[Mr. Emmanuel Roche, France]

COMDYNA.WS4
-----------

(ROCHE> I had the surprise to discover that one Ray Spiess had built
2,000
copies of one analog computer, the Comdyna GP-6, during... 36 years! Since
I
am a book worm, I made a search, and found 4 articles where he explained
it.
This file contains them.)


- "Midwestern Simulation Council meeting"
Ray Spiess
"Simulation" magazine, Vol.10, No.3, March 1968, pp.105-107

(not retyped)


- "Midwestern Simulation Council meeting"
Ray Spiess
"Simulation" magazine, Vol.11, July 1968, pp.3-5

(not retyped)


- "A Viewpoint on Analog Simulation"
Ray Spiess
"Simulation" magazine, Vol.29, No.5, November 1977, pp.181-182

(Retyped by Emmanuel ROCHE.)

Comdyna is in the business of building analog computers. Since 1968, we
have
been producing a variety of analog computing systems -- general-purpose
analog
computers for student training and small simulations, special-purpose
analog
computers for signal processing. A recently developed microprocessor
interface
system has added microhybrid computing to our analog capabilities.

If we had listened to the experts, back in 1968, we would have quit before
we
started. The prevailing wisdom said that analog computers were
obsolete,
anachronisms relegated to the past -- totally superseded by the
digital
computer. We are certainly glad, now, that we believed otherwise. In the
past
nine years, we have participated in a technology revolution that
has
dramatically changed analog computer concepts.

In our brief existence, we have lived through the entire evolution of
linear
integrated circuits. A remarkable array of integrated circuits has given a
new
perspective to present and future designs. There has never been a better
time
than now to apply analog computing techniques.

Although the principles of analog computing remain essentially unchanged,
the
scope of application concepts is completely different.
Amplifiers,
integrators, multipliers, etc. still perform the same tasks, but there is
a
new attitude concerning their use.

When they were vacuum-tube or discrete transistor designs, these
fundamental
components cost hundreds of dollars, and were renowned for their
fickle
operation. As integrated circuits, the same components cost pennies, and
give
predictable and reliable performance. In years past, the range of
analog
computer applications was severely restricted by the costly nature
of
computing components. Present-day designs freely apply analog methods with
an
almost complete disregard for the number of components used.

It is ironic that, at a time when analog component technology has reached
such
a high degree of perfection, the analog computer suffers a decline
in
popularity. The preoccupation with digital computer simulation languages
has
convinced so many individuals of the analog computer's demise that there is
a
general reluctance to admit its virtues.

Analog simulation still excels where operator experimentation and judgment
are
the keys to solving problems. For an inductive analysis or for
experimenting
on a model operating on the fringes of validity, operator involvement is
as
important as ever. The direct contact with, and close interaction between,
man
and simulation produces subtle understandings of a problem, and
generates
intuitive solutions. Digital computer simulation still inhibits
operator
contact enough to impede the analysis of certain problems.

Analog computers are unmatched for teaching the development of
mathematical
models and the principles of simulation; thus, they continue to be
an
important part of educational programs. Hands-on experimentation and
direct
contact with a simulation are especially valuable for student training.
Analog
computer programming also teaches the basic electronics of signal
processing,
instrumentation, and control.

Comdyna has concentrated on developing products for the educational market.
In
1968, we introduced our Model GP-6, a low-cost general-purpose analog
computer
that was designed for undergraduate instruction. The GP-6 is a
traditional
analog computer that is as old as an ancient Boeing differential analyzer,
and
as new as the most modern present-day model. It handles up to
fourth-order
simulations, offers a complete assortment of non-linear functions, and has
the
conveniences of a large system. Since its training function is basic to
all
engineering and scientific education, we have maintained the GP-6
model's
integrity, and intend to do so for the foreseeable future. We believe that
the
GP-6 has been, and will continue to be, valuable in perpetuating
analog
computer techniques, and preparing engineers to apply the new analog
component
technology.

