.0, 2.0), which can be shown to be the correct solution. If a value ofX
greater than 1.0 is chosen as the starting point, the series of a pproxima­
tions converges from above to the same limit. If on the other hand Eq.
(2-7a) were solved for Y and Eq. (2-7b) were solved forZ, and the process
repeated as outlined above, the series of approximations will diverge
from the correct solution regardless of where the process is started unless
of course one assumes the correct answer. The demonstration of this
divergence is left for the student to explore. A sketch of a graphical solu­
tion for these simultaneous linear equations may be helpful in seeing the
process of convergence and divergence.
The analog computer model can be thought of as solving the equations
by successive approximations. If the process is convergent, a correct
solution is obtained, but if the process is divergent, no solution is possible,
and the operational amplifiers will overload. Once again it is demon­
strated that "black magic" is only a phenomenon that is imperfectly
understood by the person in whose eyes it appears to be magic.
Consider the matrix A for the coefficients of the unknowns for the set
of E qs. (2-7a) and (2-7b):
 SOLU TION OF S IMUL TANE OUS LIN EAR EQUATION S 25

A = +5 -1
-1 +3

As a general rule the analog solution and also the analytical solution by
successive approximations will be stable if the larger coefficients are
located on (or near for a larger than 2 by 2 array) the diagonal from the
upper left corner to the lower right corner, and if the equations are solved
from top to bottom for the unknowns from left to right. That is, solve
equation 1 forXi, equation 2 for unknown X2,. .., and the nth equation
for X n. A discussion of the question of the convergence of the solution for
systems of simultaneous equations from yet another point of view is given
in Appendix IV.
Another reason for the amplifiers to overload is that the solution re­
quires numbers that with the assumed amplitude scaling cannot be
represented within the ratings of the amplifiers.

EXAMPLE 2-5

Consider the following set of linear algebraic equations:

4.0x — 3.0j> = 3.0 (2-7e)
— jc + y = 4.0 (2-7f)

FIG. 2 -17

Block Diagram for Analog Computer Model for Equations 2-7e and 2-7f.
 ANALOGS OF ARITHMETIC OPERATIONS
26

These equations are solved by the same process as in the previous ex­
ample, and the resulting block diagram is shown in Fig. 2-17 The amph-
fiers will overload under these conditions because the solution to the
problem is * = 15 and y = 19 and hence the problem is not properly
scaled The same problem would arise if this problem were solved
graphically on a graph 10 in. by 10 in. if the scale were assumed to be one
unit of x or y equal to 1 in. The lines would intersect somewhere above
and to the right of the upper righthand corner of the page. We can solve
this problem either by taking a sheet of graph paper 20 in. by 20 in., or by
rescaling the problem to make 10 in. on the graph equal to 20 units of
x or y.

= max |*| = 20 units = 2 Q units/volt

d Cm 10 volts

Therefore, x and y are related to the computer variables X and Y by

Eqs. (2-7g) and (2-7h).

x = Kd*X = 2.0*X (2-7g)

y = Kd * Y = 2.0 * Y (2-7k)

0.375

FIG. 2-18

Block Diagram for Analog Computer Model of Equations 2-7i and 2-7j.
 E XERCISE S 27

These last two equations are substituted into the original set of simul­
taneous equations, Eqs. (2-7e) and Eq. (2-7f), to yield the amplitude
scaled set of equations that can be modeled on the analog computer.

4.0(2.0 * X ) - 3.0(2.0 * Y) = 3.0 (2-7i)

-(2.0 *X) + (2.0 *Y) = 4.0 (2-7j)

After these equations are processed, as in the previous examples, the

block diagram shown in Fig. 2-18 is obtained. The ARM of P OT 01 has
been connected to an input on summer 01 with a gain of 0.1. If the lowest
gain avail able is 1.0, then the ratio set on POT 01 must be 0.0375. The
accuracy of lo w ratios is not as great as for larger ratios, and, therefore,
low ratios should be avoided when possible. The analog model obtained
with the analog computer connected as shown by Fig. 2-18 yields a value
of 7.5 forX and 9.5 for Y. These values are converted by use of th e ampli­
tude scale factor Eqs. (2-7g) and (2-7h), to a value of 15.0 for x and 19.0
for y. These values satisfy the original set of Eqs. (2-7e) and (2-7f).

2-8 Exercises

1. An operational amplifier has a voltage gain of —200,000 and an input

impedance of 2 megohms. What is the input voltage and input current
when the output voltage is +8 volts?
2. The hull of a ship is 500 ft long, has a beam of 60 feet, and a height from
keel to the main deck of 80 ft. A model is to be carved from a block of
wood which is 20 in. long, 2 in. wide, and 4 in. high.
a. Compute appropriate amplitude scale factors Kd and kd. Show
complete units.
b. Compute the dimensions of the model that correspond to the 500 ft,
60 ft, and 80 ft.
3. The variable x for a problem ranges from —20 to +30 ft. The variable y
ranges from —50 to +40. Compute appropriate amplitude scale factors
kx, ^jc, and Ky.
4'. The scale factor Kx has been determined as 4 ft/volt.
a. Compute the values of the problem variable, x, which is represented
by a voltage of 6 volts.
b. Compute the value of the voltage, X, which corresponds to a value of
20 ft for x.
5. An operational amplifier is connected as shown in Fig. 2-19. What is the
output voltage when the input voltage is 2 volts?
 ANALOGS OF ARITHMETIC OPERATIONS

100,000 200,000
ohms SJ ohms

ground

FIG. 2-19

6. Sketch a block diagram representation of the analog computer circuit in

Exercise 5.
7. An operational amplifier is connected as shown in Fig. 2-20. What is the
output voltage when the input voltages are as shown.

200 K 200 K
ohms 5j ohms

ground

FIG. 2-20

8. Sketch a block diagram representation of the analog computer circuit in

Exercise 7.
9. a. Sketch an analog computer circuit diagram showing how to connect
an operational amplifier to implement the block diagram shown in
Fig. 2-21. Assume that the feedback resistor is 1 megohm,
b. What is the output voltage for an input voltage of +20 volts?

FIG. 2-21
 EXERCISES 29

10. a. Sketch an analog computer circuit diagram showing how to connect

an operational amplifier to implement the block diagram shown in
Fig. 2-22. Assume the feedback resistor to be 100 Kohms.
b. What is the output voltage D when A = 5 volts, 5 = 2 volts, and
C — — 8 volts?

FIG. 2 -22

11. Sketch an analog computer block diagram to implement the mathematical

relations given. Assume that gains of 1, 2, and 5 are available. Use the
minimum number of operational amplifiers. No potentiometers are to be
used for this problem.
a. SA = A - 25 + 5C - D
b. SB = 4A + 25 — 5C
12. Determine the mathematical relation between the output and the inputs
for each of the block diagrams in Fig. 2-23.

13. The block diagram representation of a potentiometer is shown in Fig.

2-24. What is the output voltage when the input voltage is as follows:
a. 10 volts
b. Y volts
 A N A L O G S OF AR I T H M E T I C OP E R A T I O N S
30

0.3

> - Q^—>
F I G . 2- 2 4

14. Sketch the analog computer block diagram to implement the mathe­
matical relation given. Assume that gains of 0.1,1.0, and 10.0 are available
on the inverter-summers. Use a minimum number of operational ampli­
fiers and potentiometers.
a. SA = —0.8,4 + 0.5B + 2.0C - 0.05D
b. SB = 3.0A + B - 0.15C - D
15. Determine the mathematical relation between the output and the inputs
for each of the block diagrams in Fig. 2-25.

16. Can the block diagram for part b of Exercise 15 be simplified to use fewer
amplifiers? Assume the gains of 1, 2, and 5 are available on the summer-
 EXERCISES 31

inverters. Sketch the simplified block diagram and verify that the relation
between the output voltage and the input voltage is the same as for the
original system.
17. Show all of the connections required to implement the block diagram
shown in Fig. 2-26 on one or more of the analog computers described in
Appendix I. Sketches of the panels will be found in these appendices.
Assume that the voltages X and Y are available.

FIG. 2 -26

18. Sketch the analog computer block diagrams to implement the equations
listed below. Assume that voltages to represent all variables on the right
are available.
a. P = 8.0 - 4.00
b. P = 6.0 + 5.00
19. Set up on the analog computer available to you the block diagram shown
in Fig. 2-13. If your computer has a reference voltage other than 10 volts,
use that voltage rather than 10 volts. Measure Y for the following values
of k:

k = 0.1, 0.2, 0.3, 0.4, 0.5

Compare the measured values of Y with the expected values.

20. Set up on the analog computer available to you the model described by the
block diagram of Fig. 2-15. Verify the solution given.
21. Set up on the analog computer available to you the model described by
the block diagram of Fig. 2-16 and confirm that the model is unstable.
22. Sketch the analog computer block diagram to solve the system of equa­
tions given below.

4.0X + 0.8F = 7.0

0.5X + 3.0F - 9.0

32 ANALOGS OF ARITHMETIC OPERATIONS

23. Solve algebraically the system of equations for Exercise 22. Would a
computer with a Cm of 10 volts be able to model this system adequately?
24. Set up on an analog computer a model of the system of equations given in
Exercise 22. Rescale if necessary to fit the model to your computer. Com­
pare the answers obtained with those computed in Exercise 23.
25. Determine the general relationship between Y and X for the system shown
in Fig. 2-27. For what purpose might this circuit be useful?

FIG. 2-27
CHAPTER Q_ ANALOGS OF
CALCULUS OPERATIONS

Many problems and systems of interest to engineers in­

volve rates of change with respect to time or some other independent
variable. T hus models for such systems require models for integration
and differentiation. For example, consider velocity, v, and displace­
ment, x:

v= ~ and x = Jj vdt
dt

Velocity, v, i s the derivative with respect to time of the related displace­

ment, x. Displacement, x, is the integral with respect to time of the
velocity, v. This chapter considers analog computer models for the
processes of differentiation and integration. These processes are then
applied t o show how various functions can be generated on the analog
computer.

33
34 ANALOGS OF CALCULUS OPE RATIONS

3-1 Differentiation

Consider a tank for storing water such as is shown in Fig. 3-1. Assume
that this tank is a right-circular cylinder with circular cross section located
in the horizontal plane. The amount of water, Q, stored in the tank is

Q = AH
F-
dQ

side
= A dH
dt

FIG. 3-1

Water Storage Tank Analogy to Capacitor.

proportional to the depth of the water, H. The factor of proportionality is

the cross-sectional area, A, of the tank. Thus the relation between the
quantities Q, A, and H is as follows:

Q = A* H (3-la)

If the area of the tank is expressed in square feet and the depth of the
water is expressed in feet, then the volume of water in the tank is ex­
pressed in cubic feet. The rate at which water is flowing into the tank, F,
is the derivative of Q with respect to time, that is,
 DIFFERENTIATION 35

A capacitor consists of two electrically conducting plates separated by

an in sulator. For example, two pieces of aluminum foil separated by a
sheet of paper form a capacitor. Frequently strips of foil separated by
strips of insulating material are rolled up to form capacitors. A capacitor
stores electrical charge in a manner analogous to the way in which a tank
stores water. The amount of electrical charge, Q, stored in a capacitor is
proportional to the potential difference, V, across the capacitor, that is,

Q = C*V (3-lc)

The factor of proportionality is the electrical capacitance, C. A capaci­

tance of one farad charged to a potential difference of one volt contains a
charge of one coulomb. Since typical values of capacitance are usually
very, very m uch smaller than one farad, capacitance is usually measured
in units of o ne millionth of a farad, that is, one microfarad.
The time rate at which charge flows past a point is called electric current
and is represented by the symbol I. Thus the rate of flow of c harge into a
capacitor is the current flowing into the capacitor. Electric current is
defined by the relation

/=f (3.14)

If the charge is measured in coulombs and if time is measured in seconds,

the current is measured in amperes. On the assumption that the capaci­
tance is constant, the current flowing into a capacitor is the time rate of
change of c harge in the capacitor, that is,

7 =f = c *f <3-le>

If the voltage across a capacitor is constant, the current flowing is equal to

zero, b ecause the derivative with respect to time of the voltage is zero.
See Fig. 3-2 for a summary of the relations for a capacitor as a model of

Q = CV
c
I= = C-
' dt c dt

FIG. 3-2

Electrical Capacitor as a Model for Differentiation.

36
ANALOGS OF CALCULUS OPER ATIONS

differentiation. Before considering how the capacitor can be useful as an

analog for the process of differentiation, let us consider the analogy be­
tween the capacitor and the water tank described previously.
The amount of charge stored in the capacitor is analogous to the
volume of water stored in the water tank. The potential difference across
the capacitor is analogous to the depth of water in the tank. Thus a pres­
sure gauge connected to the bottom of the tank could be calibrated to
read the depth of water in the tank. This pressure gauge would function
in a manner analogous to a voltmeter connected across the capacitor to
read its electrical pressure, or potential difference. The capacitance of the
capacitor is a measure of the number of coulombs of charge stored in the
capacitor per volt of potential difference. Capacitance is analogous to the
area of the tank that is a measure of the volume of water stored per foot
depth of water in the tank. The rate of flow of charge into the capacitor,
coulombs per second, is analogous to the rate of flow of water into the
tank, cubic feet per second. These ideas of analogy are summarized in
Table 3-1.

T A B L E 3- 1 . ANALOGY BETWEEN ELECTRICAL CAPACITOR

AND WATER TANK

CAPACITOR WATER TANK

Symbol Name Unit Symbol Name Unit

Q charge coulomb Q volume of cubic ft

water
V potential volt H depth of ft
difference water
c capacitance coulomb/volt A area of tank cubic ft/ft
or or
farad square ft
I current coulomb/sec F flow of cubic ft/sec
or water
amp

Q —C * V Q = A* H
dV
I = C* ~r
*

II

dt

The capacitor could be used as an analog of the process of differentia­

tion if a current proportional to the change of voltage was satisfactory.
 DIFFERENTIATION 37

The most desirable kind of analog, however, would be one which pro­
duced a voltage proportional to the derivative of the input voltage. By
Ohm's law the potential difference across a resistor is proportional to the
product of the resistance and the current that flows through the resistance.
The combination of a capacitor, a resistor, and an operational amplifier
as shown in Fig. 3-3 produces the desired analog model. The relationship

Vb

SJ
-AA/V

01

ground —

FIG. 3-3

Analog Computer Model for Differentiation.

between the input voltage, vi, and the output voltage, vj, can be derived
as follows assuming that the operational amplifier is ideal, that is, Vb = 0,
ib = 0.

vc = vi — V b = vi, since Vb = 0 (3-lf)

The current through the capacitor, i\, is computed by Eq. (3-le) as

„dvc „ dvi
/1 — L/» — C/ -j (3-lg)
dt dt

The current h is equal to the current, z'i, since the current into the input of
the o perational amplifier, ib, is zero for an ideal operational amplifier.
Therefore

dv i
n = li — lb = n (3-lh)
dt

The potential difference across the resistor, R f , is by Ohm's law

Vb — v2 = h * Rf (3-li)

Since the input voltage, Vb, for an ideal operational amplifier is zero, the
output voltage, vi, is as follows:
38 ANALOGS OF CALCU LUS OPE RATIONS

— V2 = +*2 * Rf
or (3-lj)
V2 = —12 * Rf

Substituting from Eq. (3-lh) the desired result is as follows:

v>=-Cd^*R,
(3-lk)
= - R f * C dvi
dt

Appendix II-C derives the relationships in greater detail to show the

effect of a nonideal operational amplifier.

EXAMPLE 3-1

Consider the analog computer circuit for differentiation shown in

Fig. 3-4. For the numbers given for the capacitor, C , and the resistor, R f ,
the Rf * C product has a numerical value of 2.0. Thus the relationship
between the output voltage, V2, and the input voltage, v\, is as follows:

v 2 = - R f * C ^ = -2.0 (3-H)

The input voltage, 01, is defined as a function of time by the upper graph
shown in Fig. 3-4. This is called a trapezoidal waveform. The derivative
with respect to time of t he voltage, v\, is plotted in the middle graph of
Fig. 3-4. During the time interval from 0 to 4 seconds, the derivative of
i?i, th at is, the slope of the graph, is as follows:

dvi (20 - 0) 20 c lA , ,
= (4^0)" = St
4~ volts/sec O-lm)

The slope of the graph for 01 in the interval from 4 to 8 seconds is equal to
zero since the function is constant. For the interval from 8 to 16 seconds
the derivative is as follows:

dv1 (-20 - 20) (-40) c 7 ,

St = (16 - 8) = ~W = "5 V°ltS/Se° (3"'n)

The remainder of the graph for dv\/dt is computed in the same way.
Eq. (3-11) relates the output voltage, 02, and the derivative of 01 with
respect to time, and thus relates the lower and middle graph of Fig. 3-4.
 DIFFERENTIATION 39

FIG. 3-4

Example of Differentiation of a Voltage.

40 ANALOGS OF CALCUL US OPERAT IONS

The reversal of sign in the process of differentiation occurs for the same
reason as in the case of the process of summation and in fact occurs with
all operations involving operational amplifiers.
Although the analog model of the process of differentiation can be
created easily as shown above, this analog is not used very frequently in
analog computation because differentiation accentuates noise. In most
cases the model can be created using only the process of integration. This
process is preferred because the averaging inherent in integration tends
to reduce noise that may be present. The creation of the electrical analog
to the process of integration also uses a capacitor-resistor network and
will be considered next.

3-2 Integration

The processes of integration and differentiation are inverses of each

other. That is, if A is the derivative of B, then B is the integral of A. Since
the current, i, through a capacitor is proportional to the derivative of the
voltage across the capacitor, then the voltage across the capacitor would
be proportional to the integral of the current through the capacitor. Since
current is the derivative of charge with respect to time,

dq
1
(3-2a)
dt

then charge, q, is related to current, i, by:

q = / idt (3-2b)

Since q = Cv by definition, then

v =2 (3-2c)

The substitution of the integral relation for q into the last equation yields
the following:

' -= i^ fI idt (3-2c)

These relations are summarized in Fig. 3-5. This analog model for integra­
tion has a disadvantage similar to that of the simple capacitor circuit for
 INTEGRATION 41

. _ dq
' ~ ~dt
q = J idt
Vc

-1 'c 5«"
•• =

f i g . 3-5

Electrical Capacitor as an Integrator.

differentiation, that is, a voltage is proportional to the integral of a

current. The most desirable process would produce an output voltage
that was proportional to the integral of the input voltage. The combina­
tion of a capacitor, a resistor, and an operational amplifier as shown in
Fig. 3-6 produces the desired result. The validity of t his circuit will now
be demonstrated by the following derivation.

ground -=r-

RC

f i g . 3-6

Analog Computer Model of Integration.

This derivation assumes that the operational amplifier is ideal, that is,
the input voltage, Vb, a nd the input current, h, are small enough to be
considered to be zero. Refer to Fig. 3-6 for the circuit diagram. The input
current, zT, an d the output current, ii, are as follows:

. _ f(Pi - Vb)1 _ vi
L R J
R
(3-2e)
42 ANALOGS OF CALCULU S OPERATI ONS

Vi
12 = l\ ~ lb ~ ll = ~jZ (3-20

The potential difference across the capacitor, vc, is as follows:

vc = vb — V2 (3-2g)

Since the voltage, Vb, is negligible compared to the voltage, V2, the output
voltage, V2, is a s follows:

(3-2h)

This result can be remembered as follows:

1. the input voltage, i?i , divid ed by the input resistor, R, determines the
current that is flowing,
2. this current is integrated with respec t to time to determine the cha rge
on the capacitor,
3. this charge divid ed by the value of capacitan ce, C, gives the value of
output voltage.

The block diagram symbol for the analog model of integration is also
shown in Fig. 3-6. The difference between this symbol and that for the
inverter is the addition of the rectangular box on the left side of the sym­
bol. The number 01 identifies which operational amplifier is to be con­
nected as an integrator. The number represented by the symbol k is the
reciprocal of the RC product of the integrator. This constant is called the
gain of the integrator. When the capacitance is measured in microfarads
(10~6 farad) and the resistance is measured in megohms (106 ohms), the
RC product has units of f arad-ohms or seconds. The analog model for
integration has associated with it the sign inversion seen previously for
the analog models of the process of s ummation and differentiation. Al­
though the sign inversion for these processes could be eliminated by in­
cluding an inverter with each of the analogs, this addition would need­
lessly increase the cost of t he equipment. As will be seen later, the sign
inversion is taken into account in setting up the problem and an inverter
is added only if needed.
 INTEGRATION 43

EXAMPLE 3-2

Consider the analog integrator set up as shown in Fig. 3-7, which is

integrating the voltage, v\, which has the waveform shown in the upper
graph. The second graph from the top shows the result of integrating the
voltage, vi. The integral of a function can be considered to be the area
under the graph of that function. Thus the area between the graph for vi
and the ?-axis is -10 volts per sec, and the slope of the graph for the
integral is —10 volt-sec per sec over the interval from 0 to 4 sec. The value
of the integral at 4 sec is the area between the graph for vi and the f-axis
over the interval from 0 to 4 sec. This area is -40 volt-sec. Since the
voltage, vi, is zero over the interval from 4 to 8 sec, the value of the in­
tegral is constant at -40 volt-sec during this interval. In the interval
between 8 and 16 sec, the area under the vi curve is +10 volts per sec,
and the slope of th e graph for the integral of v\ is +10 volt-sec per sec.
The area under the v\ curve over the interval from 8 to 12 sec is +10 volts
* 4 sec or 40 volt-sec. Thus the value of the integral at time equal to 12 sec
is zero. After 4 more sec the value of the integral is +40 volt-sec, and the
process continues in similar fashion to complete the curve. The output
voltage from the integrator is related to the integral of the input voltage
by the reciprocal of t he RC product for the integrator and the sign in­
version common to all analog operations with operational amplifiers. The
output voltage vi is shown on the lower graph of Fig. 3-7. Compare the
voltages vi and V2 shown in Fig. 3-4 and Fig. 3-7. The voltage v\ to be
integrated in Fig. 3-7 is the same as the voltage vi obtained by differen­
tiating the voltage vi in Fig. 3-4. Furthermore, the output voltage V2 of
Fig. 3-7 is the same as the input voltage vi of Fig. 3-4. These relations be­
tween the input and output voltages demonstrate the inverse nature of
the operations of integration and differentiation.
Frequently it is desirable to integrate the sum of two or more variables.
This could be accomplished by first summing the variables with a summer
and then integrating the result with an integrator. The two processes can,
however, be combined and performed by one operational amplifier.
Since problems may be frequently large enough to use all of the com­
puting equipment that is available, it is appropriate to economize wher­
ever possible. Such a combination is shown in Fig. 3-8. The result can be
inferred by considering each input separately and then adding up the
result. The process suggested is an example of the principle of super­
position. The principle of superposition states that the net effect of several
independent causes is the same as the sum of the effects of each cause
44 ANALOGS OF CALCU LUS OPERA TIONS

1. 0 X 10 6 ohms Vb 2.0 X 10 - 6 farads

f i g . 3-7

Example of Integration of a Voltage.

 INTEGRATION 45

*t SJ c
> wv

Ri B 0
v2> WV

vo — ~

f i g . 3-8

Combined Summation and Integration.

acting separately. For example, consider a scale for measuring the weight
of p eople. If one person weighs 125 pounds and a second person weighs
150 pounds, the scale registers 275 pounds if both are on the scale at the
same time. The block diagram for the combined operation of summation
and integration is shown in Fig. 3-9.

-J(£,V, + k 2 v 2 ) d t

f i g . 3-9

Block Diagram for Combined Summation and Integration.

The integrals that we have been considering are not complete. In

calculus you learned that the integral of 6x2 is as follows:

J 6x2dx — 2x 3 + IC (3-2i)

The symbol IC stands for the initial value of the integral. This initial value
will be a constant and its derivative will be zero. Thus the integral of 6x2
could be 2x3, 2x3 + 3, or 2x3 + any constant. The derivative of all these
functions is 6x2. Thus the analog model may also require provision for an
initial condition. An initial value for the output of a n analog integrator
requires that the capacitor be charged initially to a suitable voltage. This
is accomplished by connecting a suitable voltage to the terminal on the
integrator marked IC. The nature of the process of o btaining the initial
46 ANALOGS OF CALCU LUS OPER ATIONS

condition involves the sign change common to all operations using opera­
tional amplifiers. Thus the initial condition obtained by connecting a
voltage +A to the IC terminal is — A. The block diagram for an integrator
with an initial condition connected is shown in Fig. 3-10. If no initial con-

Vo = -(v/c + kt f v,dt)
Vic = ~ vo(0)

f i g . 3-10

Block D iagram for Integration with Initial Condition.

dition is desired, the connection is omitted on both the block diagram and
in the connections made on the analog computer. The initial condition is
then zero. The value of voltage needed for an initial condition may not be
directly available on the computer. In this case a potentiometer can be
used to attenuate or reduce the reference voltage to provide the desired
value. The coefficient required on the integration process may be a value
other than the standard gains available on the patch panel. In this circum­
stance the potentiometer is used to obtain the desired value. The general
case of integration with an arbitrary gain and initial condition is shown in
Fig. 3-11. A specific example of this general case will be used in the next
section to generate a ramp function.

<-R
0
a IC

V,>
0 k

»'o = - ( ~ a 2 R +*1«I $ V i d t )
= a2 R — k i a l ^ v x d t

f i g . 3-11

Block Diagram for Integration with General Constants.

 GENERATION OF RAMP FUN CTIONS 47

3-3 Generation of Ramp Functions

The simplest function after constants are linear functions. The graph of
a linear function is shown in Fig. 3-12 in the familiar "slope-intercept"
form. The slope m is the ratio of Ay to Ax, and the intercept is the value
of b. Because the graph of such functions looks similar to ramps used to
move from one level to another, these functions are called ramp functions.

Ax

f i g . 3-12

Graph of Linear Function or Ramp.

Consider the block diagram shown in Fig. 3-13. Potentiometer 01 at­

tenuates the —10 volt reference to —0.5 volt. Since the input to an inte­
grator is the derivative of the output of the integrator, the output will

0.5

f —10 vo lts

0.05 IC

—10 volts )>-

f i g . 3-13

Ramp Function Generator.

48 ANALOGS OF CAL CULUS OPERAT IONS

change at the rate of 0.5 volt per second. Thus the ramp generated will
have a slope of —0.5 volt/second. Potentiometer 02 attenuates the —10
volts to —5.0 volts, and the inversion makes the initial value of the out­
put voltage equal to + 5.0 volts. The output voltage can be calculated
as follows:
t>o = -[(-10.0 * 0.5) + (1.0 * 0.05 J -1 0.0^0]
v0 = 5.0 + 0.5 J dt = 5.0 + 0.5/ (3-3)

The result is plotted in Fig. 3-14. Since the analog computer is assumed to
start calculating at time equal to zero, there is no curve to the left of the
origin. The limited range of t he output voltage of th e operational ampli­
fier provides a limit on the magnitude of the function that can be gen­
erated. This limit was assumed to be 10 volts for the present example.

f i g . 3-14

Ramp Function Generated by Model Shown in Fig. 3-13.

The usual process of interest is not to compute the function that will be
generated by a given analog computer configuration, but rather, given the
function that one wishes to generate, to create the analog computer
configuration that will do the job. The former process is known as
analysis, and the latter process is known as synthesis. The usual activity
of engineers is that of creating or synthesizing a system or device to per­
form a specified j ob. The analog computer is a useful tool for synthesis
because models can be easily created on the analog computer and tested
to see whether the performance is satisfactory. Consider the problem of
generating the function shown in Fig. 3-15. The slope is positive and has a
 GENERATION OF POLYNOMIAL FUNCTIONS 49

f i g . 3-15

Ramp Function Desired.

value of (18/50) or 0.36 volt per second. Therefore, the input voltage
should be —0.36 volt, which can be obtained by potentiometer 01 set for
0.036 and connected to —10 v olts. The initial condition is —9.0 volts,
which can be obtained by connecting +9.0 volts to the IC terminal. This
is obtained from potentiometer 02 set for a ratio of 0.90 and connected to
+ 10 volts. See Fig. 3-16 for the block diagram of the analog model to
generate the desired ramp function. Ramp functions are frequently
needed in analog computer models to provide a voltage that is a linear
function of time to drive one axis of an X-Y plotter or other display device.
This application will be discussed in the next chapter.

0. 9

f i g . 3-16

Block Diagram of Analog Model for Generating Ramp Function of Fig. 3-15.

3-4 Generation of Polynomial Functions

Consider the integration of a linear function y such as generated in the

last section:
50 analogs of calculus operations

z =J ydx + C — J [mx + b\dx + C

mx2 . , , n
c
z = —2— + ^

z = ax2 + ^ + C (3-4a)

Thus the integration of a linear function is a quadratic function. This

process can be repeated as many times as desired to obtain any desired

<D

ic IC

e > 01 01 -» 02 02 ">vo
k-,

f i g . 3-17

Block Diagram for Generation of a Quadratic Function.

polynomial function. Consider the block diagram shown in Fig. 3-17.

The output V2 of amplifier 01 is by application of the principles of
Section 3-2.

V2 = —(D + hai J Edt)

V2 = —D - [Ekiai]t (3-4b)

The output voltage v0 of a mplifier 02 is the integral of the voltage v2 as

follows:

vo = —[d- C + k2ci2 J v2dt\

= -C - k2a2 f [—D - Ek\a\t\dt

„, , , Ekik2aia2t2
= — C + k2a2Dt 4 2

vo = — C 4- [k2a2D]t 4- [0.5 Ek\k2aia2]t2

V o = - C + Bt+ At2 (3-4c)

 GENERATION OF POLYNO MIAL FUNCTIO NS 51

Suppose that it is desired to generate the polynomial below:

v0 = -10 + 2.01 - 0.05/2 (3-4d)

A comparison of Eq s. (3-4c) and (3-4d) shows that:

C = +10.0

The product of the constants k2, a2, and D m ust be 2.0. If D is chosen to
be +10 and k2 chosen to be 1.0, then a2 is 0.20, which is an appropriate
value to set on potentiometer 02. The value of 10 for D is appropriate if
the reference voltage is 10 volts. The value of 1.0 is generally an appropri­
ate gain for an integrator. From Eq. (3-4c) it can be found that:

— 0.05 = A = (0.5[E ki ai\ k2 a2) (3-4e)

Since k2a2 = 0.2,

Therefore: kicnE = —0.5.

Assume: E = —10

Therefore: k\a\ = +0.05

Assume: ki = 1.0
Therefore: a\ = 0.05

The block diagram for the analog computer model to generate the
polynomial of Eq. (3-4d) is shown in Fig. 3-18. A graph of the voltages vo
and v2 plotted as a function of time is shown in Fig. 3-19.

f i g . 3-18

Block Diagram for Generation of v 0 = —10 + 2.01 — 0.05/.2

f i g . 3-19

Graphs Produced by Analog Computer Model of Fig. 3-18.

 GENERATION OF SINUSOIDAL FUNCTIONS 53

3-5 Generation of Sinusoidal Functions

The sine and cosine functions have an interesting property in that re­
peated integrations will produce the function with which one started
except for a multiplicative constant. For example:

J a; sin {wt)dt — J sin (cot)d(wt) = —cos (cot)

J w cos (cot)dt = J cos (ut)d(cot) = sin (coO
The constants, GO, were added to facilitate the integrations. The implemen­
tation of the integration of the cosine function on an analog computer is
shown in Fig. 3-20. Because of the inherent inversion (sign change) the
resulting voltage V2 is a negative sine function.

CO

f i g . 3-20

Integration of a Cosine Function.

Next we will consider the integration of the negative sine function, since
we wish to make an iterative type operation. See Fig. 3-21. Again the in­
put voltage, V2, has been multiplied by the constant co to facilitate the
integration. The result, V3, is a negative cosine function. The inverter 03
has been added to produce a positive cosine function.

CO

f i g . 3-21

Integration of a Sine Function.

The analog models shown in Figs. 3-20 and 3-21 can be combined to
form a sinusoidal function generator. Before reading on, the student
should try to put the two parts together to form this generator.
The terminals marked V2 are connected together and the terminals
marked and v\ are connected together. The complete system is shown
va
54 ANALOGS OF CALCULU S OPERATION S

f i g . 3-22

Sinusoidal Function Generator.

in Fig. 3-22. The feature added here that was not in the preceding two
figures are the initial conditions A and B. If the initial conditions are both
zero, nothing is generated. If A is zero and B is —10 volts, the voltage, 113,
and the voltage, V2, will be as follows:

V3 = 10.0 * cos (ut)

V2 = 10.0 * sin (a>0

Although the reason for these results is not obvious, the correctness of the
statements can be easily verified by successive integrations and considera­
tion of the sign inversion of the analog integrators. In similar fashion the
result for initial conditions of A = —10.0 volts and B = zero are as
follows:
V2 — +10 .0 * cos (cat)
V3 = —10.0 * sin (a>t)

Sinusoidal generators such as described are very useful for analog com­
puter modeling because the value of the function at the start of the
computation can be adjusted and can be depended upon to be repeatable
for a fixed value of adjustment. Most signal generators, such as those
used in electronics laboratories, cannot be started in this fashion. The
conditions at the start would vary in a random fashion.

3-6 Exercises

Note: The real problems that an engineer or other person solves in practice are
not usually like the typical textbook problems that have just the right amount of
information and one unique answer. One of the first steps in solving such real
problems is to sort out the relevant material from that which is irrelevant. On
 EXERCISES 55

occasion the information provided is not sufficient and certain assumptions must
be made. If it is necessary to make assumptions, they should be clearly labeled as
such. Some of the problems in this book will have too much or too little informa­
tion. The student should therefore be alert.

1. A water tank is a rectangular parallelopiped 100 feet high with a base

which is 20 ft by 30 ft. Compute the rate at which water is flowing into
the tank when the rate of change of height is 2 ft/min.
2. Water is 20 ft deep in the tank of Exercise 1 at time equal to zero. Water
flows into the tank at a constant rate of 1000 cu ft/min. Compute the rate
at which the height is changing at:
a. time equals 30 minutes
b. time equals 90 minutes
3. Compute the depth of the water for the conditions of Exercise 2 at
a. time equals 20 minutes
b. time equals 80 minutes.
4. Does an answer to Exercise 3 suggest a possible error in an answer to
Exercise 2? If so, what is the error? If this problem were modeled on an
analog computer, what would occur at the problem time equal to 80
minutes?
5. a. What is the relationship between the output voltage and the input
voltage for the analog computer circuit shown in Fig. 3-23?
b. Sketch the block diagram representation of this analog computer
circuit.

2 X 10 5 ohms 10 X 1 0 - 6 farad

f i g . 3-23

6. a. What is the relationship between the output voltage, V2, and the in­
put voltage, vi, for the analog computer circuit represented by the
block diagram shown in Fig. 3-24?
b. Sketch the analog computer circuit represented by this block diagram.
Assume that the feedback capacitor has a value of 1 X 10~6 farad.

FIG. 3-24
56 ANALOGS OF CALC ULUS OPE RATIONS

volts

0 1 5 6 10 11 15 16 t - sec

f i g . 3-25

7. a. The voltage vi has the waveform shown in Fig. 3-25. This voltage is
operated on by an analog computer circuit represented by the block
diagram shown below. Sketch the output voltage V2 ove r the time
interval 0 to 10 seconds,
b. Assume that the operational amplifier used is rated for operation over
a range of output voltage from —10 to +10. Does the 20 volt input
signal create any problems? Why or why not? How long can this
system operate before problems will arise? What is the problem?
8. a. What is the relationship between the input voltage and the output
voltage for the analog computer circuit represented by the block dia­
gram shown in Fig. 3-26?
b. Sketch the analog computer circuit represented by this block diagram.
Assume that the feedback capacitor has a value of 10 X 10~6 farad.

f i g . 3-26

9. What is the relationship between the output voltage and the input voltage
for the analog computer circuit represented by the block diagram shown
in Fig. 3-27?

0.2

FIG. 3-27
 EXERCISES 57

10. What is the mathematical expression for the voltage vz i n the analog
computer circuit represented by the block diagram shown in Fig. 3-28?

0.7

f i g . 3-28

11. A voltage waveform as shown in Fig. 3-29 is desired. Sketch the block
diagram representation of the analog computer circuit that will generate
this waveform. Assume that the reference voltage on the computer to be
used is 100 volts. Assume that the integrators to be used have gains of
1, 2, and 5.

f i g . 3-29

12. What is the mathematical expression for the output voltage, vz, in the
analog computer circuit represented by the block diagram shown in
Fig. 3-30?
 ANALOGS OF CALC ULUS OPERA TIONS

<^+10 volts

va = 6 +2t
Vb = 3r-4r
f i g . 3-30

13. What is the mathematical expression for the output voltage, v2 in the
araloyomputer circuit represented by the block diagram shown in

0.3
0.5

+ 10 volts
— 10 volts
0.2 JC
0.4 IC
— 10 volts" 08
09

f i g . 3-31

14. What is the mathematical expression for the output voltage, v2, in the
Fig 3S32?mPUter CirCUlt represented by the block diagram shown iin

v a = I0e~ 2 1 )>.

Vb — — io voits y-

FIG. 3-32
 EXERCISES 59

15. Sketch the block diagram representation of the analog computer circuit
which will generate the given function. Assume that the analog computer
used has a reference voltage of 100 volts, and that the integrators used
have gains of 1, 2, and 5.

[72 = 30 - 41 + 212

16. a. What is the general mathematical expression for the voltage vi

generated by the analog computer circuit represented by the block
diagram shown in Fig. 3-33?

f i g . 3-33

b. What is the specific mathematical expression for the voltage V2

generated for each set of potentiometer ratio listed below?

POTENTIOMETER 01 02 03 04

CASE 1 0.20 0.80 0.20 0.00

CASE 2 0.20 0.00 0.20 0.60
CASE 3 0.20 0.80 0.20 0.60
CASE 4 0.40 0.80 0.40 0.60

17. Sketch the block diagram representation for the analog computer circuit
that will generate each of the voltages listed below. Assume that the
reference voltage on the computer being used is 10 volts and that the
integrator gains are 1 and 10.
a. 5 cos (21)
b. 8 sin (0.5r)
18. Show by means of detailed sketches how one or more of the analog com­
puters described in Appendix 1 would be connected to implement the
block diagrams shown in Fig. 3-34.
 ANALOGS OF CALCU LUS OPE RATIONS

03 ) < + ref

01 IC
C
1 01 02

10
02

(«)

02 ref 04 ref

IC

01 03 02 03

(•b)

f i g . 3-34

a. Create an appropriate mathematical model for the system shown in

Fig. 3-35, and show by means of a block diagram how this can be
implemented on an analog computer. Assume that the analog com­
puter to be used has a reference voltage of 100 volts and amplifier
gains of 1, 2, and 5. The variable of interest is the height, h(t), of the
water in the tank.
b. Since the rates at which the water flows into the tank, qm, and out of
the tank, qoxlt, are constant and not equal, the tank will either run dry,
or overflow. Which of these happens and at what time does it happen?

pump

qin = 60 cu ft/sec

q^ ou t, = 50cu ft/sec
'

50 feet h{0) = 30 ft
A = 40 sq ft

f i g . 3-35
chapter
RECORDING AND DISPLAY
OF SOLUTIONS

The answer produced by an analog computer is usually a

nonconstant function of the independent variable of the problem. Such
an answer is most easily interpreted when plotted as a graph showing the
functional relationship between the variables. Three of the most com­
monly used methods of presenting the answer are cathode ray oscillo­
scopes, strip chart recorders, and electromechanical X-Y plotters. The
human ear is less sensitive to sounds as the frequency increases, and
ultimately the ear has no response for sufficiently high frequencies. In the
same way, these recorders and display devices are limited with regard to
the highest frequency of electrical signal that they can handle without
excessive error. The cathode ray oscilloscope can reproduce signals of
much higher frequency (0.5 megacycle and higher) than are produced by
analog computers. This is true even when the analog computer is operat­
ing in the REP-OP mode, which increases the frequencies by a factor of
500 to 1000. The usual strip chart recorders can record signals with fre­

61
 62 RECORDING AND DISPLAY OF SOLUTIONS

quencies up to 100 cycles per second. The electromechanical X-Y plotter

can record signals up to a few tenths of a cycle per second. For signals
with frequencies less than a few tenths of a cycle per second the X-Y
plotter is capable of the greatest accuracy. Such plotters may have an
accuracy of 0.05 to 0.1 per cent. The strip chart recorder and the oscillo­
scope are useful where the signals have higher frequencies and accuracy
of the order of 1.0 per cent is adequate.

4-1 Strip Chart Recorder

These recorders have provision for moving a strip of chart paper under
one or more pens at one of several fixed speeds. The motion of the pen is
perpendicular to the direction of motion of the paper. In some cases the
tip of the pen moves in a circular arc, rather than a straight line. Since in
this case the ordinate lines on the chart paper are also arcs of t he same
radius, the coordinates on the graphs can be r ead in spite of this distor­
tion. See Fig. 4-2 associated with Exercise 1. The deflection of the pen is
proportional to the voltage applied to the terminals of the recorder.
Therefore, the trace on the paper is a graph of the voltage applied to the
terminals as a function of time. If a recorder has more than one pen, then
several different variables can be recorded simultaneously, and the values
of these variables can be conveniently compared. These recording systems
usually have amplifiers with an attenuator calibrated in volts/cm. The
setting of this control should always be written on the chart to permit the
graph to be interpreted in terms of computer voltage levels. For example,
if the control were set for 2 volts/cm, then a distance on the chart of 3 cm
corresponds to a voltage of 6 volts. There is usually a mechanical center­
ing control on each pen, and the pen should be centered with zero volts
applied to the recorder. There is also an electrical positioning control that
can be used to set the pen off ce nter with zero volts applied to the ter­
minals of the recording system. After the pen has been properly po­
sitioned, the reference voltage of the computer should be applied to check
on the correctness of the calibration. For example, if the attenuator is set
for 2 volts/cm, then a computer reference voltage of 10 volts would cause
a deflection of 5 cm. The setting of the speed control on the strip chart
recorder should also be written on the chart to permit interpreting the
time axis properly in seconds per cm.

4-2 X-Y Plotters

One of the disadvantages of the strip chart recorders is the limited
chart width available, usually 4 cm. X-Y plotters are designed for standard
 X-Y PLOTTERS 63

graph paper such as 8| by 11 in. or 10 by 15 in. Larger sizes such as 30 by

30 in. are also sometimes used. The position of t he pen is controlled by
two electric motors that cause the pen to move parallel to each axis. The
position of the pen along an axis is proportional to the voltage applied at
the input terminals for that axis. The factor of proportionality is deter­
mined by the attenuator control for that axis. Suppose that the controls
for both theX and the F axes are set for one volt per inch. A voltage of 5
volts applied to the X-axis a nd 8 volts applied simultaneously to the F-
axis would cause the pen to move to the position (5, 8) if the zero position
were at the origin of the plot. The position of the pen with zero volts
applied to the input terminals can be adjusted to any position desired by
means of centering controls. These controls can also be used to draw the
axes of the graph directly on the graph paper. This method of drawing the
axes checks the correctness of th e alignment of the paper on the plotting
table. The procedure is as follows:

a. With the pen lifted off the paper, center the pen at the position desired
for the origin.
b. Lower the pen and cause it to move along the X-axis by changing the
X-axis centering control.
c. Return the pen to the origin and repeat with the F-axis control to draw
the F-axis. Return the pen to the origin and proceed with recording the
solutions.

The setting for the attenuator controls for each axis should be written on
the graph paper to permit proper labeling of the axes with voltages and
problem units. Suppose that the F-axis attenuator was set for 2 volts per
in. and the scale factor for the y variable was 5 ft per volt. Then 1 in. on
the F-axis equals a distance of 10 ft in terms of the problem variable
plotted on the F-axis. The scale factor on the F-axis is 10 ft/in.
The independent variable in many problems solved on the analog
computer is time. Thus a voltage proportional to time is needed to drive
one axis of the plotter. This voltage can be obtained by integrating a
constant voltage as discussed in Section 3-3. Suppose that the reference
voltage on the computer being used is 10 volts and that the solution is to
be plotted over a time interval of 20 seconds. The voltage on the time axis
of the recorder should change at the rate of 10 volts in 20 sec or 0.5 volt
per sec. The attenuator on the X-axis of the recorder should be set for
1.0 volt/in. assuming that the X-axis is 10 in. long. The circuit shown in
Fig. 4-1 will provide the proper operation of the pen on the X-axis of the
recorder. It has been assumed that the pen is located at the lefthand side
of the paper when zero volts is applied to the recorder. When the com­
puter is switched to OPERATE, the time-base integrator 01 begins
 64 RECORDING AND DISPLAY OF SOLUTIONS

0.05 1 v olt/in.
X-Y
—10 volts
>-© 0.5 volts
plotter

f i g . 4-1

Generation of Time Base for X-Y Plotter.

integrating and the pen moves to the right at the rate of 0.5 in. per sec.
The X-axis would be marked to a scale of 2.0 sec per in.
Some plotters have an automatic pen lift connected to the analog
computer, and others have a manually operated pen lift. In the latter case,
the pen must be lowered before the computer is placed in OPERATE.
When the solution has been completed, the computer should be placed in
HOLD. After the pen has been lifted, the computer can be switched to
RESET, and the pen will return to the origin. If the computer is switched
directly to RESET, the pen will trace an unwanted line across the graph
paper. If the automatic pen lift feature is available, one should pause
briefly in the HOLD mode to give the pen lift time to function.
SomeX-F plotters have a built in time-base generator. The rate at which
the pen moves on the T-axis is controlled by a knob calibrated in in. per
sec. The desired rate is selected by positioning the knob. This time-base
generator must be placed in operation automatically by the computer
MODE control so that the plotter will start when the problem starts.
This synchronization is accomplished by connecting a special cable be­
tween the plotter and the computer.
On the patch panel of some computers is a special set of terminals that
connect the plotting equipment. The voltages to be recorded are patched
into these terminals. A special cable between the plotter and the analog
computer completes the connection. On other computers the connections
are made directly from the appropriate amplifier outputs on the patch
panel to the recorder input terminals. It is appropriate also with theX-F
plotters to check the calibration of the plotter by connecting the reference
voltages to the plotter input terminals and observing that the proper de­
flection is obtained.

4-3 Cathode Ray Oscilloscope

The cathode ray oscilloscope is usually referred to by the shorter name,

oscilloscope. The general procedure for using the oscilloscope to display
computer results is similar to that for the X-Y plotter. The deflection of
 EXERCISES 65

the spot on the screen is proportional to the voltage applied to the input
terminals. The voltage applied to theZ-input moves the spot horizontally,
and the voltage applied to the F-input moves the beam vertically. Assume
that the attenuators are set for 2 volts per cm. If a voltage of 5 volts is
applied to theX-input and a voltage of 10 volts is applied to the F-input,
the spot will move 2.5 cm to the right and 5 cm up from the initial posi­
tion. The same comments made above for the X-Y plotter apply to es­
tablishing the proper scales on the display in terms of problem variables.
The oscilloscope may have a built in time-base generator to provide the
voltage to deflect the spot linearly with time on the horizontal axis. This
sweep must be started, or triggered, by a proper voltage from the analog
computer to the trigger input on the oscilloscope. Frequently the oscillo­
scope will be used to check out the setup of the problem with the com­
puter operating in the REP-OP mode, and then the computer will be run
in the regular mode while the solution is being recorded by the X-Y
plotter. In this case it is desirable to generate the time base for the oscillo­
scope presentation on the analog computer because the same time-base
generator can then be used to provide the time voltage for theX-Y plotter.
The oscilloscope also has controls for centering the spot at the desired
location when zero voltage is applied to the input terminals. There also
will be controls for brightness (intensity) of the spot and for focus of the
spot. Some oscilloscopes have two or more input channels on the vertical
axis. The spot may make one trace for the first signal, and a second trace
for the next signal, and then repeat the sequence. This switching is per­
formed automatically in an ALTERNATE sweep mode of operation.
There may also be a CHOPPED mode in which the two inputs are
switched alternately onto the spot. This switching is done at a high
enough rate so that the two traces appear to be continuous.