Design concepts that apply the new technology are found with GP-6
accessories
and other simulation systems. A number of devices, special simulators,
and
functional components aid student experiments, and increase
problem-solving
capabilities. A microhybrid expansion system interfaces the GP-6 and
a
microcomputer. The microhybrid provides AD/DA conversion, plus
parallel
operations such as comparators, DA switches, and patchable logic. Its
primary
objective is to teach hybrid computing, and to train students to
apply
microprocessors to analog systems.

In other developments, we have departed from the traditional analog
computer
organization. Specialized simulation systems so structure their
internal
components that programming is oriented to the intended user. For example,
our
"Reaction Kinetics Simulator" is designed for the non-engineer scientist
who
shuns analog simulation because traditional analog computer
programming
requires some understanding of electronics. For many life
science
applications, analog simulation is an ideal problem-solving medium, but
the
presence of amplifiers, resistors, and capacitors poses a
psychological
barrier to its use. To overcome this barrier, the "Reaction
Kinetics
Simulator" is organized so that programming follows a set of equations.
Patch
panel symbols and programming procedures avoid any reference to electronics.

Our "Math Model Simulator" follows the traditional analog
computer
organization, except that operational elements are presented as
summers,
integrators, and attenuators, rather than amplifiers, resistors,
and
capacitors. By structuring the "Math Model Simulator" in this manner, a
young
mathematics student can bypass the electronics part of analog
computer
programming, and immediately apply calculus and simulation concepts.

Tailoring simulators to user programs give a new direction to analog
computer
design. Custom simulators can simplify programming, provide special
features
for the operator, and reduce costs. It is now practical to combine many
analog
operations into groups of Input/Output functions. When simulations are
divided
into categories (mechanical, chemical, physiological, etc.), there are
usually
recurring loops that can be treated as independent subprograms.
These
subprograms may, then, be programmed as physical or mathematical
functions,
rather than as analog computing components. Many traditional programming
and
operating difficulties are thus eliminated.

Analog simulation has an excellent future if prospective users will
take
advantage of the new technology. To do so will require users to recognize
new
concepts, and to discard opinions based on the old analog computer
structure.
Powerful analog or analog/hybrid simulators can now be built at a fraction
of
the old analog computer's cost. Those with an open mind, those who are
not
totally committed to the digital computer, will find that an analog
approach
offers efficient, inexpensive, and accurate solutions to particular
problems.
With innovation from users and producers, the full potential of
analog
simulation can be realized.


- "Scottsdale, Again"
Ray Spiess
"Simulation" magazine, Vol.41, No.?, July 1983, pp.36-39

(Retyped by Emmanuel ROCHE.)

Editor's note: The Western Simulation Council Conference, Scottsdale,
Again,
was held April 21-22, 1983 at the Scottsdale Quality Inn, Scottsdale,
Arizona.
The program consisted largely of vendor presentations. The address by
Ray
Spiess of Comdyna, Inc., was based on the paper reproduced here.


- "The analog/hybrid computer... an ideal microcomputer instrumentation and
controls laboratory"
Ray Spiess
Comdyna, Inc.

A microcomputer in real world control and signal processing applications
must
apply its discrete operations in a non-discrete environment. The
continuous
analog variables of real systems and the discrete digital data
of
microcomputers are opposites. It takes special instrumentation skills
to
combine variables and data into real time, real world
microcomputer
applications.

The design, development and testing of real systems presents a problem:
the
continuous systems to be controlled or instrumented, with their widely
ranging
variables, are not easily duplicated on a designer's bench. Hybrid
computers
solve the problem by incorporating analog simulation to synthesize
continuous
variables, and by applying analog-logic functions to increase the
parallelism
of real time programs.

Practically any microcomputer can easily and inexpensively be made into
a
hybrid computer. A Comdyna analog/hybrid unit is added as a peripheral with
no
more difficulty than installing a printer, memory expansion, etc. With
a
Comdyna analog/hybrid peripheral add-on, a system becomes a
microcomputer
instrumentation and controls laboratory. The microhybrid laboratory is
ideal
for conducting data acquisition and control methods research,
testing
interface designs, developing real time programming software, and for
teaching
real system applications concepts.


The need for an instrumentation and controls laboratory
-------------------------------------------------------

Real time, real world microcomputer applications concern systems in which
the
microcomputer's role is to supervise and/or perform data acquisition,
decision
making, and control. The microprocessor is but one component within
the
design. Its application is just one part of the system's operation. To
produce
a working design, it is not enough to know only the microprocessor.
Also
needed is an understanding of the environment to which the processor
is
applied, and how the processor interacts with the environment.

A systems approach to any problem requires that the whole be divided
into
components parts, so that each part may be evaluated independently, and
then
interdependently with other parts. For either an inductive analysis
to
understand a natural component's behavior or a deductive analysis to
evaluate
a designed component's performace, there is a need to analyze and test
under
controlled conditions. To study and perfect applications, some type
of
laboratory is required. An ideal choice for the laboratory is an
analog/hybrid
computing system.

Figure 1. Microprocessor-based system

Figure 2. Schematic of a hybrid computer simulation of
a
microprocessor-based system.

Figure 1 is a schematic of a typical microprocessor-based system. Figure 2
is
a schematic of a hybrid computer simulation of Figure 1. The system of
Figure
2 is composed of parts that are simulated by analog, digital, or
combinations
of both methods. As a microprocessor system simulator, the
analog/hybrid
laboratory offers a general-purpose programming structure that enables
the
system's components to be isolated and evaluated under predictable
controlled
conditions.

The hybrid computer offers analog linear circuit functions and a
computer
organization that focuses on the analog/digital interface. Working from
analog
computer programs that synthesize the continuous variables of a process
or
product: Transducers and sensors may be simulated and designed with
analog
computer components; Controllers may be simulated or designed with
analog-
logic components; Real time microcomputer programs may be developed and
tested
under near real operating conditions.

Laboratory work may be conducted on systems that are total or
part
simulations. The components of Figures 1 and 2 may be interchanged where
parts
are simulated and parts are real. Hardware may be tested with
simulated
products or processes. A real product or process may be controlled with
a
hybrid computer program. By interchanging simulated and real components,
the
microcomputer instrumentation and controls laboratory aids a
project's
development from the investigative through the design and testing phases.


Hybrid Computer simulation
--------------------------

Hybrid computer simulation is distinctive in that both analog and
digital
functions may be applied to model a system's dynamic behavior. Under a
single
programming organization, a wide assortment of methods and devices is
made
available. A programmer picks and chooses the best way to handle the job.

The choice of methods is normally determined by whether the model's
natural
behavior is best described by continuous functions or discrete
calculations.
Generally, analog computer simulations excel where the behavior is
an
action/reaction response that is defined by a set of ordinary
differential
equations. Digital computer simulations best model behavior that is based
on
logic determinations, statistical interpretations of data, and
static
calculations. Hybrid computer models simulate behavior that is
basically
continuous but include the discontinuities of logic determinations,
memory
requirements, complex parameter calculations, and empirical functions.

Assuming that the user can work effectively with both analog and
digital
methods, hybrid computing offers an efficient handling of dynamic
models.
Fast, real time simulations are possible as central processor burdens
are
reduced to a minimum, and model solutions are not restricted to
step-by-step
instructions. Hybrid computer simulations are networks of electronic
methods
that individually relate to natural phenomena. As such, they achieve
the
closest possible approximation to real behavior.


What is a hybrid computer?
--------------------------

A hybrid computer offers, within one programming structure, an
analog
computer, digital computer, and linking interface that includes
analog-logic
functions. The primary objective of any hybrid computer is to combine
into
single operating programs digital data and analog variables. Attention
is
equally given to software development and hardware operation.

Figure 3. Microhybrid I system organization

Figure 3 is a schematic of the Comdyna Microhybrid I, which is offered as
an
example of a hybrid computer. It consists of a general-purpose
analog
computer, a general-purpose digital computer, and an interface
unit.
Programming is performed by: 1. Patch cord connections at the analog
computer.
2. Normal electronic data processing by the digital computer. 3. A mix of
the
two within the interface unit.

An examination of the schematic shows that the Microhybrid I interface
section
bears a strong resemblance to typical microcomputer-based product or
process
control designs. A multiplexer and analog/digital converter transform
analog
signals into digital data. Multiplying digital/analog converters
transform
digital data into analog signals. Comparators produce indicator logic
bits
from continuous analog variables.