4-4 Exercises

1. The computer variables X and Y were recorded on a strip-chart recorder.

The calibration on the X, Y, and T axes is shown. The amplitude scale
factors on X a nd Y are given below. Determine the values of the problem
variables x and y for times t — 2, 6, and 12 sec. The strip-chart record is
shown in Fig. 4-2.

Kx = 20 ft/volt

Ky = 5 ft/volt
cS
X5
u
a
•c
 EXERCISES 67

2. The computer variables P and Q were recorded on an X-Y plotter as a

function of time as shown in Fig. 4-4. The calibration is shown on the
axes. The ramp function to drive the time axes is shown in Fig. 4-3. The
amplitude scale factors are also listed below. Determine values of the
problem variables p and q for times t — 5, 8, and 10 sec.

Kp = 25 gal/volt

Kq = 2 gal/sec-volt

X- Y plotter

•10 vo lts^>

f i g . 4-3

3. Select appropriate gains and attenuator settings for generating a 0 to 30-sec

time base for the 15-inch axis of a 10 by 15 inch X-Y plotter. Assume that
the computer being used has reference voltages of ±100 volts; that the
amplifiers have gains of 1, 2, 5; and that the X-Y plotter has attenuator
settings of 1, 2, 5, 10, 20, 50 volts/in. A positive voltage moves the pen to
the right. See Fig. 4-5.
4. Specify appropriate settings for the F-axis attenuator on theT-F plotter of
Exercise 3 above to record the following signals. Where would the pen be
positioned for F = 0? All limits are given in volts.
a. 0 < F < 50
b. 0 < F < 100
c. — 30 < F < +75
d. — 50 < F < +25
5. A cathode ray oscilloscope was used to make the record shown in Fig. 4-6.
The block diagram and attenuator settings are as shown. The amplitude
scale factors are as shown below. Determine the values of the problem
variable y at t = 2, 6, and 15 sec.
 EXERCISES 69

oscilloscope

f i g . 4-6

Cathode Ray Oscilloscope Recording for Exercise 5.

chapter
.Jm
FIRST-ORDER SYSTEMS

The word SYSTEM was used in the title of this chapter

to indicate generality. Suitable adjectives will be used as appropriate
to limit the scope in particular applications. A system is as simple or as
complex as needed. For example, the life support system for a manned
spacecraft is very complex. Complex systems may usually be broken
down into smaller, simpler subsystems. For example, the life support
system for the spacecraft may have a temperature regulating system,
a CO2 regulating system, a humidity regulating system, and probably
many others. By way of contrast, the brakes of an automobile are a
relatively simple system. The words "first-order" in the title limit the
scope of this chapter to systems in which no derivatives of higher order
than the first appear in the mathematical relations. Second-order and
higher-order systems are the subject of Chapter 6. A very strong thread
of similarity runs through all of analog simulation regardless of the
subject matter of the problem. The analog computer model for a hydrau-

70
 CREATION OF ANA LOG MODELS OF DYNAMIC SYSTEMS 71

lies problem may have the same form as the analog computer model for
cooling of a hot body. This of course results from the similarity of the
mathematical models. Examples will be chosen from thermodynamics,
dynamics, hydraulics, transient electrical circuits, and chemical rate
dynamics.
The analog computer cannot make up for any lack of understanding
of the problem on the part of the user. If the user creates a good mathe­
matical model and implements it correctly on the analog computer, the
resulting analog model will be a good model. A poor mathematical
model, even if co rrectly implemented on the analog computer, will still
result in a poor analog model. Needless to say, a correct mathematical
model incorrectly implemented will result in an erroneous analog model.
Careful checking at all stages of the process of analog simulation is
necessary to insure the validity of the results that will b e obtained.

5-1 Creation of Analog Models of Dynamic Systems

Chapter 2 was concerned with the creation of analog models of static

systems. The procedures for amplitude scaling and for arithmetic opera­
tions developed in Chapter 2 together with the procedures for integration
from Chapter 3 form the basis for creating models of dynamic systems.
A dynamic system is one in which the rates of change of quantities, that
is, their derivative or derivatives, are important in addition to the values
of these quantities. For example, the question of determining the max­
imum speed of an automobile is a static problem. It is a static problem
because the force developed by the engine to cause the car to move can
be related to the speed of the automobile by an algebraic equation. Like­
wise an algebraic expression relates the drag force which opposes the
motion of the car to the speed at which it moves. The maximum speed
of the automobile will be that speed for which the force developed by
the engine to propel the car is equal to the drag force opposing the
motion of the automobile. The determination of the intersection of
curves of drag force and engine force is an algebraic problem, and hence
a static model is adequate for this problem. If the question was to de­
termine how long a time is required for the automobile to accelerate
from 10 to 50 miles per hour, the problem requires a dynamic model
because in addition to the speed of the car, the rate of change of speed,
or acceleration is important. Generally the analog computer is more
useful as a tool for modeling dynamic systems than for modeling static
systems.
72 FIRST-ORDER SYSTEMS

The first step in creating an analog model is to create the mathematical

model for the system. The mathematical model is the collection of mathe­
matical relations between the variables which result from physical laws
and the relationships which result from the structure of th e system. The
relation between the rate at which heat is transferred across a boundary
and the temperature difference across the boundary results from a phy­
sical law. The fact that in a particular problem the temperature on one
side of the boundary is constant results from the structure of the system.
The second step in creating an analog model is to suitably scale the
mathematical model so that the scaled model will fit within the limita­
tions of t he analog computer being used. This scaling will depe nd upon
the specifications of the particular analog computer being used. A prob­
lem may require either amplitude scaling or time scaling, or may require
both amplitude and time scaling. Some procedures for scaling operate
in a way such that these two scalings if re quired are interdependent. The
procedures used in this book treat the two processes of scaling as in­
dependent processes. Therefore, changes can be made in either the
amplitude scaling or in the time scaling without changing the other. The
subject of scaling involves making estimates about the amplitudes of
problem variables. The student should not spend too much time making
these estimates because the trial solution obtained on the analog com­
puter will tell how satisfactory the estimate was. On the basis of the
trial solution, the estimate can be improved if necessary. There is a
relatively wide range of suitable scalings for any problem.
The third step in creating the analog model is to implement the scaled
mathematical model on the analog computer. For simple systems the
user may work directly from the scaled equations to the connection of
the analog computer modules. Generally, however, the creation of a
block diagram of the analog computer setup is helpful and may even
save time in the long run. The block diagram shows how the analog
computer modules are to be connected, the assignment of the modules,
and the settings of all the potentiometers. Such a block diagram contains
all the information needed to connect and adjust the analog computer.
In complex systems it is very easy to miss one or more of the connections
when connecting the analog computer from the block diagram. Each
connection should be marked with a check on the block diagram as it is
made on the analog computer to insure that no connections are missed.
The fourth step in creating the analog computer model is to run a
trial solution. The resulting solutions are compared with what is ex­
pected, and the range of a mplitudes of al l of th e computer variables are
observed to see that the scaling selected was appropriate. If a variable
 HYDRAULIC SYSTEM 73

has a range from —2 volts to +5 volts, then the scaling on that variable
could be adjusted to give a range from —4 volts to +10 volts, and thus
give greater accuracy, assuming a range of computer voltage of —10
volts to +10 volts. If the range of a variable was —0.5 volt to +1.0
volts, then a change in scaling would certainly be needed to improve the
accuracy of the results, although the analog model would function. On
the other hand, if the scaling is in error such that the voltage representing
a variable exceeds the limit of the voltage possible for the particular
computer being used, then the computer may be stopped because of the
overload. Even if th e computer is not stopped automatically, the results
will be in error. The presence of an overload is indicated by a light, an
audible alarm, or perhaps both. An overload indicates that the problem
must be rescaled to reduce the offending variable. The initial values of
all variables should be checked to see that they are in agreement with
the problem specifications. The final values of some or all of the variables
may be known and should be checked. The only easy check on inter­
mediate values of th e variables is in cases where the sum of several vari­
ables is a constant. The repetitive mode of operation is very useful in
running trial solutions.
The fifth step is to operate the analog computer model for the various
values of the parameters which are of interest and record the solutions.
The attenuator settings and scale factors used should be carefully re­
corded so that the results can be interpreted quantitatively. The graphs
should be completely labeled with variable names, problem units, and
titles.
The application of these five steps to the creation and testing of analog
models for several physical systems will be demonstrated by means of
some completely worked out examples. The student should look for the
thread of similarity in these examples because the same thread will be
found in all models of first-order systems regardless of the context of the
problem.

5-2 Hydraulic System

STEP 1. Formulation of Mathematical M odel

Consider a hydraulic system consisting of two tanks, A and B, which
are connected at the bottom by a pipe and valve as shown in Fig 5-1. The
water level in Tank A is maintained constant at a level of 20.0 ft by an
automatic system which is not shown. The valve is des igned so that the
rate of flow of water through the valve is proportional to the difference
74 FIRST-ORDER SYSTEMS

Tank A Tank B

f i g . 5-1

Hydraulic System.

in pressure on the two sides of th e valve. This difference in pressure can

be considered as proportional to the difference in the level of the water
in the two tanks. Let q represent the quantity of water in Tank B and
let y represent the height of water in Tank B. Eq. (5-2a) relates q t o y
and the area of T ank B.
q = A b *y = 2.0 *y (5-2a)

The time rate of flow of water into Tank B, dq/dt, is related to the time
rate of ch ange of t he quantity of water in the tank by Eq. (5-2b).

$ = 2.0 * % (5-2b)
dt dt

The time rate of flow of water into Tank B is also related to the difference
in the level of the water in the two tanks by Eq. (5-2c).

^ = K * (20.0 - y) (5-2c)
dt

The proportionality constant, K, has units of cu ft/sec-ft or sq ft/sec.

Eliminating dq/dt by equating Eq. (5-2b) and Eq. (5-2c) gives the mathe­
matical model for the system in terms of depth of water in Tank B as
the dependent variable.

2.0 * ^ = K * (20.0 - y) (5-2d)

 HYDRAULIC SYSTEM 75

Since this mathematical model involves derivatives, it is called a differ­

ential equation. This form of the differential equation is convenient to
program on the analog computer, and the relation of this equation to
the physical system being studied is easy to see. Differential equations
are frequently written in canonical form, that is, standard from a mathe­
matician's point of view, with all terms in the dependent variable and
its derivatives on the left side and all other terms on the right side as
shown in Eq. (5-2e).

2.0 * ^ + (K * y) = 20.0 * K (5-2e)

STEP 2. Scaling of the Mathematical Mod el

For this problem the maximum value for y occurs when the derivative
is e qual to zero, and hence the maximum value of y is 20 ft. For the
TR-20 analog computer the value of Cm is 10.0 volts. The amplitude
scale factor for the TR-20 computer is therefore as follows:

*' =;S = 2-0ft/volt <5-20

The general relation between depth, y, on the real system and the voltage,
Y, representing depth on the computer model is:

y ft = (Ky ft/volt) * Y volts (5-2g)

Substitution of E q. (5-2f) and (5-2g) into Eq. (5-3d) produces:

2 0 ft2 * d^Ky ft/volt) * (Y volts)]

dt
= K ft2/sec * [(K y ft/volt) * 10.0 volts — (.Ky f t/volt) * Y volts]
(5-2h)
The units on each quantity were carried to show the dimensional con­
sistency of the equation, and when the units are consolidated the result
is cubic ft/sec for all the terms. Since Ky is a constant, the differentiation
on th e left side yields
m,ft/volt) *r volts] _ ^ft/volt) , dY_ (volts/sec) (5 2i)

Eq. (5-2h) can be simplified by dividing through by Ky to give

2.0 * ~ = K * (10.0 - Y) (5-2j)

76 FIRST-ORDER SYSTEMS

For the range of values of K from 4.0 to 1.0, there is no need for us to
consider time scaling of the problem, since the time required for the
real system to function, that is, to fill Tank B, ranges from 2.5 to 10.0
sec. This fact is demonstrated by the trial solution of the next section.
Values of K such that time scaling is required will be considered at the
end of t his example.
STEP 3. Implementing the Scaled Model
The next step in creating the analog computer model of the actual
system is to solve for the highest-ordered derivative, which in this case
is (dY/dt), since a first-order differential equation represents the system.

^ ^ * (10.0 - Y) (5-2k)
at l
If the output of an analog integrator is defined as —Y, then the input is
(dY/dt). The connection of an integrator, a summer, an inverter, and a
potentiometer as shown in the block diagram of Fig. 5-2 produces the

+10 volts
V

f i g . 5-2

Block Diagram of Analog Computer Model of Hydraulic System.

terms required to compute the quantity Y. If necessary refer to Chapter

2 or 3 for details on the operation of these components. The output of
Potentiometer 01 is the quantity (dY/dt), and since this is the input re­
quired for Amplifier 01, the connection shown by the dotted line completes
the analog model of th e system of tanks and valves. The range of values
of th e constant, K, which can be accommodated with the amplifier gains
of unity as shown is from 0 to 2.0, since the maximum coefficient which
can be set on the potentiometer is 1.0. If the quantity (dY/dt) is not of
interest, then the computer model can be simplified as shown in Fig. 5-3
by combining the operation of summation with integration and elimi­
nating the two inverters which remain. If Y is desired rather than — Y,
the desired result is obtained by changing from +10 to —10 volts as
 hydraulic system 77

Analog Model Using One Amplifier, — Y Output.

Analog Model Using One Amplifier, -f Y Output.

(Y 0 /10)

Block Diagram of Complete Analog Computer Set-up.

 HYDRAULIC SYSTEM 79

shown in Fig. 5-4. The model of Fig. 5-2 has more amplifiers but is
easier to use since the ratio (AT/2) o ccurs only once, rather than twice
as with the alternative models shown in Fig. 5-3 and Fig. 5-4. The block
diagram for the complete analog computer setup including the time
base generator and the X-Y plotter is shown in Fig. 5-5. Potentiometer
04 provides an input of -1.0 volt to integrator 02. The output of ampli­
fier 02 is a ramp voltage with a slope of +1.0 volt per sec. P otentiometer
03 has been added to provide an initial condition on the depth of the
water in the tank. Since the output voltage of amplifier 01 is —Y, a
positive voltage is used for the initial condition. This block diagram is
now complete, and the analog model can be implemented on the analog
computer by making the indicated connections. The connections on a
TR-20 computer for an analog model based on Fig. 5-5 is shown in Fie
5-6. 5*

STEP 4. Running a Trial Solution

After making the connections shown in Fig. 5-6, a trial solution was
run for the value of K = 1. When observed on the oscilloscope display
with the computer running in the repetitive mode, the solution had the
expected form. The initial value of Y was 0, and the final value was —10
volts. The time required for the tank to fill up was approximately 10
sec. When the value of K was increased (POT 01 and POT 02) corre­
sponding to opening up the valve, the tank filled up faster as expected.
STEP 5. Solving the Problem
The analog computer model was operated in the normal mode and
solutions recorded for three values of K on the X-Y plotter. The results
are shown in Fig. 5-7. The axes are marked both in voltage units for the
computer variables Y and T, and in problem units, ft and sec, for the
variables y and t.
Curves were also run for two nonzero initial conditions for a value
of K = 1. The initial height is adjusted by setting potentiometer 03.
These results were plotted on the X-Y plotter and are shown in Fig. 5-8.
The case for zero initial height of water was rerun here for comparison.
These curves show the interesting fact that the time to fill the tank is
independent of t he initial depth of water in the tank. The reason for this
perhaps unexpected result is the specification that the rate of flow of
water is proportional to the difference in the depth of water in the two
tanks.
When the mathematical model is expressed as follows:
dy 1*
- = - (y, - y) (5.21)
80 FIRST-ORDER SYSTEMS

o.
O

ctf

o
SJJOA-^ 3 §BJ[OA ,£
CM GO
a
OJ
y-X qjcbp D
 HYDRAULIC SYSTEM 81

T-volts

f i g . 5-8

Height of Water vs. Time for Several Initial Heights.

then the time required for y to attain its final value is approximately
5 r . The constant, r, is called the time constant. The general solution to
Eq. (5-21) is shown in courses in differential equations to be as follows:
y = yo + 0/ - yo)(l - e~"T) (5-2m)
The constants yo and y/ are the initial and final depth of the water in
the tank. The correctness of th is solution can be easily demonstrated by
differentiating Eq. (5-2m) and substituting y and dy/dt into Eq. (5-21).
When t is zero, the exponential term e° equals one, and y is equal to the
initial value of y or >>o. When the exponent t/r is greater than five, the
exponential term e~"T is less than 0.00674. Under these conditions the
value of y is very nearly equal to the final value y/, and the solution is
said to have attained a steady-state or static condition. Comparison of
Eq. (5-2d) and Eq. (5-21) shows that
82 FIRST-ORDER SYSTEMS

This equation can be solved for r:

r = 2.0 (5-02°),

When K equals one, r equals 2.0 sec, and the solution is expected to
attain steady-state in 5 * 2.0 or 10.0 sec. When K equals 2.0, r equals 1.0
sec, and the solution is expected to attain steady-state in 5 * 1.0 or 5.0
sec. The solutions plotted in Fig. 5-7 demonstrate these ideas. The value
of t he time constant provides an indication of whether time scaling will
be required. For systems with values of r from 0.5 to 10 sec, no time
scaling is needed for recording with electromechanical X-Y plotters.

5-3 Time Scaling

Let us consider some other values for the parameter K in the water
tank example in order to see the process of t ime scaling. If the value of
K were 0.05 sq ft/sec, then the time constant for this system would be
(2/0.05), or 40 sec, and the tank would require 200 sec to fill. See Fig.
5-9 for the block diagram. This period of time poses no problems from a
computational point of view, since the time-base generator could be

0.5

+10 volts)>-

0.025

+10 volts)*-

0.005

• 10 volts)>- T
zero to ten volts
in 200 seconds

f i g . 5-9

Block Diagram for Analog Model of Tank Problem with K = 0.05 and no Time
Scaling.
 TIME SCALING 83

slowed down, but since the model on the computer can "fill the tank"
in 10 sec, the additional 190 seconds of computer time are wasted.
Furthermore, the accuracy of the potentiometer settings is low for low
ratios. Let T represent time on the analog computer model of the system
and let t represent time on the actual system. In a manner consistent with
the definition of amplitude scale factors, Kt is defined as the ratio of the
maximum time for the physical system, tm, to the maximum time desired
on the model, Tm, that is,

= (5-3a)
m

If the desired model t ime runs to 10 sec when the system time runs to
200 sec, then K, = 20.0 and has units of sec per sec. Problem time, t,
and computer time, T, are related by the scale factor K, in the following
ways:

t = Kt * T
or

T = ~ *t (5-3b)

The time rate of ch ange of a variable Y with respect to these two time
bases are related in the following way:

dY _ dY ^ dT
dt dT dt
(5-3c)
dY = dY^l_
dt dT * Kt

Substitution of E q. (5-3c) into Eq. (5-2k) and solving for (dY/dT) gives
the time-scaled equation:
dY K* K
M = - ^ *(10.0-7) (5-3d)

Since in the present example the value of K is 0.05, and the tank on the
model is to be filled in 10.0 sec, the value of Kt would be chosen to be
20 (the student should check this), and the time axis of the plot produced
on the X-Y plotter would be labeled to show 10.0 in. as representing 200
sec of time on the real system. Since the settings on potentiometers 01
and 02 are [(K * K,)/2] for the mathematical model after the time scaling
84 FIRST-ORDER SYSTEMS

—10 volts
0 to 10 volts
in 10 seconds

f i g . 5-10
Block Diagram for Analog Model of Tank Problem with K = 0.05 and Time
Scaling Kt = 20.

has been performed, the time scaling is accomplished on the analog

computer model by multiplying the original integrator gains by the
factor K,. Note the changes in potentiometers 01, 02, and 04 in Fig. 5-10
from those in Fig. 5-9. Note also that the setting of the potentiometer
03 that establishes the initial height of the water in the tank is not
changed. The initial condition is not affected by changing the time scale
on the integrators.
The need for time scaling in the other direction may also arise. Suppose
that the value of K for the hydraulic system was 20 sq ft/sec. The tank
on the real system would now fill in 0.5 sec, but the X-Y plotter could not
accurately follow the results produced by the analog computer model.
If the model tank is to fill in 10 sec, a value of K, = 0.05 is chosen, and
potentiometer settings are multiplied by this factor. The time axis of t he
X-Y plotter is labeled to show 10.0 inches as being 0.5 sec.

5-4 Freely Falling Body

STEP 1. Formulation of Mathematical Model

Consider a body with mass M falling under the influence of gravity.
Assume for simplicity that the friction of the atmosphere is negligible.
 FREELY F ALLING BODY 85

f i g . 5-11

Freely Falling Body.

This would not be a bad assumption at high altitudes. This situation is

shown pictorially in Fig. 5-11. N ewton's second law states that the rate
of change of momentum is proportional to the impressed force and takes
place in the direction of that force. The momentum of a body is the
product of the mass and the velocity, or

momentum = Mv (5-4a)

Since mass is assumed constant, the rate of change of momentum with

respect to time is as follows:

If the units are consistent, such as the meter-kilogram-second (MKS),

then the factor of proportionality is one, and

force = (5-4c)

For the present case the force is due to gravity, and therefore,

M^ = Mg (5-4d)

After dividing both sides by the mass, M, we find:

do N
at - « (5"4e)

This result is the mathematical proof of Galileo's famous experiment

at the Leaning Tower of Pisa. The acceleration experienced by a body
as a result of gravity is independent of its mass.
 86 FIRST-ORDER SYSTEMS

STEP 2. Scaling the Mathematical Mod el

Assume that this problem is to be worked on an analog computer
with a reference voltage of ±10 volts. If this experiment is to be per­
formed around the earth, the maximum acceleration is 32.2 ft per sec2.
For convenience assume max|a| = 40 ft per sec2.

max \a\ 40 ft/sec2 . ., , , .. ...

K. = = 4 ft/sec2-volts (5-4f)

Assume that the problem is to run for 25 sec. The velocity will be 25
sec times 32 ft per sec2 or 800 ft per sec. Therefore, assume max|u| =
800 ft/sec.
„ 800 ft/sec ... .
" = i q volts = ft/sec-volt (5-4g)

Assume that a computer voltage G represents the problem variable g,

therefore,
g = KaG (5-4h)

Assume that a computer voltage V represents the problem variable v,

therefore,
v = KvV (5-4i)

Substitution of Eq. (5-3h) and (5-3i) into Eq. (5-3e) yields

= *f = (5-4j)

Division by Kv yields

ft 1G
-
(5"4k1
STEP 3. Implementing the Scaled Model
The computer voltage G which represents g is computed by using
Eq. (5-4h).

~ g 32.0 ft/sec2 . .n
G = ~ = . n c. , -t- ^ = 8 volts (5-41)
Ka 4.0 ft/volt-sec2
 FREELY F ALLING BODY 87

Alteration of Eq. (5-3k) by multiplying both sides by — 1 yields:

_ d V = __ K a
dt Kv (5-4m)

Assume that the output of an integrator is V. Then the input is —dV/dt.

Hence the appropriate block diagram representation of the amplitude
scaled analog computer model is as shown in Fig. 5-12. The computer

(K/K r ) dV_
dt
~G> 02 »
f i g . 5-12

Block Diagram Relating — G and V.

voltage G should be obtained from a potentiometer followed by an in­

verter to eliminate loading. Therefore, the complete block diagram is as
shown in Fig. 5-13. A ramp function generator has been added to provide
a voltage going from 0 to 10 volts in 25 sec.

+ 10 volts^-

0.04

-10 volts

f i g . 5-13

Complete Block Diagr am.