Digital/analog switches turn variables on and off with control logic.
Gates,
counters, monostables, buffers, etc. are applied as one would find in
many
common logic circuits. The hybrid computer applies, as general procedures,
the
same techniques that are used in the design of dedicated electronic
circuits.

Complete system programs consider three computing media: 1.
Continuous
variables; 2. Data; 3. Binary indicators and control logic.
Continuous
variables are produced by networks of analog computer components. Data
is
collected, processed and distributed as digital computer programs.
Single
logic bits that indicate binary states and provides binary control
are
produced and handled by an assortment of devices that may be purely digital
or
analog, but are likely to be composites of both.

While continuous variables and discrete data are well understood
opposites,
single bit interface is a gray area middleground. Indicator and control
bits
may originate from analog variables, digital data, or as strictly
interface
functions. The bits may be processed with networks of gates, flip-flops
and
other logic components, or as part of the digital computer program.

Many opportunities to optimize overall program run speeds and meet frame
time
requirements exist with the handling of indicator and control bits.
An
integration of single bit programming with analog computing networks
and
digital data processing is a major hybrid computing strength. An ability
to
utilize binary logic as a link between the analog and digital worlds is
an
advantage the hybrid computer has in meeting application objectives.


The benefits of analog computer simulation
------------------------------------------

Natural systems are fundamentally continuous. It may or may not be
realistic
to simulate continuous systems with discrete data. It may or may not
be
workable to sample a natural system as discrete data, and control it
with
discrete commands. Digital computer simulations and control designs
inevitably
demand near zero program frame times and a near infinite resolution of
system
variables. At some point, a digital method will fail because of a
program's
excessive frame time, or a variable's inadequate resolution.

As the electrical analogs of real systems, analog computer simulations
produce
the same continuous, infinite resolution variables as those found in
natural
environments. Analog simulations, therefore, synthesize continuous
variables
with an accuracy and realism that is unattainable by digital or other
methods.

Analog computer simulations produce a predictable and controlled response
to
stimuli. Dynamic characteristics are determined by user programs which
are
based on analytical models or transfer functions. A dependable response
to
inputs, in many cases, makes analog simulations more desired testing
grounds
than real environments. When an instrumentation or control design is
tested
with a simulation, it can be developed to operate with theoretical
validity
before being exposed to real world perturbations.

Analog simulations also enable a control design to be tested under a range
of
conditions that may not be possible, practical or safe under real
world
conditions. Model parameters are easily changed. Non-linearities, noise,
etc.
can be added progressively as the testing program matures. Designs may
be
tested to operate outside the safety limitations of real devices.

Finally, testing with analytical models rather than their real
counterparts
induces a generalized approach to control design. Designs are developed
to
control general transfer functions, rather than specific products
or
processes. Any natural system that has similar characteristics may utilize
the
design. As methods, techniques, and designs become more applicable to
general
rather than specific models, they become more adaptable to
future
requirements.


Asynchronous logic
------------------

Physical systems, even when they are fundamentally continuous, have
discrete
operations in the form of events and variable states. (For example: A pump
is
either on or off; a liquid is either above or below a given level.)
Such
operations are likely to be based more on the manipulation or cause and
effect
of variables than on analytical programs. How they are handled is important
to
the simulation programmer or control system designer.

If a status is dependent on space or time relationships rather than a
clocked
sequence, it makes sense to perform operations with asynchronous logic
in
preference to a central program. (For example: It would be inefficient
to
digitize a liquid level measurement and run a computer routine to turn a
pump
off and on.) By processing signals asynchronously, without sending them to
the
microcomputer, a hybrid program produces functions that are
basically
continuous, but include the discontinuities of discrete, binary operations.

Two hybrid computer components, the comparator and the digital/analog
switch,
interface continuous variables with digital devices. The comparator produces
a
binary logic bit that indicates if the polarity of the sum of two or
more
analog inputs is positive or negative. The digital/analog switch turns
analog
variables on or off from a binary command.