STEP 4. Running a Trial Solution

The trial solution verifies t hat the variable V goes from 0 to 10 volts
in 25 sec and that T goes from 0 to 10 volts in 25 sec. C an you see from
the block diagram that V — T as would be expected from the nature of
the scaling?
 88 FIRST-ORDER SYSTEMS

STEP 5. Solving the Problem

The solution produced by the analog computer was recorded on the
X-Y plotter and is shown in Fig. 5-14.

T-volts (2 volts/in/

f i g . 5-14

Plot of Velocity vs. Time for Freely Falling Body.

5-5 Falling Body with Viscous Drag

STEP 1. Formulation of Mathematical Model

Consider that there is opposition to the motion of the falling body
considered in the previous section. This opposition to motion is called
damping or friction. In some cases this damping may be deliberately
increased by adding a parachute. The functional relationship between
 FALLING BODY WITH VISCOUS DRAG 89

the velocity of the object and the opposing force may be very complex.
In some cases the frictional force may be considered proportional to the
velocity and is computed as

f = Dv (5-5a)

If the force is a linear function of velocity, the drag or friction is called

viscous. The direction of this force is opposite to the direction of m otion
of the body. The parameter D is the damping coefficient. The force acting
on the body is now the difference between the force of gravity and the
damping force.

dv
M -jt = Mg - Dv (5-5b)

The velocity of the body in the previous example increased without

bound. Eq. (5-5b) shows that the object in this case will attain a limited
velocity because the impressed force decreases as the velocity increases.
Ultimately the damping force, Dv, approaches the force of g ravity, Mg,
and the acceleration of the body approaches zero. This limiting value
of velocity is called terminal velocity.

STEP 2. Scaling the Mathematical Model

Assume that the mass, M, is 20 kg and that the damping coefficient
D is 10 newtons per (m per sec). The mathematical model then becomes:

2 0 ^ = 2 0g — 1 0 o (5-5c)

The limiting velocity is ob tained by setting dv/dt = 0.

20 * g = 10 v

v = 2 * g = 2 *9.8 = 19.6 m/sec

For convenience assume max \v\ = 20 m per sec. The amplitude scale
factor Kv is for a 10 volt analog computer

K _ max |d | _ 20 m/sec
Cm 10 volts

= 2 m/volt-sec (5-5d)
90 FFRST-ORDER SYSTEMS

The amplitude scale factor Ka is as follows (assuming 10 m per sec 2 as

the maximum acceleration):

max \a\ 10m/sec2 , , u ,

*• - -or- -msa 1 m/volt-sec2

G .A » |9'8/m/sec222 = 9.8 volts (5-5e)

Ka 1 m/volt-sec

The amplitude scaled model is therefore:

M<M?D = MKaG - DKvV

at

K*dit = K-G-T,K

~ G v (5"5f)
dt =Y
Kv ~Tr
M

STEP 3. Implementing the Scaled Model

Assume that the output of an integrator is + V; then the input to that
integrator is —dV/dt. Therefore, an appropriate block diagram repre­
senting the analog computer model is as shown in Fig. 5-15. The output
of amplifier 04 provides the voltage needed as the input for amplifier 01.
If the voltage dV/dt is not needed, this block diagram can be simplified
as shown in Fig. 5-16. This block diagram also includes a ramp func­
tion generator to produce a voltage changing from 0 to 10 volts in 20 sec.

G / 10 (Ka/Kv)

f i g . 5-15

Block D iagram for Falling Body with Viscous Drag.

 FALLING BODY WITH VISC OUS DRA G 91

(K a G/\0K,) = 0.49

0.05

f i g . 5-16

Simplified Block Diagram for Falling Body with Viscous Dr ag.

T-volts (2 v/in.)

f i g . 5-17

Plot of Velocity vs. Tim e for Falling Body with Viscous Drag.
 92 FIRST-ORDER SYSTEMS

STEP 4. Running a Trial Solution

When the circuit of Fig. 5-16 is connected. The voltages V and T behave
as expected. If the ratio set on potentiometer 03 is made larger, the
terminal velocity decreases as it should when either the damping D is
increased, or the mass M is decreased.

STEP 5. Solving the Problem

Solutions run for various settings of potentiometer 03 are shown in
Fig. 5-17. The terminal velocity decreases by 50 per cent as the ratio set
on potentiometer 03 is increased to 1.0. This ratio corresponds to re­
ducing the mass by 50 per cent or doubling the amount of damping.
Notice that the axes are marked both in terms of computer variables
V and T in volts, and in terms of problem variables v and t in m per sec.
and sec.

5-6 Thermal Behavior of a Semiconductor Device

STEP 1. Formulation of the Mathematical Model

The current which most electrical devices can safely carry is limited
by the resulting temperature rise of the device. The rate at which heat is
produced is measured for direct current devices by the product of voltage
and current, which is called electric power. When voltage is measured
in volts, and current is measured in amps, the power is given in watts.
The thermal capacity of a device is a measure of the quantity of heat
energy absorbed by the body per unit change in temperature. The thermal
resistance of the device is a measure of the rate at which heat energy
is lost by the device per unit temperature difference. The ideas are ex­
pressed mathematically as

q= c,e (5-6a)

(5-6b)

d is the difference between the temperature of the device and

its surroundings (°C)
O,

q is the amount of heat energy stored (watt-sec)

 t h e r m a l be h a v i o r o f a s e m i c o n d u c t o r d e v i c e 93

Ct is the thermal capacity (watt-sec/°C)

Rt is the thermal resistance (°C/watt)

= f is the rate of transfer of heat energy by thermal conduction

(watts)

Consider the problem of determining the time for a semiconductor de­

vice to attain a critical temperature for a fixed power input. Assume the
following parameters for the device:

Ct = 20 watt-sec/°C

Rt = 4°C/watt

Also assume that the maximum permissible temperature for this de­
vice is 85°C and that the ambient temperature is 25°C. The difference
between the thermal power supplied and the thermal power conducted
serves to raise the temperature of the device. Thus:

d(CtO)
P'-P' = sr

C'It = P'-T, <5-&>

The temperature attains equilibrium when the thermal power conducted
away is equal to the thermal power supplied. Hence the device can
continuously handle a thermal power input:

max |0| (85 - 25)°C 1C

~RT = 4°C/watt =15 watts <5"6d)

STEP 2. Scaling the Mathematical Model

Since the temperature difference over which the device can operate is
limited to 60°C, an appropriate value for Kg is as follows:

K _ max |fl| _ 60°C

6 Cm 10 volts

= 6°C/volt (5-6e)
 94 FIRST-ORDER SYSTEMS

Assume that the maximum thermal power of interest will be 100 watts.
Thus the amplitude scale factor Kp is as follows:

K _ max \p\ _ 100 watts

p Cm 10 volts
= 10 watts/volt (5-6f)

The general relations between problem variables and computer variables

are as follows:
9 = Kg * 0

p — Kp* P (5-6g)

Substitution of these scale factor equations into Eq. (5-6c) yields the
amplitude scaled equation:

n d(K$Q) ^ n KgQ
c' ~dT ~ KpPs ~ ~r;

C,K,f t = K p P,~fe

do = KpPs 1
(5-6h)
dt CtKg R,Ct

STEP 3. Implementing the Amplitude Scaled Model

Assume that the voltage — d Q/dt is available. Integrate this to obtain
0, and create additional terms to form — dQ/dt. An appropriate block
diagram representing the analog computer model is as shown in Fig.
5-18. As shown before, the analog model can be simplified as shown in
(Kp/ C,Kg)(Pj/10)

— 10 volts )>•

f i g . 5-18

Block Diagram of Amplitude Scaled Model.

 THERMAL BEH AVIOR OF A SEM ICONDUCTOR DEVICE 95

( P s / 120)

— 10 volts )>-

0.0125 01

f i g . 5-19

Block Diagram of Simplified Analog Model.

Fig. 5-19 if the variable dd/dt is not needed. The potentiometer settings
are computed as:

1 1
POT 02
RtC, (4°C/watt)(20 watt-sec/°C)
= 0.0125/sec

(10 watts/volt)(Pi volts)

POT 03 = ^4
C/Kg 10 (20 watt-sec/oC)(6°C/volts)(10 volts)
= (Ps/120)/sec

( P s / 12)

f i g . 5 -2 0

Block Diagram After Time Scaling.

96 FIRST-ORDER SYSTEMS

STEP 4. Trial Solution

When a trial solution is run, the temperature is found to change very,

very slowly. This fact is evident from the very low gains on the inputs to the
integrator 01. Multiplying all integrator gains by 10.0 causes the solution
to be generated 10 times faster. Thus one second of computer time is 10
sec of problem time. The block diagram of t he analog model after time
scaling is shown in Fig. 5-20. The time-base generator has been chosen
to produce a voltage T that changes from 0 to 10 volts in 40 seconds.
Since this would have required a gain of 0.025 on potentiometer 01, an
inverter with a gain of 0.1 was used to reduce the voltage.

6 — 60 °C (above ambient)

Ps = 40

t-sec.
16 24

4 6
T-volts (2 volts/in.)

f i g . 5-21

Plot of Temperature vs. Time as a Function of Power Dissipation.

 THERMAL BEH AVIOR OF A SEMIC ONDUCTOR DEVICE 97

100 90 80 70 60 50 40
6 = 60°C
(above ambient)

Note: parameter for

these curves is ps - watts

/-sec.

1.0 1.5

T-volts (0.5 volts/in.)

f i g . 5-22

Plot of Temperature vs. Time as a Function of Power Dissipation.

STEP 5. Solving the Problem

The parameter of i nterest is the power input P s . Solutions are shown

in Fig. 5-21 and Fig. 5-22 for various values of ps. Table 5.1 lists the
values of ps in problem units, the corresponding value of the model
variable Ps, and the potentiometer settings. The dotted line is drawn
for 6 = 60°C or 0 = 10 volts. The time required for the solution to
reach this value is the time interval over which the power can be held at
the indicated level without overheating. For example, this semiconductor
device can dissipate 50 watts for 5.5 sec before the temperature rise
exceeds the permissible 60°C rise.
 98 FIRST-ORDER SYSTEMS

t a b l e 5-1. p otentiometer s et t in gs f o r st udy of ther­

mal behavior o f semiconductor device

Ps WATTS Ps VOLTS P . / 12 = RATIO OF 03

100 10 .833
90 9 .750
80 8 .667
70 7 .583
60 6 .500
50 5 .417
40 4 .333
30 3 .250
20 2 .167

5-7 Hydraulic System of Three Tanks

The examples considered thus far have involved a single dependent

variable. Some situations require two or more dependent variables to
define the system. These dependent variables are interrelated in some
fashion. Such systems are referred to as coupled systems.

STEP 1. Formulation of the Mathematical Model

Consider the hydraulic system of t hree tanks shown in Fig. 5-23. The
rate of transfer of liquid out of Tank A and into Tank B is p roportional
to the difference in depths, (ya — yb).
q\ = K\(ya - yb) (5-7a)
Tank A Tank B Tank C

ya

>'b

r i L h L
K,
f i g . 5-23

Hydraulic System with Three Tanks.

 HYDRAULIC S YSTEM OF THRE E TANKS 99

Similarly the rate of tra nsfer out of Tank B into Tank C is proportional
to the difference in depths yb and yc.

q2 = K2(yb — yc) (5-7b)

The rate of transfer out of T ank C is proportional to depth yc.

<73 = Kiyc (5-7c)

The rate of change of hei ght is related to the net rate of flow out of the
tanks in general terms as

A ^ = (7in (?out (5-7d)

This relation becomes for each tank:

Ac ^ = -qi = -Kiiya - yb) (5-7e)

= qi - q2 = Ki(ya - yb) - K2(yb - yc) (5-7f)

Ac ^dt = q2 ~ q3 = K2^b ~ y<^ ~~ Kiye (5"7g)

These equations can be solved for the derivatives of t he height of water

in e ach of the tanks.

dya Ki Ki
~dt = ~~Aa <5"7h>

dy„K, (I<C2 ...

W + T ) y b+ - A y
i b

dyc K2 (K2 . Ki\

W- ~ U + zz)y»t (5. - 7j)

These three equations constitute the mathematical model of the system

of th ree tanks.
 100 FIRST-ORDER SYSTEMS

STEP 2. Amplitude Scaling the Mathematical Model

The parameters of the problem are as follows:
Area of Tanks
Aa — 100 sq ft
Ab = 75 sq ft
Ac = 50 sq ft
Flow Coefficients
K\ = 40 sq ft/sec
K2 = 30 sq ft/sec
Kj, = 20 sq ft/sec

Height of all the tanks is 50 ft.

Initial Conditions
ya(0) = 50 ft
yb( 0) = 10 ft
yd0) = 5 ft
The scale factor for y is appropriately chosen to be the same for all tanks
since the depth of t he tanks is the same. Assume Cm = 10 volts.
„ max | v| 50 ft ^ x
= -Cif- lOrolts - 5ft/volt (5"7k»

Assume that computer voltages Ya, Yb, and Yc represent the problem
variables ya, yb, and yc.
ya — K yaYa

yb — FybYb

yc = KycYc (5-71)
Then the equations of the mathematical model become
_ K \ K \
~dT~ ~Tj' + Tjb (5"7m)

f <n 5 -r 7 » (5 >7o)
= - - (f + z ) <
dYc
dt
= K2 „
A
(K2 , K3
"
Note that because Kya — K yb = Kyc that the equations are unchanged.
 HYDRAULIC SYSTEM OF THREE TANK S

Note: See Table 5-2 for potentiometer settings and amplifier gains

f i g . 5-24

Block Diagram of A nalog Model for System of Three Tanks.

 102 FIRST-ORDER SYSTEMS

STEP 3. Implementing the Scaled Model

The computer voltages — Ya, — Yb, and —Yc will be produced by
integrating dYa/dt, dYb/dt, and dYc/dt. The block diagram is shown
in Fig. 5-24. The derivatives are formed by summing the appropriate
terms as defined by the mathematical model. Table 5-2 shows the ratios
to be set on each potentiometer and the amplifier gain to be used with
each potentiometer.

t a b l e 5-2. potentiometer ratios and amplifier gains

f o r a na l og computer model shown in fig.
5-24.

POTENTIOMETER COEFFICIENT GAIN AMPLIFIER

01 .05 1 01
02 (Ki/Ab) 0.533 1 05
03 CKl/Aa) 0.400 1 03
04 CKl/Aa) 0.400 1 03
05 (K2/Ab)+(Ki/Ab) 0.933 1 05
06 (Ki/Ab) 0.400 1 05
07 (K2/Ac)+(K3/Ac) 1.00 1 07
08 (K2/Ac) 0.600 1 07
09 (Yb/10) 0.200 IC 05
10 (Yc/10) 0.100 IC 07

STEP 4. Running a Trial Solution

The analog model was tested using the REP-OP mode and the CRO
display. The water level in Tank C was found to go below zero, which is
not physically possible. When the computer model was checked, the
high side of potentiometer 08 was found connected to amplifier 05 rather
than 06.

STEP 5. Solving the Problem

The voltages Ya, Yb, and Yc were plotted using the X-Y plotter. The
results are shown in Fig. 5-25. Notice that the water level in Tank C is
always less than that in Tank B, which in turn is less than that in Tank
A. This is as expected since water is flowing from Tank A to Tank B to
Tank C.
HYDRAULIC S YSTEM OF THREE TANKS 103
104 FIRST-ORDER SYSTEMS

5-8 Exercises

1. A water tank is in the shape of a right circular cylinder with the circular
cross section in the horizontal plane. The tank is 100 ft high and has a
diameter of 20 ft. Water is initially 80 ft deep. The discharge of water
through a pipe in the bottom is found to be proportional to the depth of
the water in the tank. The proportionality factor is 20 cu ft per sec-ft.
Create a suitable mathematical model (differential equation) which
describes this system.

Tank A Tank B

valve
f i g . 5-26

2. One portion of an industrial process involves two tanks, A and B, con­

nected by piping and a valve as shown in Fig. 5-26. After Tank A is filled,
the valve is opened and liquid is allowed to flow into Tank B until equi­
librium is attained. Create a mathematical model describing this system
on the assumption that the rate of flow is proportional to the difference
of the depth of liquid in the two tanks.
3. In order to speed up the transfer of liquid from Tank A to Tank B in
Exercise 2, a pump is added to the system to pump the liquid from Tank A
to Tank B as shown in Fig. 5-27. It is reasonable to assume that the liquid
will be transferred at a greater rate initially because of the added pressure
difference resulting from the liquid being deeper in Tank A than in Tank B.
Assume that the transfer rate is given by the following relationship:

q = [K\ + K2(ya — yb)\ cu ft/sec

Create a mathematical model describing this system.

4. Another possible arrangement for transferring the liquid from Tank A to
Tank B would be that shown in Fig. 5-28. Create a mathematical model
 EXERCISES 105

Tank A Tank B
area = A a area = A b

f i g . 5-27

Tank A Tank B

f i g . 5-28

describing this system. Assume that the pump is the same one used in
Exercise 3. Is this model valid for yb > yP
Will the system of Exercise 4 transfer the fluid faster than the system of
Exercise 3? Explain why or why not on a qualitative basis.
Assume the parameters listed below for the system of Exercise 2. The
mathematical model for this system then becomes:

— = — 0.60(ya - yb)

— = -f0.50(yfl - yb)
106 FIRST-ORDER SYSTEMS

a. Check your results for Exercise 2.

b. Amplitude scale these equations for solution on an analog computer
which has reference voltages ±100 volts.
c. Draw the block diagram of the amplitude scaled analog computer
model for this system.

K — 30 sq ft/sec ya(0) = 70 ft

Aa = 50 sq ft yA(0) = 5 ft

Ab = 20 sq ft

7. Assume that Exercise 3 has the parameters listed below. The mathematical
model for the system becomes

dy a
= -2.0 - 0.1(ya - yb)

~ = +5.0 + 0.25(ya - yb)

a. Verify the correctness of your model for Exercise 3.

b. Can Tank B overflow? Explain quantitatively.
c. Amplitude scale the model to fit on an analog computer with reference
voltages of ±100 volts.
d. Draw the block diagram for the amplitude scaled analog computer
model for this system.

A'i = 100 cu ft/sec ya(0) = 70 ft

K2 = 5 sq ft/sec y6(0) = 5 ft

ha = 80 ft

hb = 100 ft

8. Assume that the height yc for Exercise 4 is 60 ft. The mathematical model
for the system is then

dya
= - 2 - 0.1(ya - 60)

= 5 + 0.25(ya - 60)
 EXERCISES 107

a. Verify the correctness of your answer for Exercise 4.

b. What is the maximum possible depth of water in Tank B? Explain
your answer qualitatively.
c. Draw the block diagram of the amplitude scaled analog computer
model for this system. Assume that the analog computer used has a
reference voltage of ±100 volts.
Make an analog computer study of the effect of varying the parameters of
the pump for Exercise 7. Make two series of tests. Present the results on
two graphs.
a. For K\ = 100 cu ft/sec
K.2 = 0, 5, 10, 20 sq ft/sec
b. For K.2 = 5 cu ft/sec
K\ — 0, 50 , 100, 200 sq ft/sec
Note: The parameter K\ is a measure of the volume of water which this
pump delivers at zero pressure difference. The parameter K2 is a measure
of the rate of back flow through the pump as a function of pressure rise
across the pump. The ratio of Ki to K2 is a number with units of feet.
Thus the ratio (K1/K2) is the pressure measured in feet of water at which
the net flow through the pump is 0.
A projectile weighs 2 kg and is launched vertically with an initial velocity
of 20 m per sec. Assume that the drag resulting from air friction is viscous
and that the coefficient is 10~n newton-sec per m.
a. Produce the block diagram representation for the amplitude scaled
analog model for this system. Scale the model for an analog computer
with reference voltages of ±100 volts (or to fit the computer which
you are going to use for Part c).
b. A simple modification of the analog computer model will permit the
vertical height to be displayed. What modification is required?
c. Set up the analog model, and determine the time the projectile will be
in the air and the maximum height that it will attain.
A cannon has a barrel 8 m long and a bore (internal diameter) of 15 cm.
The projectile weighs 12 kg. Assume that the gas pressure in the barrel is
constant at 5 X 104 kg per sq m. Assume the friction of the projectile in
the barrel to be viscous with a coefficient of 2 newton-sec per m. One
other assumption is needed.
a. Produce the block diagram representation for the amplitude scaled
analog model for this system. Scale the model for an analog computer
with reference voltages of ±10 volts (or to fit the computer available
to you if you are going to do Part b).
b. Set up the analog computer model to determine the velocity of the
projectile as a function of time until the projectile reaches the muzzle
(the end of the barrel). What is the velocity of the projectile as it
leaves the end of the barrel?
Consider the electrical circuit shown in Fig. 5-29. The rectangular box at
the left is the current generator which develops a constant current in-
108 FIRST-ORDER SYSTEMS

ig — 0.02 amps C, = 5,000 X 10 6farads

R, = 15,000 ohms C2 = 3,000 X 10" 6farads
R2 — 5,000 ohms v, (0) = v2 (0) = 0

f i g . 5-29

dependent of the potential difference across it. By Kirchhoff's Current

Law the sum of the currents flowing out of a junction is equal to the sum
of the currents flowing into that junction. The application of Kirchhoff's
Current Law to junctionC1 j yields

^ dvi , Of - 02
". +, /.. - c, —+ =h

Application to junction 2 yields

. . . _ dvi , V2 Vl —V2

ig = 0.02 amp Ci = 5000 X 10~6 farad

Ri = 15,000 ohms C2 = 3000 X 10-6 farad

Z?2 = 5000 ohms Ul(0) = £72(0) = 0

Create the block diagram representation for the amplitude scaled

analog computer model of this electrical network. Amplitude scale
for an analog computer with reference voltages of ±100 volts or for
the analog computer which you will use for part b.
b. Set up the analog model, and obtain plots of vi and v2 which show
the time from t = 0 until equilibrium is attained.
chapter SECOND-ORDER SYSTEMS

Some physical processes involve not only the first deriva­

tive of a variable, but also the second derivative of this variable. Such a
system is called a second-order system. This situation arises when there
are two different forms of energy storage. For example, in a mechanical
system with a spring, mass, and dashpot, mechanical energy is stored as
kinetic energy in the motion of the mass, and as potential energy in the
deformation of the spring. In an electrical system consisting of a resistor,
inductor, and capacitor, electrical energy is stored as potential energy in
the form of electrical charge in the capacitor, and as kinetic energy in the
motion of electrical charge through the inductor. Examples of several
second-order systems will be considered to demonstrate the procedure of
creating analog computer models for second-order systems. An alternate
procedure consists of defining a set of two first-order differential equa­
tions which are equivalent to the second-order differential equation and
then handling them by methods discussed in the previous chapter for sets

109
110 SECOND-ORDER SYSTEMS

of first-order differential equations. Both of these procedures will be

demonstrated. The five steps of the previous chapter are used in both
cases.

6-1 Spring-Mass-Dashpot System

STEP 1. Formulation of the Mathematical Model

Let us consider the spring-mass-dashpot system shown in Fig. 6-1. This
system can store mechanical energy as kinetic energy in the motion of the
mass. Mechanical energy is stored as potential energy, energy of position,
in the displacement of the spring. Energy is dissipated by virtue of the vis­
cous friction in the dashpot. An example of a dashpotisthe shock absorber
on an automobile. The shock absorber converts mechanical energy into
heat when there is motion, and so reduces the oscillation caused by
bumps in the road. The mathematical model for this system is formed
by summing up the three forces as defined by physical laws and equating

,d2x
kinetic M f t = Ma = M
energy dt2

Energy b/2 = Dv = D te­

loss
rn

potential^
energy*^ f3=Kx
>-v

M
-I

2/l-/l+/,+w(£WfUft-0
f i g . 6-1

Spring-Mass-Dashpot System and Differential Equation.