Binary outputs from comparators may: 1. Directly control
digital/analog
switches. 2. Be gated with other logic, trigger flip-flops, monostables,
or
counters. 3. Be monitored by digital computer sense lines or interrupt
digital
computer programs.

Digital/analog switches may be controlled from: 1. The outputs of
logic
devices. 2. The digital computer I/O bus.

Sensing and controlling the states of analog variables with asynchronous
logic
maintain the parallel nature of continuous systems. The effects of the
logic
occur simultaneously with their causes. For asynchronous logic functions to
be
handled by a digital computer, the central processor must continually
monitor
an array of state indicators, run binary logic routines, and output an
array
of control bits. Delays in producing switch control logic can restrict
a
digital computer's ability to keep up with real time demands.

When the digital computer program is busy running data acquisition,
decision
making, or control routines, the added requirements of asynchronous
logic
functions may impose unnecessary programming difficulties. The
asynchronous
logic section of a hybrid computer encourages the use of
comparators,
switches, and logic devices as a means to reduce the digital
computer's
programming burden.


The microcomputer hybrid computing system
-----------------------------------------

In the past, the digital computer section of a hybrid computing system
was
selected to support analog computer simulation. Today, analog
computer
simulation is viewed as a support for the digital computer. With the focus
on
digital data acquisition and control, analog and interface circuits are
now
treated as microcomputer peripherals. Modern analog/hybrid
products
communicate with microcomputers as standard I/O devices.

As peripherals, analog/hybrid products may be added to a microcomputer of
the
user's choosing. How a system is configured is determined primarily by
how
much responsibility the user chooses to assume, regarding the
interface
functions. Where in-house hardware and software are utilized, a hybrid
system
may be constructed by simply adding an analog computing unit.

When a complete analog/hybrid unit is the peripheral, the user is relieved
of
the interface design responsibility. Like any typical parallel
peripheral
device, the analog/hybrid unit is plugged into a parallel I/O bus,
and
programmed as a memory expansion.

The Comdyna interface employs three programmable ports, as shown in Figure
3.
A port's mode may be set to data, instructions, and sense bits transmitted
by
loading from, and storing to allocated memory locations. An example of
an
interface instruction set is given below:

Name Description
---- -----------
ADC Starts the analog/digital converter.
FUN Enables the analog/digital converter to be set by analog
time
scale clock pulses.
ADR Enables the processor to load converter data.
MUX Sets the analog multiplexer address.
MDA Latches data into a multiplying digital/analog converter.
IOR Enables the processor to load logic state indicators.
IOW Latches single bit logic commands.
RDY Resets the interrupt latch.

Machine language routines that incorporate the hardware commands provide
a
macro set that allow interface functions to be programmed as
high-level
instructions. Analog signals and logic/indicator/control bits are handled
as
program variables.


Circuits vs. software
---------------------

The spectacular availability of microprocessors as low-cost product
components
has induced a tendency to overstress the software role in meeting
design
objectives. That circuitry can be a better and less expensive alternative
is
often lost on today's young engineer. In numerous existing designs,
time
consuming and difficult software could have been avoided with the use of a
few
amplifiers, logice devices, and simple circuitry.

With its structure of general-purpose circuit devices, by its very
presence,
the hybrid computer encourages the system or product designer to apply
both
circuitry and software. The hybrid computer programmer acquires the skills
to
routinely use a wide selection of hardware devices and software techniques.


Conclusions
-----------

In the final analysis, hybrid computing is more a concept than equipment.
It
is a concept that equally stresses software and hardware solutions to
data
acquisition, instrumentation, and control problems. It instills
an
understanding of basic system dynamics as a prerequisite to solving
systems
problems. It encourages the application of analytical and empirical
methods.
However it is configured, the analog/hybrid computer is an ideal
microcomputer
instrumentation and controls laboratory for research, teaching, and
product
development.


- "The Comdyna GP-6 Analog Computer: Twenty Five Years... and still
counting"
Ray Spiess
"Simulation" magazine, Vol.59, No.5, November 1992, pp.323-326

(Retyped by Emmanuel ROCHE.)

What computer holds the record for the longest production run for all
of
simulation? Come on. You know. The answer is in the title. That's right,
the
Comdyna GP-6 Analog Computer.