 SPRI NG-MASS-DASH POT SY STEM 111

the sum to zero by v irtue of the structure of the system. The inertial
force is proportional to the product of the mass and the acceleration
experienced by the mass. In equation form this force is as follows:

The frictional force is proportional to the velocity of th e mass when the

friction is viscous. The equation form of this force is fi where D is the
damping force coefficient.

fi = Dv = (6-lb)

The force resulting from stretching or compressing the spring is propor­

tional to the displacement of the spring from its zero force position,
which is taken for convenience to be zero. In equation form the spring
force is fi where K is the spring force coefficient.

fi = K* x (6-lc)

Consider first the case where the external applied force is zero and the
mass is given an initial displacement from the zero position but no initial
velocity. For this case the sum of the forces is zero:

fi + fi + fi = M ^ -f D ~ + Kx = 0 (6-ld)

For convenience take the following numerical values for the constants
and the initial conditions:

M = 2.5 kg
D = 4.0 newtons/(m/sec)
K = 5.0 newtons/m
x:(0) = 0.5 m, u(0) = 0 (6-le)

Substitution of these constants into Eq. (6-If) yields the differential

equation for this particular system.

2.5 ~ + 4.0 ^ + 5.0* = 0 (6-If)

 112 SECOND-ORDER SYSTEMS

The following integral relations result from the definition of the process
of integration and are useful in setting up the analog computer model.

dx ,
dt + Cl (6"lg)
/

dx f (fix ,
dl + C(Wh>

STEP 2. Scaling the Mathematical Model

The displacement x will never be greater than the initial displacement

because the mass has no initial velocity and therefore the stored energy in
the system must decrease, and the displacement when the mass is at rest is
proportional to the square root of the stored energy. An appropriate scale
factor for displacement is therefore as follows:

k"= = = 005 m/vo,t (6-n)

Assume that U is the computer voltage which represents the displacement

x. The displacement x relates to the voltage U by

x m = 0.05(m/volt) * U volts (6-lj)

For lack of information on the range of velocity, let us assume initially

the same scale factor on velocity as on displacement and that V is the
computer voltage which represents the velocity, dx/dt. The velocity, v,
relates to this voltage, V, by

v
dx
= m/sec = 0.05(m/sec-volt) * V volts (6-lk)

The initial value of the acceleration will be the largest encountered

because, as before, the energy in the system will be d ecreasing. Therefore
substitute the initial conditions into Eq. (6-If) and solve for d2x/dt2.

=~Lm - 2-0(0-5)
= —1.0 m/sec 2 (5-11)
 s p r i n g - m a s s - d a s hp o t s y s t e m h3

The appropriate scale factor for acceleration is, therefore,

fca max id2x/dt2] __ 1. 0(m/sec2)

=
C™ 10 volts
= 0.1 m/sec2-volts (6.,m)

dv/dZ C°TUte r VO'lage that repreSe"tS the deration,

t
' °f d x/dtThe -
acceleration, relates to this voltage, by

IP ~0AW (6-1 n)

for": °f Eq' (6"'j)> (6"lk)> a"d (6"'n) int° and solving

0.1W = —1.6(0.05 F) - 2.0(0.05t/)

W = -°-8E - " (6-lo)

This equation in the computer variables TJ V and u/ r

:tg:sn^T ationfps are needed

differential equation for the system. Since there are three°vadablesmtwo

"aTh
Eqs Tlgl^dlS^ Thferel.at,0"s are obtained from the integral
(6q^) and:i,ltIs,follows Utl°n * Srali"8 ^

0.0517 = .v = j-jt dt + a = J0.05V dt + 0.051/(0)

n 0*5 V — ^ [(fix , f
W ~ di = J dfi + c2 =J 0l>vdt + 0.05V(0) (6-lp)

Simplifying these equations yields the following relationships:

V = f v d t + 1/(0) (6.iq)

V = / ZWdt + V(0) (6.lr)

STEP 3. Implementing the Analog Computer Model

The last two integral equations provide the starting point for creating
the analog computer model for the given differential equation Because of
the sign inversion associated with the process of integration, the equations
114 SECOND-ORDER SYSTEMS

<- f/(0)
IC
u
— v> > U = - [ f - V d t - U ( 0)]
(6.1 -q)

IC < V(0)
0.2
- V
wy V = —[2 f wd t + V ( 0 ) ]
10
(6.1 - r )

0.8

F>
w= -(0.8F+ U)
U> (6.1-0)

fi g . 6-2

Analog Computer Block Diagrams for Individual Equations.

are rewritten as shown in Fig. 6-2. From this form of the equations, the
three block diagram segments can be easily drawn. The output W from
amplifier 01 provides the input needed for potentiometer 02. The output,
— V, from amplifier 02 provides the input needed for amplifier 03. An
inverter is needed to provide the +V input for potentiometer 01. These
connections are shown in Fig. 6-3. If the value of acceleration is not re­
quired, the summer amplifier 4 may be omitted as shown in Fig. 6-4. In
this block diagram provision has been made for the initial condition on U
by connecting potentiometer 3 to the IC terminal on integrator 3.

f i g . 6-3

Analog Computer Block Diagram for Second-Order System of Fig. 6-1.

 SPRING-MASS-DASH POT SY STEM 115

•<( —10 volt s

0.16

L©
10
u
0.2
10
^iL-0

f i g . 6-4

Analog Computer Block D iagram Simplified by Elimination of W.

Note the changes in the ratios set on potentiometers 01 and 02. The
ratio on 02 is set to the value which preserves the gain around the upper
loop, that is, —10(0.8)(0.2) = —1.6. The ratio on 01 is set to the value
which preserves the gain around the lower loop, that is, —10(0.2) = —2.0.

STEP 4. Trial Solution on the Computer

When the system shown in Fig. 6-4 is connected and operated on the
computer, the solution changes too rapidly to be satisfactorily recorded

— 10 volts

U
OA ' 02

-u

/ = 0.2T

<^4-10 vo lts
IC
0.1

10 v olts

f i g . 6-5

Computer Block Di agram After Time Scaling.

116 SECOND-ORDER SYSTEMS
 NONZERO FORC ING FUNCT ION 117

on the X-Y plotter. Therefore, the time scale must be changed to slow
down the generation of the solution. To cause the solution to be generated
only 20 per cent as fast, the gains of all inputs to integrators are multiplied
by 0 .20. Now 5 sec of computer time correspond to 1 sec of problem
time. The new block diagram is shown in Fig. 6-5. The results as plotted
on an X-Y plotter are shown in Fig. 6-6. Note that vertical axes are
marked in terms of b oth the computer variable in volts and the problem
variables in m a nd m per sec. The time axes is marked in terms of the
computer variable, T volts, the computer time variable in computer sec,
and the problem variable t in problem sec.

6-2 Nonzero Forcing Function

The term or terms of a differential equation which d o not involve the

dependent variable are referred to as the forcing function, since such
terms arise when a force is applied to a system as shown in Fig. 6-7. The
differential equation then becomes

(Fx dx
MZfi + Dit + Kx=f (6-2a)

x = 0

i—a
f i g . 6-7

Spring-Mass-Dashpot System with Applied Force.

When this is solved for d2x/dt2 the result is as follows:

d2x _ f D dx K
(6-2b)
dt2 ~ M~ M~dt ~ MX

Suppose that the constants are the same as used in the previous example,
that the force applied is 2.0 newtons, and that the initial displacement is
zero. As time increases, the derivatives approach zero and the displace­
 118 SECOND-ORDER SYSTEMS

ment approaches a limiting value which can be computed in the following

way:

a _ / d
m K
M M MX

_ / _ 2.0 newtons
AT 5.0newtons/m

* = 0.4 m

Since this final displacement is less than the initial displacement of the
previous example, the maximum acceleration will be less than for the
previous example, and therefore the same ka will be valid. The constant
term f/M has a value of

/ 2.0 newtons __
j[f ~ 2 5 kg ~~ m/sec (6-2c)

Since this term has units of acceleration, the acceleration scale factor
should be used here. Assume that a voltage F on the computer represents
the term f/M. Then the voltage F a nd the term f/M are related in the
following way:

f
(m/sec2) = &a(m/sec2-volt) F(volt) (6-2d)

Solving for F and substitution of n umerical values for f/M and ka yields

«•*>

Substitution of the amplitude scaling equations into Eq. (6-2b) and solv­
ing for IT yields

0.1W = 0.1F - 1.6(0.05 T) - 2.0(0.05f/)

W = F - 0.8T - U (6-2f)

The integral equations which relate the computer variables are not
changed from the previous example because they depend on the relation­
ships between displacement, velocity, and acceleration. The block dia­
gram representation of the analog models for these three equations is
 NONZERO FORCING FUNCTION 119

U = - [ f - V d t - U ( 0)]

- V =- f[ W
2 d t + K(0)]

f i g . 6-8

Block Diagram of Analog Models for System with Force Applied.

shown in Fig. 6-8. Compare with Fig. 6-2 and note that the only change is
the additional input Finto amplifier 01.
These models can be combined as in the previous exercise. Since the
solution will be g enerated too rapidly for plotting using an X-Y plotter,
time scaling is again required. The Block diagram for the analog model is
shown in Fig. 6-9 after all integrator gains have been multiplied by 0.2 to
slow down the computer model. Again 5 sec of computer time is equal to
1 sec of system time.
Compare the block diagram of F ig. 6-5 for the previous example with
Fig. 6-9 for this example. The solution curve for this problem is shown in
Fig. 6-10. Compare the solution curves for these two examples. Notice
that the velocity curve plotted in Fig. 6-6 is negative velocity. The velocity
is negative initially in the actual system because the mass is moving from a
positive displacement toward the origin or neutral position. The velocity
 120 SECOND-ORDER SYSTEMS

Computer Block Diagram for System with Ap plied Force.

curve plotted for the second system is a plot of p ositive velocity obtained
by inverting the voltage, — V, from the analog model. The time of occur­
rence of certain events should be the same for these two systems. Some of
these pairs of events are tabulated in Table 6-1. Check the two curves to
see that the times are identical. Complete the table.

t a b l e 6-1. pairs o f events which occur a t th e same

time for the two examples

FIG. 6-6 FIG. 6-10

maximum negative velocity maximum positive velocity

zero velocity zero velocity
zero displacement * = +0.8
maximum negative
displacement ?

? JC = + 0.4

6-3 The Swinging Door System

Consider a swinging door that is 1.0 m wid e and 2 m high and which
weighs 8.0 kg. This swinging door is returned to the closed position by a
torsional spring which has a torque constant of 0.3 newton-m per radian.
THE SWIN GING DOOR SY STEM 121
 122 SECOND-ORDER SYSTEMS

The problem is to determine the amount of damping required to allow

the door to close in the minimum possible time with a maximum over­
shoot of 0.1 radian. Overshoot is the amount the door swings past the
equilibrium position of 0 = 0. If the damping is too small, the door will
close more rapidly but will have more overswing. If the damping is too
large, the door will ha ve less overswing but will re quire a longer time to
close. The problem as stated is not completely defined. This situation
frequently arises in practice where the problem is not fully defined. It is
the engineer's responsibility to find the missing information, if the in­
formation exists, or to make reasonable assumptions for the missing in­
formation. In the latter case these assumptions should be explicitly called
out. Consider what else needs to be specified.

STEP 1. Formulation of the Mathematical Model

Three torques will be present in this system: the spring torque tending
to close the door, the damping torque proportional to the angular velocity
of the door, and the inertial torque proportional to the angular accelera­
tion of the door. These torques are expressed quantitatively as

Spring torque = KQ

dQ
Damping torque = Z) — (6-3a)

Inertial torque = J
at2

The angular position of the door with respect to its equilibrium position is
measured by the angle 0. The sum of t hese three torques is equal to the
applied torque which is zero for this problem. This condition yields the
mathematical model for the structure of the system given as Eq. 6-3b.

2 torques = J^+ D^ + KQ (6-3b)

The definitions of angular velocity and angular acceleration yield the two
additional equations needed which are as follows:

df)
(6-3c)
-I ~dt + Q(0)

dQ
dt
 THE SWIN GING DOOR SYSTEM 123

STEP 2. Amplitude Scaling the Mathematical Model

The point at which the problem is not completely specified becomes
evident now, if it was not evident before. The initial displacement and
initial velocity are needed. For the door in question the maximum dis­
placement is 100°. Since the remainder of the problem is stated in terms of
radians, consider for convenience that the maximum displacement of the
door from its equilibrium position is 2.0 radians. The amplitude scale
factor fo r displacement is, for an analog computer with a range of ±10
volts, found as follows:

„ max [el 2 radians „ _ ,. . . ,, _ .

Ki = -car = TovsitT = °-2 radian/volt (6-3e>

If a co mputer voltage U represents displacement, then

0 = KdU = 0.2U (6-3f)

A consideration of ou r experience with swinging doors suggests that the

door should require the order of 5 sec to close. If the door closes too
rapidly, it may be dangerous to small children, and if it closes too slowly,
its function to keep the hot air in in the winter and the cold air in in the
summer will be impaired. An estimate of the average velocity would
therefore be as follows:

vav = 2 radians/5 sec = 0.4 radian/sec

The ma ximum velocity may be three to four times the average velocity
since the initial and final velocity are zero. This gives an estimate of 1.2 to
1.6 m per sec for the maximum velocity. For convenience assume initially
that the maximum velocity is 2.0 radians per sec. The amplitude scale
factor fo r velocity is t herefore as follows:

dQ
Kv = max dt 2 radians/sec
(6-3g)
Cm 10 volts
= 0.2 radian/volt-sec

If the computer voltage, V, represents the angular velocity of the door,

then

^ = K,V(6-3h)
at
 124 SECOND-ORDER SYSTEMS

The validity of the assumptions made in arriving at these amplitude scale

factors will be tested by the trial solution run on the analog model. If they
are not valid, the trial solution will pr ovide insight as to the changes re­
quired. We will c onsider that the initial angular velocity of the door is
zero. An estimate of the maximum angular acceleration can be made by
substituting the initial displacement and angular velocity into Eq. (6-3b)
and solving for the initial acceleration. Since the initial velocity is zero,
the unknown value of D poses no problem. The numerical value for the
moment of inertia J is needed before proceeding with the solution.
Consider that the door is hinged on the >>-axis an d that the door is in
the x-y plane when at rest. The moment of inertia about the y-axis is then
defined to be as follows:

J = /// r2a dx dy dz (6-3i)

If the door is thin compared to the width, then the distance r from the axis
of rotation to the element of volume is approximately the x coordinate of
the volume. If the density, cr, is uniform and the thickness Z constant,
then the moment of inertia can be computed as

See Fig. 6-11 for a sketch of the door and the element of area. Integration
of Eq. (6-3j) yields

J = ZYX3g = (aXYZ)X2 = MX2

(6-3k)
3 3 3

The product of the dimensions X, Y, Zof the door and the density <r gi ves
the mass M of the door. The original statement of the problem gave a
value of 8. 0 kg for the mass and 1.0 m for the width X of this door and,
therefore, the inertia / is:

/ =|= 2.67 kg-m2 (6-31)

Eq. 6-3b becomes upon substituting the known constants:

(6-3 m)
 THE SWIN GING DOOR SYSTEM 125

y-meters

7 = 2

x *-j

dx

door

x-meters
X = 1

f i g . 6-11

Sketch of Swinging D oor in an X-Y Coordinate System.

Solution of th is equation for the initial value of a ngular acceleration re­

membering that dQ/dt is zero initially yields:

cPQ (0.3X100' radian/degree^

dt2 (2.67) "(lio
= 1.4 radians/sec2 (6-3n)

The nearest value that produces a convenient scale factor is 2.0

radian/sec2, and therefore the scale factor for acceleration is as follows:
d2Q
Ka = max dt
2 2.0 radian/sec2
Cm 10 volts
= 0.2 radian/volt-sec2 (6-3o)

Assume that the computer voltage, W, represents the angular acceler­

ation:
m= KW (6-3p)
a
dt2
 126 SECOND-ORDER SYSTEMS

Substitution of the amplitude scaling equations into the model equations

yields the following set of amplitude scaled equations, which the student
should verify.
2.67 W + DV + 0.3U = 0
U = f Vdt + U( 0)
V = f Wdt (6-3q)
The value of the initial condition, U(0), is computed

U( 0) = 6(0)
Kd

= 1.75 radian/(0.2 radian/volt)

= 8.75 volts (6-3r)

STEP 3. Implementing the Analog Compu ter Model

Eq. (6-3q) are implemented on the analog computer by assuming that
the voltage, W, is ava ilable as the output from a summer, by integrating

— 10 volts

o.i (y) 2 volts/in.

T
—10 vo lts)>- 07 ~(^) 1 volt/in.
—0
X — Y plotter or
oscilloscope
f i g . 6-12

Block Diagram of An alog Computer Model for Swinging Door.

THE SWIN GING DOOR SYSTEM 127
 128 SECOND-ORDER SYSTEMS

W to obtain — V, and then integrating —V to obtain displacement, U.

See Fig. 6-12. Inverter 04 is needed to provide + Vas required to make up
the inputs for the summer 01. Potentiometer 03 produces the —8.75 volts
required for the initial condition on the displacement, U. Integrator 07
provides a time-base voltage, T that goes from 0 to 10 volts in 10 sec,
which covers the expected time for solution.

STEP 4. Trial Solutions

The first trial solution was run for a setting of 0.30 on potentiometer
01. This setting is the value of the product 0.375Z). Hence this value is for
a damping constant less than 1.0. Solution Curve 1, plotted in Fig. 6-13,
has too much overshoot, which suggests that more damping is required.
However, if more damping is added, the door will close even more slowly,
and it already takes in excess of 5 sec to close. Although a value, D, could
be found which would limit the overshoot to the desired 0.1 radian with
the given spring constant, such a solution would leave the door open too
much of the time. The real problem then is to adjust both the damping, D,
and the spring constant, K, to meet the condition of clos ing the door to
less than ±0.1 radian in less than 5 sec. Trial Run 2 was made for an in­
crease in the torque constant of the closing spring. The closing time is
reduced, but the overshoot is still excessive as would be expected. Trial
Run 3 was run for an increase in both the spring constant and the damp­
ing. This curve is better, but the overshoot is still too large and the closing
is too slow. Trial Run 4 was made for a further increase in the damping
and the spring constant. Trial Run 5 is made for a further increase in the
spring constant of 20 per cent and produces what appears to be a satis­
factory closing of the door. This series of adjustments could be made
much more rapidly using the REP-OP mode. The plots were made on the
X-Y plotter to show the process of adjusting the parameters in the system
to obtain the desired response. The behavior of the door can be observed
very nicely by watching the needle on the voltmeter when the voltmeter
is measuring the voltage, U. The position of the needle is analogous to the
position of the door if you look down on the door from above.

6-4 Simple Pendulum

A simple pendulum consists of a rigid rod whose moment of inertia is

negligible compared to the inertia of the weight or bob which is attached
at the lower end of the rod. This rod is usually pivoted at the upper end in
 SIMPLE PENDU LUM 129

such a way th at the friction of rotation of the rod and bob about the
support is quite small. A diagram of such a pendulum is shown in Fig.
6-14.

f i g . 6-14

Diagram of a Simple Pendulum.

The distance from the point of support to the center of g ravity of the
bob is represented by the symbol L. The mass of the bob is represented by
the symbol M, and the angle between the supporting rod and vertical is
represented by the angle 0 measured in radians.

STEP 1. Formulating the Mathematical Model

Gravity acting on the bob produces a downward force equal to the
product of the mass, M, of the bob and the acceleration of gravity, g.
This force can be resolved into two components: one that is parallel to the
axis of the supporting rod and a second one that is perpendicular to the
supporting rod. Both of these forces act through the center of the bob.
The force perpendicular to the rod is as follows:

fa = Mg sin 0 (6-4a)
 130 SECOND-ORDER SYSTEMS

This force tends to cause the pendulum to move to the equilibrium posi­
tion where 0 = 0. The pendulum also experiences an inertial force, /„
which can be calculated as

f > ~ u t («-4b>

The velocity, v, is related to the angular velocity co by Eq. (6-4c)

v = Lw (6-4c)

The inertial force can be expressed in terms of angular velocity as

fl = M*M = LM§(6-4d)

The third force acting on the bob is the damping force, which is assumed
to be proportional to velocity,
fd = DLu (6-4e)
The sum of these three forces is equal to zero, since there is no externally
applied force in this system. Thus the mathematical model that results
from the structure of the system is obtained by summing the forces as
follows:

f i + f d + f a= LM ~ +D c oL + Mg(sin 0) = 0 (6-4f)

The following relation is used to further simplify the equation based on

the assumption that the angle is sufficiently small.

sin 0 = 0 (6-4g)

The substitution of this relation into Eq. (6-4f) yields

LM ^ + LDoo + MgO = 0 (6-4h)

This equation can be further simplified by dividing through by LM.

dw Doo gQ

The angular velocity w is relate d to the angle 0 by

dQ
(6-4j)
dt
 SIMPLE PENDU LUM 131

These latter two equations constitute a set of simultaneous first-order

differential equations representing the dynamics of th e pendulum, which
is a second-order system because there are two forms of en ergy storage,
kinetic and potential. These two equations can be organized in a manner
similar to that for simultaneous algebraic equations as follows:

a l
+- +ze=° <6*4k>

- »+ ^ =0 (6-41)

These two equations can be solved together to eliminate one or the other
of the variables. If a> is eliminated, the following second-order differential
equation results:

d29 . D dQ q n tc a \
dfi + MY, + Le = 0 (6"4m)

This process is clearly reversible and demonstrates the process of repre­

senting a second-order differential equation by a set of two first-order
differential equations. The analog computer model that results will be the
same for either approach. The following equations are needed along with
Eq. (6-4m) to complete the mathematical model:

0 = J co dt -f- 0(0 ) (6-4n)

dt + a<0) (6-4o)

STEP 2. Amplitude Scaling the Mathematical Model

For purposes of thi s example take the following as parameters for the
pendulum to be studied:

L = 1.0 m
M = 2 kg
D = 0.06 newton-sec/m
—0.1 < 0 < +0.1 radian
0(0) = 0.1 radian
g = 9.81 m/sec2
 132 SECOND-ORDER SYSTEMS

These constants are comparable to that for the pendulum of a typical

grandfather clock. Substitution of these constants into Eq. (6-4m) yields
the following:
>72 f )
+ 0.03 aff + 9.810 = 0 (6-4p)

For a reference voltage of ±100 volts, the amplitude scale factor for dis­
placement is as follows:

„ _ max |9| _ 0.1 radian

d Cm 100 volts

= 0.001 radian/volt (6-4q)

The pendulum of the typical grandfather clock makes a complete cycle

from one side to the other and returns in 2 sec. The average velocity is
therefore 0.4 radian divided by 2 sec, or 0.2 radian per sec. The maximum
velocity will be greater than the average velocity, since the velocity is zero
twice each cycle. Select a value of 0.5 radian per sec to give a convenient
scale factor.
dQ
Kv = max dt 0.5 radian/sec
Cm 100 volts

= 0.005 radian/volt-sec (6-4r)

The maximum acceleration will occur when the displacement is maximum

and the velocity is zero. Therefore, the maximum acceleration can be
computed from Eq. (6-4p) as

ePQ
max = 0.981 radian/sec2 (6-4s)
dP

Select a value of 1 radian per sec2 to give a convenient scale factor.

d2Q\
Ka = max dP |
Cm

1 radian/sec2
100 volts

= 0.01 radian/volt-sec2 (6-4t)

 SIMPLE PENDU LUM 133

The problem variables can be e xpressed in terms of c omputer variables

U, V, W as:
9 = KdU = 0.001U
df\
W = ^ - = K„ V = 0 . 0 0 5 F
at

= K°w = 0.0101F (6-4u)

at at2

Substitution of the amplitude scaling of Eq. 6-4u into the equations for
the mathematical model, Eqs. (6-4n), (6-4o), and (6-4p), yields the ampli­
tude scaled equations:

0.00IU = f (0.005V)dt + 0.001t/(0)

0.005 F = J (0.010W)dt + 0.005 F(0)
(0.010 IE) + 0.03(0.005 F) + 9.81(0.001U) = 0 (6-4v)

These equations can be solved for the three computer variables W, U,

and F:
U = J 5.0Vdt + U(0)
F= J 2.0Wdt + F(0)
W = —0.015F - 0.981C/ (6-4w)

The initial velocity is assumed to be zero, and therefore F(0) = 0. The

initial displacement is 0.1 radian and therefore £/(0) = 100 volts.

STEP 3. Implementing the Amplitude Scaled Model

The amplitude scaled equations are implemented as shown in Fig. 6-15.
A convenient place to start is to form the voltage, W, by summing appro­
priate fractions of U an d F. The voltage, — F, is then formed by integrat­
ing the voltage 2W. The voltage U is formed by integrating the voltage
— 5F. An inverter is r equired to invert — F to obtain +F. A time-base
generator is included to provide a voltage increasing at 10 volts per sec.
The IC on integrator 07 creates a voltage, T, which starts at —100 volts.
Thus a period of 20 sec is available for recording solutions.

STEP 4. Running Trial Solutions and STEP 5. Solving the Problem

These steps are left as an exercise for the student.

 134 SECOND-ORDER SYSTEMS

f i g . 6-15

Block D iagram of Analog Model for Grandfather Clock Pendulum.

6-5 Exercises
1. a. Connect the analog model for the pendulum of the grandfather clock
and determine the period of oscillation, that is, the time required for
one complete cycle. This can best be obtained by recording the dis­
placement for at least ten cycles. Check the computer variables V and
W to verify that the scaling is satisfactory,
b. Reduce the damping to zero and observe the period of oscillation. As
long as the damping is small enough, the damping should not much
influence the period of oscillation.
2. a. Adjust the length of the pendulum to obtain a period as near to 2.0
seconds as possible,
b. In physics books it is shown that the frequency of oscillation in radians
per sec of a simple pendulum is -\Jg/L. The frequency in cycles per sec is
obtained by dividing the frequency in radians per sec by 2ir. The period
of an oscillation is the reciprocal of the frequency in cycles per sec.
Using this series of calculations, check to see that the length which you
determined does indeed have a period of 2 sec.
 EXERCISES 135

3. An assumption was made in the formulation of the mathematical model for

the simple pendulum that the sin 0 was approximated by the angle 0 in
radians for small angles. Reduce the initial displacement to 0.05 radians
and run a solution for comparison with the solution for an initial displace­
ment of 0.1 radian. A convenient way to compare the solutions is to change
the sensitivity of the recorder so that the two curves will be coincident.
4. Consider a Blackburn Double Pendulum which consists of a weight sus­
pended on a string attached to the midpoint of a second string. This second
string is attached to two points at the same horizontal elevation. The par­
ticular pendulum of interest for this problem has a string 100 cm long
attached to two points on a bar which are 60 cm apart. A string 40 cm long
is attached by a knot to the midpoint of the first string. Consider that the
support is arranged parallel to the x-axis as shown in Fig. 6-16. If the

2t

f i g . 6-16
Blackburn Double Pendulum.

pendulum swings only in the y-z plane, the length of the pendulum is
effectively 40 cm plus the distance a. If the pendulum swings in the x-z
plane, the length is only 40 cm. If it swings in both planes, the total motion
is simply the sum of the individual motions.
136 SECOND-ORDER SYSTEMS

a.
Produce an amplitude scaled block diagram representation for an
analog computer model for this pendulum. The initial displacement in
the x-z plane should be 0.08 radian, and the initial displacement in
the y-z plane should be 0.04 radian.
b. Set the analog model up on an analog computer and study its behavior.
Hint: You will need to use small angle approximation to convert from
analog model variables for the angles with respect to the z-axis into dis­
placements along the ^ and y axis for plotting.
5. Consider the problem of stopping a rolling freight car such as occurs at the
end of a spur track on the railroad. A schematic representation of this sys­
tem is shown in Fig. 6-17. The mass of the car is 20,000 kg. The rolling

D
f i g . 6-17
Schematic Representation for Stopping a Rolling Freight Car.

friction of the car at low velocities is 1000 newton-sec per m. The spring co­
efficient is 5000 newtons per m. The velocity of the car as it strikes the
bumper is 0.5 m per sec.
a. Create a block diagram of the amplitude scaled analog computer
model for this system. Chose scaling appropriate to the analog com­
puter which you have available.
b. Run the analog simulation and determine
(1) the position of the car vs. time after impact.
(2) the velocity vs. time after impact.
(3) the acceleration of the car vs. time after impact.
c. What is the maximum deflection of the spring, and what is the maxi­
mum deceleration of the car?
d. Based on the results of your analog simulation, should more damping
be added to reduce the rebound?
e. For the system as modeled, determine by trial and error the maximum
initial velocity with which the car can strike the bumper for a maximum
deceleration of 0.2 g.
6. Modify the model of Exercise 5 and determine how far the car would roll
on level track in the absence of the bumper for initial velocities of
a. 2 m/sec.
b. 1 m/sec.
c. 0.5 m/sec.
7. In what way can the analog model of Fig. 6-12 for the swing door be simpli­
fied to use less equipment? Draw the simplified block diagram.
chapter NONLINEAR OPERATIONS

The assumption that systems are linear makes the

mathematical analysis much simpler. Many real life systems however
are not linear, and hence the analysis may be quite difficult without the
assistance of computers. If the analog computer is to be of aid in these
situations, means must be found to model nonlinear phenomena. The
electrical diode is a nonlinear resistor and is very useful in creating
analog models.

7-1 Diode Limiters

The electrical diode has the property that the resistance is very low
for current flow in one direction, called the forward direction. The re­
sistance to current flow in the opposite direction, called the reverse
direction, is very high. Symbols for diodes are shown in Fig. 7-1. The

137
 138 NONLINEAR OPERATI ONS

f i g . 7-1

Symbols for Electrical Diodes.

symbol on the left is that commonly used for semiconductor diodes and
the symbol on the right is that used for vacuum diodes. Semiconductor
diodes are widely used in analog computers today because they have
adequate ratios of backward to forward resistance and they do not have
the heaters of the vacuum diodes. The forward direction is that indicated
by the arrow. It is customary to mark the cathode end with a K, or per­
haps just with a band. The operation of diodes in a simple circuit is
illustrated in Fig. 7-2. In the left circuit the current is flowing in the re-

100 i f -
\
current small current large
diode reverse biased diode forward biased

f i g . 7-2

Simple Diode Circuits.

verse direction through the diode, and hence the diode exhibits a very
high resistance. Since the back resistance of the diode is very large com­
pared to the resistance of the lamp, the lamp does not light. The diode
is said to be reverse biased. The connection of the diode has been re­
versed in the circuit on the right; and the current now is flowing in the
forward direction of the diode, which is the low resistance direction.
Since the forward resistance of the diode is small compared to the re­
sistance of the lamp, the lamp lights. The diode is now said to be forward
biased. The diode thus functions as a switch in an electrical circuit. If
the current tries to flow in the preferred direction, the switch is closed.
When the current tries to flow in the reverse direction, the diode acts
as an open switch.
Consider now the analog computer circuit shown in Fig. 7-3. A diode
has been connected in series with the feedback resistor. When the input
voltage is positive, the output voltage is negative, and the diode is con­
ducting current in the forward direction. The forward resistance of the
 DIODE L I M ITERS 139

R m =0
'.>—wv
0
\
SJ e

f i g . 7-3

Inverter with the Diode in the Feedback Path.

diode is small compared to the resistance, R, and hence the inverter

functions in the usual manner. However when the input voltage becomes
negative, the output voltage tries to go positive and the current in the
feedback resistor is now trying to flow through the diode in the reverse
direction. The reverse resistance of the diode is very large compared to
the feedback resistor, R, and hence the voltage gain of the inverter is very
low. A plot of the output voltage, ei, vs. the input voltage, e\, is also
shown in Fig. 7-3. Such a plot of an output variable vs. an input variable
is called a transfer characteristic. The circuit functions as a limiter since
the output voltage is limited to negative values of v oltage. The switching
point of actual diodes may be a few tenths of a volt positive, rather than
zero voltage as is the case for an ideal diode. If the diode is placed in
parallel with the feedback resistor as shown in Fig. 7-4, the nature of the

f i g . 7-4

Inverter with Diode in Parallel with the Feedback System.

limiting is reversed. When the output voltage is positive, the diode is re­
verse biased, and the circuit functions as a normal inverter with unity
gain. When the input voltage goes positive, the output tries to go negative,
but the diode is then forward biased and hence has a low resistance. Since
the gain is the ratio of the feedback resistance to the input resistance, the
 140 NONLINEAR OPERATIO NS

V
+1
H-wv1!

e>>—WV
R

o
SJ o

- V
m = 0

f i g . 7-5

Inverter with a Diode and Battery in the Feedback Path.

gain becomes very small, and the output voltage, ei, is constrained from
going negative.
Consider now the circuit shown in Fig. 7-5, which has added a battery
in series with the diode and the feedback resistor. This battery is assumed
to have a voltage, V, and to have an internal resistance that is negligibly
small compared to R. As long as the output voltage, C2, i s positive, the
diode will be forward biased and so will have a very low resistance. When
the output voltage is equal to — V, the voltage across the diode will be
zero because the summing junction is at a potential very near to zero.
When the input becomes more positive, the diode is reverse biased, and
the output voltage is prevented from going more negative than — V. Al­
though this circuit is simple to understand, it is not convenient to supply
the voltage, V. Therefore, the circuit that is usually used is that shown
in Fig. 7-6. This circuit is a variation of that of Fig. 7-4 where the diode

— 10 volts

m = —1

f i g . 7.6

Practical Limiting Circuit for Positive Voltages.

 DIODE LI MITERS 141

was in parallel with the feedback resistor. Assume that the potentiometer
is set for a ratio of a = 0.40. The voltage at the ARM of the potenti­
ometer would be -4.0 volts when e2 = 0, and therefore the diode is
reverse biased.
The diode will be reverse biased as long as the potential at the arm
of th e potentiometer is negative. When the voltage, e2 has become suffi­
ciently positive, there will be a potential difference of ten volts across the
upper portion of the potentiometer, and the voltage at the arm will be
zero. Under these conditions the voltage across the diode is zero, and
any further increase in the voltage, e2, will cause the diode to be forward
biased. The voltage across the entire potentiometer is e2 + 10. The
voltage across the upper part is:

(1 - a)(e2 + 10) (7-la)

The switching point, or zero voltage condition on the diode, occurs when
this voltage is equal to 10 volts. This condition gives Eq. (7-lb).

(1 - a)(e2 + 10) = 10 (7-lb)

Substitution of 0.4 for a and solving for e2 yields

e2 = +6.67 (7-lc)

For any input voltage more negative than —6.67, the diode will be
forward biased, the feedback resistance will be very low, and the
output voltage is constrained to be not more than +6.67 volts. Rather

+ 10 volts

-V

f i g . 7-7

Limiting Circuit for Negative Output Voltages.

 142 NONLINEAR OPERATION S

f i g . 7-8

Block Diagram Representation of Limiting Circuit.

than computing the ratio to be set on the potentiometer, the usual pro­
cedure is to connect the circuit, supply a —10 volts for e\, and adjust
the potentiometer to give the desired output voltage. The circuit would
be set up as shown in Fig. 7-7 to limit in the other quadrant. The block
diagram representation of such a circuit is shown in Fig. 7-8. These
circuits demonstrate an application for the potentiometers that have
the lower end ungrounded. Note that the anode end of the diode is
connected to the summing junction, SJ, rather than to an input resistor.
Consider now the functional relationship between y and x that is
described by the graph of Fig. 7-9. This function is to be modeled on a
10-volt analog computer. This function can be broken into two parts
called y\ and as shown in Fig. 7-10. If y\ is added to yi, the result

f i g . 7-9

Graphical Definition of y = f(x).

 DIODE LIMITE RS 143

y =/(*)< i
m — 0
7.5

2.5

10 X
f i g . 7-10

Decomposition of f(x) into Two Functions: yi and yi.

will be the desired function. When x equals 2.0, y\ equals 0.5, and }>2
equals 7.5, these two values add to give the desired value of 8.0 as re­
quired by Fig. 7-9. When x equals 10.0, y\ equals 2.5, y2 equals 7.5, and
the sum of y\ and y2 equ als 10.0 as required. These functions can be
created as shown by the block diagram of F ig. 7-11. Inverter 01 produces
— x, which is converted to —yi by potentiometer 03. Amplifier 03 inverts
this —yi to produce the +yi component of y. The inverter 02 with a
gain of 10.0 produces — y2 with a slope of —3.75. The diode and poten­
tiometer 02 constrains y2 to be not more negative than -7.5 volts.
Amplifier 03 inverts this -y2 to produce the component of y.

+10 volts
H

0.375 L

+x
>

f i g . 7-11

Generation off{x) by Summing Negative yi and Negative y i.

144 NONLINEAR OPERATION S

7-2 Diode Function Generators

Although it would be possible to approximate functions by using a

number of simple diode limiter circuits discussed in the previous section,
this would not be particularly convenient. Collections of diodes, re­
sistors, and potentiometers are assembled into units called diode function
generators. Some of these units are designed to be adjustable so that
the user can approximate special functions. For example, the relation
between the variables x: and y is defined by the curve of Fig. 7-12. This

f i g . 7-12

Variable Diode Function Generator.

 MULTIPLIERS 145

curve can be approximated by a series of straight lines such as the dashed

lines shown. In this example five straight line segments are used to
approximate the actual function. The general procedure for setting up a
diode function generator is the following:

a. Move all break poin ts to the upper limit.

b. Adjust the slope and intercept of the first se gment.
c. Move the first breakpoint in to the proper location.
d. Adjust the slo pe of the sec ond segment.
e. Continue until all of the segment s have been set pro perly.

The manuals for your computer should be consulted for specific details
on patching and adjusting the diode function generators on your com­
puter.

7-3 Multipliers

The operation of multiplying two variables together is frequently

desired. This process has been implemented in at least three different
ways on analog computers: quarter-square multipliers, time-division
multiplication, and servo multipliers. With the development of eco­
nomical diode function generators for producing the square of a voltage,
the quarter square multiplier has become very popular. This multiplier
is based on the algebraic identity

XY = [ j x + Y y - ( x - Y y ] (7-3a)
4

This equation can be implemented in principle by the block diagram

shown in Fig. 7-13. This circuit requires five summer-inverters. The

*>

f i g . 7-13

Block-Diagram of Principle of Quarter Square Multiplier.

 146 NONLINEAR OP ERATIONS

scheme is implemented somewhat more efficiently in practice by com­

bining some of these operations. A multiplier may still require three
operational amplifiers unless both the positive and the negative repre­
sentation of both variables is available. The block diagram representation
for quarter-square multipliers such as used by EAI on the TR-20 is
shown in Fig. 7-14. The patch panel connections are shown on the right

mult 7.045

FIG. 7-14
Block Diagram of Actual Multipliers.

side of t his figure. Inverters are required to produce the inverted signals
if they are not otherwise available. The simplified block diagram rep­
resentation is often used to show multipliers, but all of the detail is
required for actually connecting the multiplier. The connections for the
BURR-BROWN Model 600 multiplier is shown in Appendix I-B.
Notice that the multipliers when implemented as a unit for the TR-20
or the BB-600 have a factor of 10 i ncluded in the design. This is done to
permit two voltages of 10 volts to be multiplied together without going
beyond the range of the computing equipment. This factor of ten must
be taken into account in creating the analog model. If the computer
was designed for 100 v olt reference voltages, the dividing factor would
be 100 so that two voltages of 100 volts would produce a product which
was 100 volts. The quarter square multiplier units can obviously be used
to produce the square of a variable by making Y = X. Diode function
generators are available for producing the square of a variable. These
special units may be more economical for some applications. The quarter
square multipliers can also be used to implement the process of division
as shown in Fig. 7-15.
 MULTIPLIERS 147

FIG. 7-15
Block Diagram of Analog Division Circuit Using a Multiplier.

The operation of the circuit is evident from the constraint that the
current, h, is equal to the current, i\.

. _ Y _ . QX
11 R 12 10 R

101 Q = - (7-3b)

This process of division has the property that the quotient of two voltages
which are 10 volts will be a voltage which is also 10 volts. This amplitude
scaling must also be taken into account when the analog model is created.
The division process is implemented on the TR-20 computer as shown
in the diagram of Fig. 7-16. The nature of the process requires that the

FIG. 7-16
Circuit for Division on the TR-20 Analog Computer.
 148 NONLINEAR OPERA TIONS

indicated limitations be placed on the variables. See the manual for

your analog computer for details of this and other applications such
as square root of the quarter-square or other multipliers available to
you. The principles outlined here for the applications of the quarter-
square multiplier also apply to time-division multipliers and to servo
multipliers. An example showing the use of the multiplier will be con­
sidered after the sine-cosine generator is presented in the next section.

7-4 Sine-Cosine Generator

The problem of the pendulum involved the need for a sine function
in order to model the pendulum for other than small deflections. Sine
and cosine functions arise in many places where forces or motions must
be broken into components. Special diode function generators are made
for this purpose. These functions can also be generated by electro­
mechanical servo type devices with special potentiometers which have
an output voltage proportional to the sine or the cosine of the angle of
rotation of the shaft of the potentiometer. The diode type generators
are generally preferred today because they can operate at much higher
speeds than the servo type generators. This process also must be scaled
to produce the reference voltage when the angle is 90°. The generation
of sine and cosine functions can be represented as shown in Fig. 7-17

1oo \ cos (X)

IT
77" sin (X)

FIG. 7-17
Generation of Sine and Cosine Functions.

for a computer with reference voltage of 10 volts. The angle must be

scaled according to the reference voltage of t he computer and the range
of the input angle for which the generator is designed. Typically the
range of i nput angle is ± 180° o r expressed in radians ± ir. Th e imple­
mentation of the sine-cosine generator usually requires the use of several
summer-inverters in a manner similar to that for the multiplication and
division circuits previously considered. Consult the manuals for your
particular computer for details on patching the sine-cosine generator.
 SINE-COSINE GENERATOR 149

EXAMPLE 7-1

Consider the problem of computing the position or a vehicle in terms

of the speed and bearing of t he vehicle. The parameters of th e problem
are as follows:

Speed: 0 to 1800 knots

Bearing: ±180° relative to north
Distance: N-S ± 100 nautical miles
E-W ± 70 nautical miles

The knot as a unit of speed is one nautical mile per hour. For conve­
nience take the same scale factor on distance in both the E-W and N-S
direction.

= K,- 'jff K, = 10 miles/volt (7-4a)

The amplitude scale factor for speed is as follows:

Ks = = 180 knots/volt (7_4b)

The amplitude scale factor for the angle /3 is determined by the con­
struction of the sine-cosine generator. Let us assume that a 10-volt
input corresponds to 180°.

„ _ max |/3| _ 180 degrees

" ~ Cm 10 volts
= 18 degrees/volt (7-4c)

Assume that the computer voltages X, Y, S, and B represent problem

variables jc, y, speed, and bearing. Then the amplitude scale factors
relate problem variables to the analog model variables as follows:

x = 10T

j; = 107

^ = 180S
0 = 18B (7-4d)
 150 NONLINEAR O PERATIONS

W-

FIG. 7-18
Geometry of Relations Between Speed, Bearing, and Velocity Components.

The speed s can be resolved into components as shown in Fig. 7-18.

The components dx/dt and dy/dt of the speed, s, of the vehicle are
computed as follows:

dx . „
•Jt = s sin ,3

dy
di = scos (7-4e)

The x and y coordinates of the vehicle are computed by integrating these

components:

x= dx
dt + x(0) = f 0 sin B)dt + x(0)
/

y = / % dt + XO) = J 0 cos (3)dt + y(0) (7-40

Substitution of the amplitude scaling Eqs. (7-4d) into Eqs. (7-4e) and
dividing by ten yields the equations for the analog model.

X = J 185 sin (18B)dt + Z(0)

Y = J 185 cos (l$B)dt + 7(0) (7-4g)

 S I N E - C O S I N E GE N E R A T O R 151

£•, = +10 for 0 < /? < +18 0°

E | = - 1 0 for - 1 8 0 ° < B< 0

FIG. 7-19
Block Diagram of Analog Computer Model Demonstrating the Use of Multi­
pliers and Sine-Cosine Generator.

The block diagram of Fig. 7-19 shows how the analog computer model
is c onnected. The polarity, plus or minus, of the voltage E conne cted
to the input side of potentiometer 01 determines whether the bearing
(8 is east of north or west of n orth. The sine-cosine generator G1 produces
the functions of 185 degrees multiplied by the reference voltage of 10
volts. Notice that this factor of ten and the divisor of t en on the multi­
pliers, Ml and M2, cancel to give the inputs shown for the integrators
01 and 02. Gains of 18 are used on the integrators to meet the conditions
specified b y Eqs. (7-4g).
As a check on the analog model assume that the vehicle is moving at a
speed of 900 knots at a bearing of 90° East of North. Further assume,
for simplicity, that x(0) = j>(0) = 0- T he ratio set on potentiometer 01
is 0.50 and the voltage B is 5 volts. This 5 volts input to the sine-cosine
generator G1 produces functions of 90°, and hence Ei eq uals zero and
E3 = 10 volts. The voltage S is minus 5 volts. The voltage E4 is zero
because the voltage E2 is zero. The voltage Es is as follows:

E5 = = 1Q(~5) _ _5.o volts (7-4h)

5 10 10
 152 NONLINEAR O PERATIONS

The voltage Y remains at zero because the input to the integrator 01 is

zero. The voltage X increases at the rate of 90 volts per second since
the input voltage is 18( — 5) or —90 volts. Thus after one second of
operation (except that the computer would overload after approximately
0.1 second of operation) the voltage X would have increased from 0 to
90 volts. This voltage corresponds to a distance of 900 nautical miles.
A distance of 900 nautical miles should be reached at 900 knots after
one hour, not one sec. The error arose because the unit of speed, the
knot is based on the hour as the unit of ti me, and the unit of time used
to compute the gains of t he integrators is the sec. The analog computer
model is thus running 3600 times faster than real time. One hour or
3600 seconds of problem time corresponds to one sec of c omputer time.
This kind of e rror is easy to make unless care is taken with the units of
rate variables.
The gains on the integrators should be reduced by a factor of 3600
to 0.005 as shown by Eq. (7-4i)
10 1
= 0.005 (7-4i)
3600 200

This value will make 1 sec of problem time equal 1 sec of compu ter time.
A more appropriate scaling might be to make 1 minute of problem time
equal to 1 sec of computer time. Such scaling gives integrator gains of
0.30, which are more feasible than either of the previous values of 0.005
or 18.

7-5 Comparators

In some problems it is necessary to compare two variables. For

example, suppose that we were interested in creating an analog model
for a sump pump operation. Water collects in the sump until the level
of the water reaches a depth, U, at which time the pump is turned on
and water is pumped out. When the water level falls to a depth, L, the
pump is turned off. The operation of a comparator can be thought of
as that of a polarized relay. An ordinary relay operates when a suffi­
cient current flows through the winding in either direction. A polarized
relay, however, operates only when there is sufficient current flowing in a
particular direction. The symbol frequently used on the patch panel of
alog computers for comparators is a relay symbol. The two input
TTTTTTTTTTTTTTTTT5 fninals are distinguished by polarity marks, + or —, or perhaps by
eeeeeeeeeeeeeeee^t^erent num bers. See Fig. 7-20. If the voltage applied to the terminal
 COMPARATORS 153

FIG. 7-20
Representation of Analog Comparator.

marked + is indeed positive with respect to the other terminal, then

the relay in the comparator is energized and the circuit is completed
through the normally open contacts from a to c. If the voltage applied
to the terminal marked + is negative with respect to the other terminal,
then the relay is not energized and the circuit is completed through the
normally closed relay contacts from a to b. Comparators may have more
than one set of c ontacts, and some comparators are all solid state, that
is, these comparators do not have electromechanical relays to perform
the switching. These solid state comparators operate much faster than

ground + 10 volts + 10 volts

FIG. 7-21
Analog Model for Sump Pump Problem.
 154 NONLINEAR OPE RATIONS

relays and so are particularly desirable for REP-OP type solutions.

Consult the manual on your computer for details on the terminal mark­
ing of the comparators you are using. It is also important to note the
particular limitations that may exist as to the amount of current that
the contacts can safely carry. For example, in the analog model shown
in Fig. 7-21, it may not be desirable to place the contacts on comparator
K3 between the +10 reference voltage and the HI end of POT 03 al­
though the switching effect would be the same. The current to the HI
terminal is much greater than is the current which flows in the ARM of
the potentiometer.

EXAMPLE 7-2
Consider the problem of the sump pump proposed earlier. A block
diagram of the analog model is shown in Fig. 7-21. Potentiometer 04
regulates the rate at which water flows into the tank and potentiometer
03 regulates the rate at which water is pumped out of the tank. When the
contacts K3a are in the position shown, the pump is turned off because
the ARM of potentiometer 03 is not connected to the integrator input.
When the contacts K3a shift to the other position, a positive voltage
is applied to the integrator input that will reduce the voltage, H, rep­
resenting the depth of water in the sump. Three comparators are re­
quired to model this system. Comparator K\ compares the height of
water, H, with the lower reference level, L, set by potentiometer 01.
When the water is deeper than L, the comparator energizes its relay
causing the contacts to take the position opposite to that shown on the
sketch. The comparator K2 compares the depth of the water, H, with
the upper reference level, U, which is set by potentiometer 02. When the
water is above the level, U, both A3 and K2 are energized. This completes
the path from the +10 reference voltage through K2a and Kla which
energizes comparator A3. When A3 is energized, the "pump" is turn ed
on and the water level be gins to fall. The contacts K3b prevent the pump
from being turned off when the water level falls below U. When the
water level falls below the lower limit, L, comparator ATI is de-energized
which opens the circuit through Kla to comparator A3. When this
circuit is opened, K3 is de-energized and the "pump" is turned off by
opening the circuit through K3a.

7-6 Exercises
1. Sketch the transfer characteristic for the analog circuit shown in Fig. 7-22.
2. Sketch the block diagram for an analog circuit that will provide the transfer
characteristic shown in Fig. 7-23.
 EXERCISES 155
R

FIG. 7-22

y = /(*)

10

~ 7 ]

/}
71

\ v
-10 -3
2 10 X

-6

-8

FIG. 7-23

3. Check the manuals for your computer and sketch the panel connections for
a. forming the product of two variables,
b. forming the square of a variable which is always positive,
c. forming the square of a variable which is always negative,
d. forming the sine and cosine function of a variable,
e. forming the quotient of two variables for which the divisor is always
positive,
f. forming the absolute value of a variable.
4. Connect up the analog model of the sump pump and observe how the
cycling of the pump is influenced by the rate at which water runs into the
sump. Also investigate the effect of reducing the difference between the
turn-on level, L, and the turn-off level, U.
 BIBLIOGRAPHY

Textbooks

1. Ashley, J. Robert, Introduction to Analog Computation, John Wiley &

Sons, New York, 1965.
2. Jenness, Roger R., Analog Computation and Simulation: Laboratory
Approach, Allyn and Bacon, Boston, 1965.
3. Korn, G. A.; Korn, T. M., Electronic Analog Compu ters, McGraw-Hill
Book Co., New York, 1956.
4. Peterson, Gerald R., Basic An alog Computation, The Macmillan Com­
pany, New York, 1967.
5. Rekoff, Michael G. Jr., Analog Computer Programming, Charles E.
Merrill Publishing Co., Columbus, Ohio, 1967.
6. Stice, James E.; Swanson, Bernet S., Electronic Analog Computer Primer,
Blaisdell Publishing Co., Waltham, Mass., 1965.
7. Warfield, John N., Introduction to Analog Computers, P rentice-Hall,
Englewood Cliffs, N.J., 1959.

156
 BIBLIOGRAPHY 157

Periodicals

IEEE Transactions on Education, Institute of Electrical and Electronic

Engineers, 345 E. 47th St., New York, N.Y. 00017.
IEEE Transactions on Electronic Computers, I nstitute of Electrical and
Electronic Engineers, 345 E. 47th St., New York, N.Y. 10017.
Newsletter, Analog/Hybrid Computer Users Group, P.O. Box 582, Prince­
ton, New Jersey 08540.
Simulation, Simulation Councils, Inc., P.O. Box 2228, La Jolla, California,
92037.
 A. BASIC OPERATION O F T HE
APPENDIX ELECTRONIC ASSOCIATES T R-20
ANALOG C OMPUTER

I-A-l General Description

The Electronic Associates TR-20 is an all solid-state

(transistorized) analog computer. The reference levels are ± 10 volts. The
standard computer has ten chopper stabilized operational amplifiers, four
integrator networks, ten potentiometers, a reference potentiometer, a volt
meter, a set of overload indicators, and a mode control switch. This basic
unit can be expanded up to a total of twenty operational amplifiers and
twenty potentiometers. Integrator networks, comparators, function
generators, multiplier networks, and other special functions can also be
added. All of these additional functions can be added without any addi­
tional wiring in the console because the connectors into which these

158
 ELECTRONIC ASSOCIATES TR -20 ANALOG COMPUTER 159

modules plug are wired at the factory when the computer is manu­
factured. The integrator networks may be obtained with provisions for
REP-OP (repetitive operation) for an additional charge. A REP-OP
control unit must also be added to provide for this type of operation. All
of th e components for the computing networks are contained inside the
computing modules. The connections are made by plu gging patch cords
in an appropriate manner between terminals on the front panel, or by
plugging in special jumper plugs. A prepatch panel can be added to per­
mit programs to be patched together away from the computer, and then
the panel is inserted to connect the units on the computer as desired. This
prepatch capability permits wired programs to be sa ved, and the com­
puter can be used for other problems by using a different prepatch panel.
The sections which follow describe how the patching (connecting) of the
modules is done on the TR-20 analog computer and how the computer
is operated.

I-A-2 Multiplication by a Constant

The potentiometers are mounted two to a panel, and the connections

are shown in Fig. 1-1. These potentiometers are multiturn potentiometers
and are set to a desired ratio by comparison to a master potentiometer.
The lower end of the potentiometer whose terminals are on the left side is
internally connected to ground or zero reference. The lower end of the
other potentiometer must be grounded by a shorting plug for normal
operation. The ground terminal is located to the left of the terminal for
the lower end of the right potentiometer. To set a ratio of 0.75 on po­
tentiometer 1 p roceed as follows:

a. Set the switch on the master potentiometer to +10.

b. Set the master potentiometer for a reading of 0.75.
c. Turn the voltmeter function switch to POT BUS.
d. Press the button beside the knob for potentiometer 1 and turn the po­
tentiometer knob to obtain a zero reading on the voltmeter.

If the output of the potentiometer is connected to an input resistor, the

effect of "loading" is automatically compensated. The upper knob con­
trols the potentiometer connected to the left set of terminals and the
lower knob controls the potentiometer connected to the right set of
terminals. The voltmeter will deflect off sca le unless the potentiometers
are very nearly to the same ratio. If the meter deflects down scale, the
knob on the potentiometer should be turned clockwise to equalize the
 160 APPENDIX I- A

Note: Total resistance of the potentiometer

is 5000 ohms

FIG. 1-1
Potentiometer Panel on TR-20.

ratios. (Counterclockwise if the meter deflects up scale.) The meter is

protected such that off-scale deflection does not damage the meter in the
POT BUS mode of operation.

I-A-3 Inversion and Summation

The TR-20 computer is connected for summation or inversion as shown

in Fig. 1-2. The resistors connected to terminals marked "1" have a re­
sistance of 100,000 ohms. The resistors connected to the terminals marked
 ELECTRONIC ASSOCIATES T R-20 ANALOG COMPUTER 161

FIG. 1-2

Amplifier Panel on TR-20 With Connections for Summation.

"10" have a resistance of 10,000 ohms. The terminals in the red area of
the panel are all connected together and are the output terminals of the
operational amplifier. The upper amplifier, 1, has the summing junction
connected to the input and the output connected to a feedback resistor by
means of a double bottle plug. The numbers by the input terminals (green
region on the panel) are the factors by which the input is multiplied in the
process of summation. If only one input is used, then the operation is that
of inversion. The lower amplifier, 2, has been connected by using two
single bottle plugs. The feedback resistor is now one tenth the previous
value, and so the gains are reduced by a factor of ten. The student should
verify these gains from the principles of Chapter 2.
162 APPENDIX I-A

I-A-4 Integration

n.T'Ti t'!"®, ca P acitors on the TR-20 are located in units named

DUAL INT. The lower pair of terminals on the INT unit must be shorted
by single bottle plugs as shown in Fig. 1-3. These connections provide

SJ

l O^W h , 0o-
1 B O o

1 o
SJ o

lo-W"
,0s IC o

10o^/VJ -10
volts
SJ o- Bo-
r-o0

1 o-

1 o-

lO o-

10 o-

1 2
dual D .C. amp. dual int. attn .

<( - 1 0 volts

V = -[/(A + 10 B)di-10 K]

FIG. 1-3

Block Diagram and Panel Connections for TR-20 for Integration.

 ELECTRONIC ASSOCIATES TR-20 ANALOG COMPUTER 163

for the operation of t he relays which switch the computer from RESET
to HOLD to OPERATE. Three connections must be made between the
operational amplifier and the integrator unit. The pairs of terminals
marked B, SJ, and O must be c onnected together as shown in Fig. 1-3.
The integrator gains are the numbers marked by the input terminals. The
initial condition voltage is connected to the terminal marked IC. The
sign change that is inherent in all analog operations occurs here also, and
therefore, —5 volts connected to the IC terminal causes the initial condi­
tion to be +5 volts. The pair of terminals just below the output terminals
must be connected by a single bottle plug for the repetitive mode of opera­
tion (REP-OP). The integrating capacitor has a value of 10 micro-farads
when the computer operates in the normal mode. This value of capacitor
and an input resistor of 100,000 oh ms gives a gain of o ne as described in
Chapter 3. When the computer is operated in the REP-OP mode, the
value of all integrating capacitors is 0.02 micro-farad, and the gains of
all integrators are multiplied by 500.

I-A-5 Normal Operation of TR-20

1. Be certain that feedback is provided on all amplifiers, that is, each am­
plifier should be connected for operation either as a summer with bottle
plugs, or as an integrator by removing the bottle plug and patching to
an integrator unit.
2. Make other connections as required for the problem.
3. Set the MODE control to RESET.
4. Set the COMPUTE TIME control to OFF.
5. Set the VOLTMETER range switch to 10.
6. Turn the POWER switch ON. The overload indicators will light
temporarily, but should go out after a few sec.
7. Set the MASTER POT switch to +10.
8. Set the VOLTMETER function switch to POT BUS.
9. Adjust each potentiometer for a zero reading on the voltmeter with the
POT BALANCE procedure described above.
10. Set the VOLTMETER function switch to AMPL.
11. Select the amplifier whose output is to be observed by setting the
AMPL switch to the number of that amplifier.
12. Set the MODE control to OPER, and the computer solves the problem
for which it is programmed. The solution is displayed by the voltmeter,
or plotted by an attached plotter. After the desired solution has been
generated, the MODE control should be placed in either HOLD or
RESET. The HOLD mode stops the solution and the value of all
voltages remains constant. The RESET mode re-initializes all initial
conditions in preparation for another solution.
164 APPENDIX I-A

I-A-6 Repetitive Mode of Operation of TR-20

When the computer is operated in the REP-OP mode, the integrator

gains are multiplied by 500, and therefore the solution which requires
10 sec in the normal mode is generated in (10/500) or 20 msec. The
COMPUTE TIME control determines how long the computer will be in
the OPERATE mode. Following this period of time, the computer is
automatically placed in the RESET mode to prepare for generating the
next solution, and the computer switched again automatically to the
COMPUTE mode. Since the solutions are generated so rapidly, an oscillo­
scope type display unit must be used to display the solutions. Problem
parameters can be adjusted while the computer is in operation and the
effect of the change observed immediately. The overload indicators glow
dimly when the computer is operated in the REP-OP mode. Any reading
of amplifier output voltages by the voltmeter will be i n error when the
computer is in the REP-OP mode because the voltmeter cannot respond
accurately to the rapidly changing voltages.
The steps 1 to 9 given for the normal mode of operation are the same
for setting up the computer for REP-OP operation. Then proceed as
follows:

10. Select the amplifier output to be observed and connect it to the display
unit. With some display units several different voltages can be observed
simultaneously.
11. Set the COMPUTE TIME control to 20 msec.
12. Set the MODE CONTROL to REP-OP and observe the solutions.
13. If a longer solution time is desired, select the appropriate time on the
COMPUTE TIME control.
14. Potentiometers can be adjusted to give the desired form of solution
while the computer is operating in the REP-OP mode. The numerical
value of this ratio can be determined by setting up the POT BAL opera­
tion, but adjust the MASTER POT for a zero deflection while pressing
the appropriate potentiometer button.

I-A-7 Reference

TR-20 Operator's Reference Manual, Electronic Associates Inc., Long

Branch, N.J.
 B. BASIC OPERATION O F THE
APPENDIX I BURR-BROWN MODEL 6 00
ANALOG C OMPUTER

I-B-l General Description

The Burr-Brown Model 600 is a solid-state, table-top

size, analog computer. The reference levels are ±10 volts. This model has
six operational amplifiers which may be patched either as summers or
integrators. Four more amplifiers are required in conjunction with the
two multiplier units, or if t he multiplier units are not being used, these
amplifiers may be used independently as unity gain inverters. The panel is
laid out so that many connections can be made by a double-banana plug
wired as a shorting plug. Other connections are made with banana plug
patch cords. Ten coefficient setting potentiometers are provided with
locking, calibrated dials. These potentiometers may be set with reference
to a REFERENCE POTENTIOMETER by a NULL Balance method.

165
 166 APPENDIX L -B

Five of the potentiometers may be operated with the lower end un­
grounded if desired. The computer may be operated in the following
modes: PATCH, RESET (this is also POT SET), COMPUTE, and
HOLD. These modes are selected by pressing a pushbutton along the
lower edge of the console. A red indicator marked DO NOT PATCH is
lighted when the computer is in any mode but the PATCH mode. In the
PATCH mode the reference voltages are removed from the front panel. A
voltmeter is located in the center of the upper panel and has three ranges;
NULL, X 1, and X 10. The desired range is selected by pressing the ap­
propriate pushbutton. The pushbuttons that have been pressed remain
lighted until another button in the group is pressed. A row of OVER­
LOAD INDICATOR lights are found on the left side of the upper panel.
If any of the amplifiers one through six has an output voltage in excess of
+10 volts, or less than —10 volts, the corresponding indicator light lights.
This is an indication that the amplifiers are operating outside their ratings
and that the problem solution may be in error. These lights will come on
momentarily when the computer is first turned on, but should go out in a
few seconds. If an overload indication persists, the associated amplifier
probably does not have a proper feedback connection with either a re­
sistor or a capacitor. Just below the OVERLOAD INDICATORS is a
row of pushbuttons and jacks. The lower row of jacks connect to the
ground terminal of th e patch panel. The two upper terminals associated
with the buttons marked AUX are connected to the voltmeter when the
AUX button is pressed. The remaining jacks are connected to the output
terminals of the amplifier having the same number. These jacks may be
used to connect these output voltages to external equipment such as
recorders. When the numbered pushbutton is pressed, the output of that
amplifier is connected to the voltmeter circuit. The output terminal of
each amplifier is also connected to the jacks on each side of the patch
panel. The BUS 1 j acks on each side are connected together internally.
The BUS 2 jacks are also connected together internally. These BUS jacks
and output jacks are useful when two or more Model 600 computers are
used jointly for solving larger problems. A control cable plugs in on the
back to connect the mode control circuits of two or more Model 600
computers together. The patch panel is well laid out and diagramed in
such a way that the proper patching of the elements is virtually obvious.

I-B-2 Multiplication by a Constant

A portion of t he potentiometer panel is shown in Fig. 1-4 below. The

distance between adjacent jacks is f- in. so that adjacent terminals can be
 'BURR-BROWN MODEL 600 ANALOG COMPUTER" 167

o o

PI P2

10> ©— 1 "l0> ©—I

FIG. 1-4
Panel Connections and Block Diagram Representation of Potentiometers.

connected together by inserting double banana shorting plugs. If shorting

plugs a re inserted as indicated by the dotted lines, the HI end of PI is
connected to +10 volts and the HI end of P2 is connected to —10 volts.
The LO end of P2 must be grounded for normal operation as shown by
the dotted line between the lower end and ground. The potentiometer can
be set with reference to the master pot, NULL, by the following pro­
cedure:

1. Place the computer in the RESET mode.

2. Set the desired ratio on the NULL potentiometer.
3. Set the dial of the potentiometer to be adjusted to approximately the
same reading.
4. Press the meter selector marked NULL.
5. Press the small button located above the potentiometer to be set.
6. Adjust the potentiometer to give zero deflection on the meter.

The potentiometer should be completely patched as it will be u sed in the

analog model prior to these adjustments so that potentiometer loading
will be properly compensated for. The lower end of the potentiometer
must be connected to ground. If the potentiometer is to be used in some
other fashion, this NULL BALANCE procedure cannot be used.
 168 APPENDIX I-B

I-B-3 Inversion and Summation

The amplifiers associated with the multipliers may be used as inverters

with a gain of one. The two resistors associated with the inverter symbol
are connected together by plugging in a shorting plug. The left terminal of
the left resistor is the input terminal for the inverter and is marked P in
Fig. 1-5. The output is available at either of the two jacks on the right.
One of these jacks is just to the right of the vertex of the inverter symbol

-o-
p

o o

o o

•o >~Q
Q

FIG. 1-5

Panel Connections and Block Diagram Representation for Inverters.

 'BURR-BROWN MODEL 600 ANALOG COMPUTER"

and the other is just to the left. These two jacks are connected together
internally. The lower inverter of the pair is connected in precisely the same
manner. The corresponding block diagram is shown in the lower portion
of Fig. 1-5.
A sketch of the portion of t he patch panel associated with a summer-
integrator is shown in Fig. 1-6. A shorting plug is connected between the
jack marked R a nd the jack marked SJ. Note that these terminals are not
marked on the panel of the actual computer. When this connection is
made, the simil arity between the appearance of t he patch panel and the
schematic diagram of a summer should be evident. Three variables, P, Q,
and R, are shown connected to inputs of this summer. The output is
labeled as T. The block diagram representation of t his summer is shown
at the bottom of Fig. 1-6.

FIG. 1-6

Panel Connections and Block Diagram Representation for a Summer.

I-B-4 Integration

A portion of the patch panel associated with a summer-integrator is

shown in Fig. 1-7. The shorting plug has been moved to connect the left
 170 APPENDIX I-B

O
PI

O ;
•

FIG. I - 7
Panel Connections and Block Diagram Representation for an Integrator-Summer'

terminal C of the feedback capacitor and the summing junction SJ. Again
the similarity between the appearance of the patch panel and the sche­
matic diagram of the integrator should be very evident. If an IC voltage is
needed on this integrator, potentiometer P\ should be us ed as shown in
the upper portion of Fig. 1-7. T he block diagram representation of this
INTEGRATOR-SUMMER is shown in the lower portion of the Fig. 1-7.

I-B-5 Multiplier

The Burr-Brown Model 600 computer has two multipliers as standard

equipment. These multipliers permit the simulation of nonlinear problems
 172 APPENDIX I-B

which involve the products, quotients, and square roots of variables. The
two inverters associated with a multiplier plus another amplifier operated
as an inverter are required to implement multiplication. The connections
for a multiplier are shown in Fig. 1-8. The construction of the multiplier
is such that the output voltage is the product of the input voltages
divided by ten. Thus the product of a voltage of 5 v olts and another of 7
volts is not 35 volts, but rather is 3.5 volts. If the same voltage is supplied
to both input terminals on the multiplier the output voltage is 0.1 times
the square of the input voltage. For example, if 6 volts is fed into both
inputs the output is 3.6 volts. Division and square root can also be per ­
formed. See the Burr-Brown manuals for details.

l-B-6 General Operation

1. The first step in setting up a simulation on this analog computer is to

patch the summer-integrators as summers or integrators according to
the assignments made on your block diagram of the analog model which
you are setting up. All operational amplifiers must have a feedback ele­
ment connected to prevent overloads when the computer is turned on.
2. Turn on the computer by pressing the button marked POWER.
3. Press the button marked PATCH to place the computer in the patch
mode.
4. Connect the computer elements according to your block diagram.
5. Adjust each potentiometer to the desired ratio according to the pro­
cedure given in Section I-B-2.
6. Press the X 10 range switch on the voltmeter and press the appropriate
AMPLIFIER OUTPUT MONITOR button to observe the result of the
simulation.
7. Press the button marked COMPUTE to place the computer in opera­
tion, and observe the solution on the voltmeter. Check other appropriate
voltages to determine if the model is performing satisfactorily.
8. Make any changes in scaling that may be required.
9. Connect the recording equipment and record the desired solutions.

I-B-7 References

Burr-Brown Model 6 00 Educational An alog Simulator Operator's Hand book,

Burr-Brown Research Corporation, Tuscon, Arizona, 1966.
Burr-Brown Model 600 Educational An alog Simulato r Laboratory Manual,
Burr-Brown Research Corporation, Tuscon, Arizona, 1966.
APPENDIX I
C. ELECTRONIC ASSOCIATES
TR-48 A NALOG C OMPUTER

I-C-l General Description

The TR-48 is an all solid-state, precision analog com­

puter. The console is wired to accommodate up to 58 operational ampli­
fiers, 115 potentiometers, 24 integrators, and a large number of nonlinear
components. The panel is divided into 12 areas, and each area has five
positions for installing computing equipment. The first and third posi­
tions accept dual operational amplifier units. The second position
accommodates multipliers, diode function generators, and other such
units. The fourth position is wired for dual integrator networks or non­
linear units except diode function generators. The fifth position is wired
for potentiometer units. The potentiometers together with control knobs
are mounted on a panel to the right of the patch panel. The panel to the
left of the patch panel has all of the control switches, voltmeters, and

173
 174 A P P E N D I X l- C

REP-OP display unit. The operating modes for this computer are: POT
SET, RESET, HOLD, OPERATE, and REPETITIVE OPERATION.
The desired mode is selected by pressing one of the MODE buttons along
the lower edge of the control panel. The integrator units may be obtained
either with relay mode control, or with electronic mode control for greater
precision at higher speeds. An array of t hree columns of buttons permits
the selection of any amplifier output, any coefficient potentiometer, or
input trunk for monitoring on the digital voltmeter, or on the panel
meter. The digital voltmeter has automatic range and polarity selection.
The zero center panel meter can also be connected to the selector circuit,
used to manually balance the amplifiers if needed, or to monitor the
power supply voltages. The voltmeter has five ranges selected by a rotary
switch. The time duration of the solution when in the REP-OP mode is
selected by a rotary switch located in the center of the panel. A vernier for
the solution time is located in the center of the range switch knob. Below
this is located the control for the motor which engages and disengages the
patch panel. On the lower left side of the control panel is the array of
OVERLOAD indicators. Below this is the power switch which turns the
computer on and off. A ll computing components are located behind the
panel and connections are made by means of bottle plugs and patch
cords. The input terminals for the REP-OP display unit are terminated at
convenient locations on the patch panel. Other recording and display
equipment may also be terminated on the patch panel. Mode pushbuttons
are illuminated when depressed to show clearly the condition of the
computer. When a button is depressed to make an alternate selection, the
prior connection is broken, the new one made and the new indicator light
turned on.

I-C-2 Multiplication by a Constant

A sketch of a C oefficient Setting Potentiometer Group is shown in Fig.

1-9. There are five potentiometers in a group. Four of these are perma­
nently grounded on the lower end and the fifth may be grounded if
desired by a single bottle plug. In EAI terminology a double banana
shorting plug is a single bottle plug. The right half of the panel for this
group contains five jacks for the T-10 volt reference supply. This area of
the panel is colored red. The middle portion of the right side is colored
black and has two grounded jacks. The lower right side contains five jacks
for the —10 vol t reference supply. This portion of the panel is colored
yellow. The potentiometers are set by the following procedure.
 "ELECTRONIC ASSO CIATES TR-48 ANALOG COMPU TER" 175

1-5

atten 42.2 83

FIG. 1-9
Panel Connections for Coefficient Setting Potentiometer Group 42.283.

1. Press POT SET.

2. Select the pote ntiometer by number with the pus h button selector.
3. Adjust the potentiometer until the desired ratio is indicated on the
digital voltmeter.

If the computer does not have a digital voltmeter, the potentiometers can
be set with reference to a master potentiometer in the same manner as on
the TR-20 computer.

l-C-3 Inversion and Summation

The terminal arrangement for the dual amplifier 6.614 is the same as i ;;
that for the amplifiers used on the EAI TR-20 analog computer. The quad
 176 APPENDIX I-C

amplifier modules have a different terminal arrangement with only two

unity gain inputs and two output terminals. When the double bottle plug
is in place to make an inverter with the QUAD amplifiers, there remain
only one input terminal and one output terminal. These amplifiers are in­
tended primiarly for use with nonlinear equipment where one input and
one output are adequate. The QUAD amplifiers may be mounted in the
lower row in place of five dual amplifiers to expand from 48 operational
amplifiers to 58 operational amplifiers. See Fig. 1-10 for a sketch of the
panel connections of the DUAL amplifiers and the QUAD amplifiers.

I-C-4 Integration

Fig. 1-11 shows a sketch of the patch panel connections for a DUAL
INTEGRATOR unit (12.1322) such as is used for electronic mode con­
trol. A bottle plug with six plugs and three circuits is used to connect the
B, SJ, and O terminals of the integrator unit with the corresponding
terminals of an adjacent operational amplifier. Single circuit bottle plugs
are used to patch the signals to the mode switches. Four are required: one
each connecting the pairs of t erminals marked IC, OP, and 0.1/3. If the
gain of an individual integrator is to be increased by a factor of ten, a
single bottle plug is removed from the pair of terminals marked 0.1/3 on
that integrator. The terminal marked SJ' makes the summing junction
available for special circuits when the other SJ terminal is in use. The
negative of the desired IC voltage is supplied to the single terminal marked
IC. This connection can be made by a single circuit bottle plug if the
proper potentiometer is assigned this function.

I-C-5 Nonlinear Modules

A very large number of modules for performing various nonlinear

functions are available for the TR-48. The major kinds of units are as
follows: Dual Limiters, Quarter Square Multipliers, Fixed Diode Func­
tion Generators, Variable Diode Function Generators, Log Diode
Function Generators, Sine-Cosine Generators, Comparators, and Func­
tion Switches. See the EAI TR-48 OPERATOR'S MANUAL for details
on the use of these units.
 ELECTRONIC ASSOCIATES T R-48 ANALOG COMPUTER' 177

> ~T

SJ<£—

'V IC^VW V

SJo- So-

c loA/V-
1 °A/V-

SJo- So- SJ o- SO-

l o-A/V-" ioA/V- o-
<7
l o-VV-" io-W-
io-A/\/-• 57 o So-

lOoAAr-' 10A/V-
<7
iooAA^ 10A/V-

dual DC quad DC
amp. 6.614 amp. 6.704-1

FIG. 1-10

Connections for Dual DC Amplifier (6.614) and Quad DC Amplifier

(o. 7U4-1).

I-C-6 General Operation

1. Make certain that appropriate bottle plugs are in place to connect all
operational amplifiers as either a summer, an inverter, or as an in­
tegrator. All amplifiers must have a feedback element connected to
prevent overloads. It is also necessary to see that all single bottle plugs
required for control of the integrators are in place.
2. Turn on the computer by pressing the POWER switch.
3. Place the computer in the POT SET mode.
178 APPENDIX I-C

Hi'
6 6 6 6 0 - - 0 6 6 6 6 6
/ +

~VV~j|' [-A/V-||i 1—A/V—J|i rA/V-jl

y> />o 6 6 6 6 6 6 c

o f f
/ /
/ /
[2 ! si
lo 1^
Io I
\
\
\
\
05 o ^ g \> \> \>
/>o Cr*\ O \>

rO rC>

V) 6-<' 6
'ELECTRONIC ASSOCIATES TR -48 ANALOG COM PUTER" 179

4. Patch the computing elements as defined by your block diagram. Careful

assignment of computing elements will maximize the number of con­
nections that can be made by the bottle plugs and hence reduce the
clutter of leads on the patch panel. It is helpful to patch all interconnec­
tions by functional groups and then to make the interconnections be­
tween groups.
5. Adjust all potentiometers by the POT SET procedure described above.
6. Select appropriate voltages for monitoring and check the system re­
sponse when the computer is placed in the COMPUTE mode. The REP-
OP mode is particularly helpful for checkout of the operation of analog
models because the entire solution can be viewed at once, and the effect
of changes in parameters is readily apparent.
7. After the model has been determ ined to function properly, record the
variables of intere st for the range of parameter s required.
APPENDIX „ DERIVATION O F
ANALOG C OMPUTER G AINS

11-A Inversion

The detailed connections of an operational amplifier as

an inverter are shown in Fig. II-1 below. Voltages and currents are identi­
fied, and their reference directions are shown. The voltage e\ is positive if
the terminal marked + is indeed positive with respect to the terminal
marked —. Th e current ia is positive if in deed current flows through Ra
from left to right. Likewise, the voltage e0 is negative if the terminal
marked -j- is indeed at a negative potential with respect to the terminal
marked —. K irchhoff's current law states that

The sum of the currents flowing into a junction is equal to the sum of the
currents flowing out of that junction.

180
 DERIVATION OF ANALOG CO MPUTER GAINS 181

FIG. 11- 1
Schematic Diagram of Analog Computer Inverter Circuit.

The application of Kirchhoff's current law to the summing junction, SJ

yields Eq. (II-l).
ia = it + i0 (II-la)

Kirchhoff's voltage law states:

The sum of the vo ltage rises in a given direction around a loop is equal to
the sum of the voltage drops around that loop in the same directio n.

The application of this law to the left hand loop in the clockwise direction,
A-SJ-G-A, yields

eA = iaRa T" ei (II-1 b)

The voltage eA is a voltage rise in the clockwise direction because the

reference d irection for eA shows terminal A at a positive potential with
respect to terminal G. Similarly iaRa is a drop in potential because ter­
minal SJ is at a lower potential than terminal A. Application of the voltage
law to the right hand loop, SJ-O-G-SJ, yields
e, = i0Rf T- (II-lc)

The nature of the operational amplifier provides two more relations:

Co (Il-ld)
€i ~ G

C;
li
(II-le)
Ri
182 APPENDIX I I

Solve Eq. (1Mb) for ia:

eA
ia = (II-1Q
R

Solve Eq. (II-lc) for io:

U = (H-lg)
Kf

Substitute Eqs. (Il-le), (II-If), (II-lg) into Eq. (II-1a).

eA ~ ei = + g' ~ g° (II-lh)
Ra Ri ^ Rf v

Collect the terms in e,- on the right and substitute Eq. (II-Id).

s - -tf i ! i -«;] - Z
en = -m (Il-lj)

Since the value of G is very large, the second term in the denominator
becomes very small when the values of Ra, Rf, Ri a nd G a re substituted.
For example, typical values of these parameters are as follows:
Rf = R a = 106 ohms
Ri = 50 X 106 ohms
G - 100,000

The denominator then becomes:

Denom = 1 + 10-5(1 + 0.02 + 1)

= 1.0000102 (II-lk)

Thus the inverter configuration of Fig. II-1 has the following gain
relation:

e„ = (n-'O

When Rf = R a the output voltage is the negative of t he input voltage.

 D E R I V A T I O N O F A NA L O G C O M P U T E R G A I N S 183

II-B Summation

The detailed connections of an operational amplifier for operation as a

summer are shown in Fig. II-2. Voltage and currents are identified and
reference d irections are shown. By application of Kirchhoff's laws in a
manner similar to that of the previous section the following equations
result:
Kirchhoff's current law at the summing junction, SJ:
ia " t~ ib ~l" ic — ii T~ io (II-2a)

ground -=-

FIG. 11-2
Schematic Diagram of Analog Computer Summer Circuit.

Kirchhoff's voltage law:

Loop A-SJ-G-A; eA = iaRa + (II-2b)
Loop B-SJ-G-B; &b — hRb 4~ (II-2c)
Loop C-SJ-G-C; ec — i cRc + d, (II-2d)
Loop SJ-O-G-SJ; e, = i0Rf + e0 (II-2e)

The properties of t he operational amplifier yield:

<?, = — eo (II-2f)
G
li ei
= % (H-2g)
 184 APPENDIX I I

Solution of Eqs. (II-2b), (II-2c), (II-2d), and (II-2e) for ia, ib, ic, and i0
yields

eA — ei
la Ra

eB e'
• = ~
lb ~ Rb (II-2h)
ec — ei
h Rc

lo ~ Rf

Substitution of Eqs. (II-2g) and (II-2h) into Eq. (II-2a) eliminates the
current variables.

•i , — c, ec — ei c,- — e0 /TT ^
J . - + X + X = «" + T ( I 1 "2 , )

The terms in e(- are col lected on the right:

eA . es . ec f 1 , 1 . 1 , 1 , 1 "1 e0 ,rr
+ +R„ = e{_R,
R„ + ^ + «;
Rc +ft+S}J-«7 (II"2J>

The voltage c, is eliminated by substituting Eq. (II-2f) into Eq. (II-2j).

Terms in e0 are also consolidated.

e^d + £* _l £c = i lii i 1 , ilm 2k)

R, R b ^ Rc Rf <?[_*« Rb^Rc Ri + R f \ u

Solve for c0:

Co = (11-21)
l + ^E + l + E' + f+ l )
The right term in the denominator is again very, very small because the
amplifier gain G is very large. Hence the output voltage is the weighted
sum of t he input voltages. Each voltage is multiplied by the ratio of the
feedback resistor to the corresponding input resistor.

e =
° + (uh + (sf)ec] (II"2m)
 DERIVATION OF AN ALOG C OMPUTER GAI NS 185

II-C Integration

The detailed connection of an operational amplifier as an integrator is

shown in Fig. II-3. The application of Kirchhoff's current law to the junc­
tion SJ yields:

ia — li to (II-3a)

<? +

ground

FIG. 11-3
Schematic Diagram of Analog Computer Integration Circuit.

The application of Kirchhoff's voltage law to the left loop yields:

eA = iaRa + <?; (II-3b)

The application of Kirchhoff's voltage law to the righthand loop yields

<?; = vc + e0 (II-3c)

The voltage across a capacitor is related by the capacitance to the charge

stored in the capacitor.

(II-3d)
= h
But the value of q i s the J idt and, therefore

Vc = 5 /<» (II-3e)
c ' iodt
186 A P P E N D I X II

The nature of th e operational amplifier again yields

e0
ei = ~G (II-3f)

e>
li =
(II-3g)
R,
Solve for ia in Eq. (II-3b):
eA —
la — (II-3h)
Ra

Substitute Eqs. (II-3g) and (II-3h) into Eq. (II-3a) and solve for i0:

eA — e t ei
10
(H-3i)
Ra Ri

Substitution of E q. (II-3i) into Eq. (II-3e) yields

1 f [~eeAA-~ eet e/~| j

= cj (II-3j)
Ra _ ~RR
a"

Substitution of E q. (II-3j) into Eq. (II-3c) yields

1 eA — e ( £/
dt T e0 (II-3k)
C Ra Ri.

Substitution of E q. (II-3f) into Eq. (II-3k) yields

e_o e
_ J_ [ _A_ e0 , e0
G C J Ra dt + e + ^ 0
R a G + RiG
dt (11-31)

Solving for eo on t he left yields

eQ =
RaC j 6Adt G e° + cf (r o + R^ dt ] ( n " 3m )
For a sufficiently large value of amplifier gain, G, the second term on the
right is negligible compared to the first term and the integrator has the
relationship:

e° = -jf?;
toI-
I eAdt (II-3n)
 DERIVATION OF ANA LOG COMPUTER GAI NS 187

The capacitor may be initially charged giving an initial condition.

(II-3o)

The reciprocal of the RC product is the gain of the integrator where R is

measured in ohms and C is measured in farads. It sometimes is more con­
venient to measure R in megohms and C in microfarads. One megohm is
106 ohms and one microfarad is 10~6 farad. The powers of 10 cancel out
and the result has the proper numerical value. The RC product has units
of sec, which is consistent because the integral of v oltage with respect to
time gives u nits of volt-sec which yield, when divided by sec, the units
of v olts.
APPENDIX in ELECTRICAL CIRCUITS

This appendix shows in detail the operation of the

electrical circuits used for POT SET and IC.

Ill-A Pot Set

The potentiometer can be set by a balancing technique that com­

pensates for the reduction in voltage from a potentiometer caused by
current flow in the ARM of the potentiometer. This method also makes it
unnecessary to have calibrated dials on the potentiometers. The desired
ratio is set on the Master Potentiometer, MP. The voltage at the arm of
the potentiometer, CP, being set is compared by a millivoltmeter, MV,
with the voltage from the Master POT as shown in Fig. III-l. The resistor
Ra is the input resistor to which the arm of the potentiometer is connected.
A relay grounds the summing junction SJ of the operational amplifiers

188
 ELECTRICAL CIRCUITS 189

+ ref O—

CP -AAA/—1 SJ MP
Rn

coefficient master
potentiometer
potentiometer
ground

FIG. 11 1-1
Schematic of Pot Set Principle.

in the POT SET mode. This circuit functions very much like the Wheat-
stone Bridge. When the potentiometer, CP, is adjusted so that the reading
on the millivoltmeter is zero, the potentiometer CP is set for the same
ratio as the master potentiometer. Since the resistor Ra is effectively con­
nected the sam e as when the computer is operating, the loading is com­
pensated properly. If the lower end of the potentiometer CP is not
grounded, the bridge circuit is not complete and hence will not function
properly.
The actual circuit is more complicated. A typical POT SET circuit is
shown in Fig. III-2. The Pot Set Relay, KPS, switches the terminal B
from the amplifier input to ground when the POT SET MODE is selected.
When SW1 is d epressed the contact arms shift from top to bottom. The

sw\
T HI
patch panel

O
° +ref Aa a ^ ^
O Wv °—
°1_
CP1
KPS <

pot set bus

+ REF

-^Mvy MP

FIG. 111-2
Typical D etailed Circuit for Pot Set.
190 APPENDIX IN

top end of the potentiometer is connected to the + Reference Voltage and

the ARM is connected to the POT SET BUS which connects to the left
side of the millivoltmeter, MV. If the computer is well en gineered the
needle on the millivoltmeter will move in the same direction, clockwise or
counterclockwise, as the knob is moved on the potentiometer, CP, which
is being adjusted. Thus if the meter shows a downscale deflection, the
knob on potentiometer CP is turned clockwise to bring the meter to a
zero deflection, and hence to set the potentiometer to the desired ratio.
The converse is the case for an upscale deflection. When the master po­
tentiometer, MP, is being set to match the setting of a coefficient poten­
tiometer, CF, the needle on the meter moves in a direction opposite to the
rotation of th e knob of the master potentiometer.
On computers which have a digital voltmeter, DVM, in addition to, or
in place of a panel meter, the POT SET is accomplished by connecting the
DVM in place of th e millivoltmeter MV and the master potentiometer.
This system is more convenient to use because the desired ratio is set only
on the coefficient potentiometer, rather than first on MP and then on CP.
These computers frequently have another position on switch SW1 which
permits the measurement of the voltage present on the arm of the poten­
tiometer when the upper end is connected to the HI terminal on the patch
panel. This feature is convenient for setting a potentiometer whose lower
end is not grounded, or for checking to see that the desired voltage is in­
deed present on the arm.

III-B Initial Condition

Consider the circuit shown in Fig. III-3 below: The resistors R have the
same value and ignoring for the moment the capacitor C, the circuit looks
like a unity gain inverter. When the capacitor is considered, the output

R
ico WV •oo
c
ft

FIG. 111-3
Principle of Initial Condition Circuit.
 ELECTRICAL CIRCUITS 191

FIG. 111-4
Typical Circuit for Establishing Initial Conditions.

voltage will ris e after a period of time to the same value as though the
capacitor were not there. By choosing a suitable value, R, for the resistor
this time can be made as short as desired. Relays or electronic circuits are
used to select the mode of operation of an integrator circuit. There are
usually three modes: 1. RESET or IC, 2. HOLD, and 3. O PERATE or
COMPUTE. One possible circuit is shown in Fig. III-4. When the com­
puter is placed in the Initial Condition, IC, mode, the relay KIC in each
integrator unit switches the arm from the compute position, C, to the
reset position, RS.
When the computer is placed in the HOLD mode, the Pot Set Relay,
KPS, opens the circuit from the terminal B to the input of the operational
amplifier. In this condition the input to the integrator is zero and the out­
put voltage HOLDS at the value which it had at the time the HOLD
condition was initiated.
APPENDIX IV
STABILITY O F SOLUTIONS
FOR A LGEBRAIC EQUATIONS

Consider the following set of simultaneous algebraic

equations.

4x + y = 6 (IV-la)

x + 5 v = 11 (IV-lb)

This system of equations would result in a stable analog computer model

if Eq. (IV-la) is solved for x and Eq. (IV-lb) is solved for y. The solution
would be found to be as follows:

x = 1
(IV-lc)
y= 2

192
 S T A B I L I T Y O F S OL U T I O N S F O R A L G E B R A I C E Q U A T I O N S

0.25

Stable Analog Computer Model for Simultaneous Algebraic Equations.

This fact can be verified by substitution into the original set of equations.
If th e analog model is amplitude scaled on a o ne unit per volt basis, the
block diagram is as shown in Fig. IV-1. If the circuit is completed, between
x and xi, the system will come to equilibrium with x = 1 and y — 2. Con
sider now that the connection between x and x\ is opened and that a
variable voltage xi is supplied to potentiometer 03. The table below indi­
cates the value of x for various values of xi. These values can be observed
experimentally on the analog computer, or can be computed in t e
following way:
11 — Xi
y = —3—

A ~ 4

*= 4
 194 APPENDIX IV

XI *

0.8 0.99
0.9 0.995
1.0 1.000
1.1 1.005
1.2 1.10

Note that if x\ is too small then x is larger than x\, but still less th an the
equilibrium value. If x\ is too large, then x is smaller than xi, but still
larger than the equilibrium value. Thus when x\ is connected to jc, the
system tends to move toward a position of e quilibrium if jc has a value
different from the equilibrium value. The same tendency would be ob­
served if th e circuit were opened between the output of amplifier 03 and
potentiometer 01. The input to potentiometer 01 would be considered
as yi.
Consider now what happens when the equations are reversed before
forming the analog model.

x+ 5y=ll (IV-le)

4x + y = 6 (IV-lf)

0.5

FIG. IV-2
Unstable Analog Computer Model for Simultaneous Algebraic Equations.
 STABILITY OF S OLUTIONS FOR AL GEBRAIC EQUATIONS

The analog model is as shown in Fig. IV-2. The assumption is again made
that the dotted connection between x and xi is opened and a variable
voltage x\ supplied. The results for several values of xi are tabulated
below based o n the relation:

jc = _ 19 + 20 xi

XI X

0.9 -1
0.95 0
1.00 +1
1.05 +2
1.10 +3

In this case note that when xi is too small, the value of x is even more
negative with re spect to the equilibrium value than is xi. When xi is too
large, the value of x is more positive with respect to the equilibrium value
than is x\. Thus when the connection is made that forces xi to be equal to
x, any departure from the solution point, x = 1, y = 2, causes the values
to depart even further from the solution point.
This demonstration of the nature of the instability of the analog model
was used because it probably is more meaningful and provides more in­
sight than the argument based on the roots of the characteristic equation
formed from the matrix of t he coefficients.
 INDEX

Acceleration: Electrical charge, 35

angular, 125 Electric current, 35
Algebraic equations, 18 Electric power, 92
simultaneous, 4, 21, 192
Amplifiers: Falling body:
operational, 7, 11 freely, 84
Amplitude scaling, 9, 19 with viscous drag, 88
Analog computer, 3
Feedback resistor, 12, 14
active, 5
First-order system, 70
passive, 4
Forward bias (See Bias, forward)
Analogs:
Forward resistance, 138
of arithmetic operations, 7 Freight car, 136
of calculus operations, 33 Function:
Angular:
cosine, 53
acceleration, 125 polynomial, 50
velocity, 130 ramp, 47
sine, 53
Back resistance (See Resistance, back) sinusoidal, 53
Balance, 5
Function generator:
Bias:
diode, 144
forward, 138
sinusoidal, 54, 148
reverse, 138
Blackburn double pendulum, 135
Gain:
Block diagram, 11, 72
integrator, 42
voltage, 8
Capacitance, 35
electrical, 35 Galilei, Galileo, 85
thermal, 92 Generator:
Capacitor, 35 sine-cosine, 148
sinusoidal, 54
Cathode ray oscilloscope, 64
Gravity, 84, 89, 129
Chopper stabilization, 5
Ground, 8
Cm, 10
Comparators, 152
Cosine function, 53 HOLD, 163, 191
Hybrid computer, 1, 6
Dashpot, 110 Hydraulic system:
Differential equation, 75 two tank, 73
Differentiation, 34 three tank, 98
Digital Computer, 2
Digital Voltmeter, 190 IC (See Initial condition)
Diode function generator, 144 Inertia:
semiconductor device, 138 moment of, 124
DVM (See Digital voltmeter) Initial condition, 45, 46, 190
Dynamic system, 71 Input resistors, 14

197
198 INDEX

Integration, 40, 162, 169, 176, 185 Resistor:

Integrator: feedback, 12, 14
gain, 42 input, 14
Inversion, 11, 160, 168, 175, 180 Reverse bias {See Bias, reverse)

Kirchhoff's Laws: Scale factor, 9

current law, 180, 181, 183, 185 amplitude, 10
voltage law, 181, 183, 185 time, 82
Scaling, 72
Limiter, 139 amplitude, 9, 19
Linear equations: time, 82
simultaneous, 21, 40, 192 Second-order system, 109
Semiconductor device:
Modeling: diode, 138
mathematical, 3, 72, 73 thermal behavior of, 92
Models: Servo multipliers, 145
differentiation, 37 Simultaneous algebraic equations, 4, 192
integration, 41 Simultaneous differential equations, 98
mathematical, 72, 73 Simultaneous linear equations, 21
Moment of inertia, 124 Sine-cosine generator, 148
Momentum, 85 Sine function, 53
Multiplication by a constant, 16, 174 Sinusoidal function, 53
Multipliers, 145, 170 generator, 54
quarter-square, 145 SJ {See Summing junction)
time-division, 145 Spring-mass-dashpot system, 110
servo, 145 Strip chart recorder, 62
Summation, 13, 160, 168, 175, 183
Nonlinear modules, 176 Summing junction, 13
Nonlinear phenomena, 137 Swinging door system, 120
Superposition, 43
Ohm's Law, 8, 12 System, 10
Operational amplifiers, 7,11 dynamic, 71
Oscilloscope {See Cathode ray oscil­ first-order, 70
loscope) hydraulic, 73, 98
second-order, 109
Pendulum: spring-mass-dashpot, 110
Blackburn double, 135 swinging door, 120
simple, 128
Polynomial function, 50
Terminal velocity, 89
POT BAL, 17
Thermal capacitance, 92
Potential difference, 8
Thermal power, 93
Potentiometer, 16
Thermal resistance, 92
three-terminal, 17
Time constant, 81
two-terminal, 17
Time-division multiplier, 145
POT SET, 17, 188
Time scaling, 82
Precision, 2
Transfer characteristic, 139
Trial solution, 72
Quarter-square multiplier, 145

Ramp function, 47 Velocity:

REP-OP mode, 61, 163 angular, 130
RESET, 163, 167, 191 Viscous drag, 88
Resistance: Voltage, 8
back, 138 Voltage gain, 8
forward, 138
thermal, 92 X-Y plotter, 62