This June, the GP-6 entered its twenty-fifth year. And, there is no end
in
sight. In spite of a depressed U.S. economy, 1991 was one of the best
sales
years ever.

When I started the business, I could only dream that, one day, my
little
company would become the world's largest and most profitable manufacturer
of
general-purpose analog computers. From meager beginnings, we fought and
clawed
our way to the top, eliminated competitors one by one: Comcor;
Systron-Donner;
Applied Dynamics; Hitashi; Dornier; Electronics Associates, Inc. (EAI);
Heath-
Kit.

Now, Comdyna dominates the market. I have achieved a monopoly.

By 1967-68, it was becoming an uphill battle for the analog
computer.
Presentiments filled the air.

Simulation is going digital, the smart voices said.

No, the analogers answered, the real world is analog, and always will be.

With digital simulation, there are no equipment constraints. Problems are
not
limited in size. Yes, yes, but everyone knows that the disciplines of
problem
simplification and organization are critical to understanding,
and
understanding is what simulation is all about.

You get rid of the patch panel. You will never have to patch another
program.
Hey, patching is not so bad. Once the analog computer is patched, it is run
as
the real system. Operator interaction is how we solve simulation problems.
How
can anyone interact with a digital computer?

Amplitude and time scaling: Who needs it? You will never have to scale
another
variable again. Now, you are really showing your ignorance. Don't you
know
that scaling brings out internal time constant variations. Scale a
program,
and you have probably solved the problem.

You will never have to battle the quirky nature of analog computer hardware.
I
love my analog computer, hardware quirks and all. More often than not,
they
are symptomatic of model definition problems, inappropriate application
of
physical laws. When the digital computer spits out garbage, how does
anyone
know?

You will eliminate the analog computer laboratory, run simulations along
with
business applications. Engineers can run their own problems. Eliminate
the
analog computer lab manager? The simulation specialist? Wait a minute.
Who,
then, will have the experience and insights to solve system problems,
problems
of dynamics, non-linearities?... And, another thing, the digital computer
is
too damn slow! Do you know what a digital computer simulation is? AN
ERROR
ANALYSIS WITH A TWENTY-FOUR HOUR TIME CONSTANT?

The digital computer people attacked zealously, as if the matter were war
and
nothing less than annihilation of the analog computer would be
accepted.
Analog simulation's misfortune, it seems, was to share with digital the
name
"computer". With some other name, analog simulation's fate may have been
to
coexist on the basis of shared strengths and weaknesses. How much
enthusiasm
would digital computer advocates have had, how much fun would it have been
to
annihilate the analog simulator, or the analog differential analyzer?

Opportunities of the times were certainly not limited to digital
computing.
The new integrated circuit miniaturization was revolutionizing analog, as
well
as digital, technologies. Fairchild, then Amelco, then others, offered
an
operational amplifier in a TO-5 package, with prices in dollars, not
the
hundreds of dollars charged by mainframe suppliers. Manufacturing
efficiencies
made possible further cost reductions. High-quality analog computers could
be
built and sold for a fraction of the old prices. It was the reason I went
into
business.

The smart entrepreneur first conducts a market study before risking
his
enterprise. Mine occurred following an Eastern Simulation Council
meeting,
sitting in a Miami airport bar with Professor Donald Martin of North
Carolina
State University. I asked Don what he thought of an analog computer that
could
be produced and sold for under $1,000. He said he would buy one, no
question
about it. In fact, he would buy a whole bunch for his laboratory. I
could
count on it.

The GP-6 was born then and there, and on the airplane home. (Don never
bought
any, by the way.)

Pursuant to what I had learned of the analog computer tradition --
insights
acquired, to some extent, while working for EAI but mostly from my
association
with Simulators, Inc. and Chuck McVey, and the manual on the subject,
Maughan
Mason's classic treatise, "The Kludge", the essence being that, while the
norm
is to first build then sell a product, reversing the order: selling it
then
seeing if it could be built, was the analog computer way -- I constructed
from
carboard, contact paper, photo-etch aluminium, and various purchased parts,
a
sort of GP-6, had a photograph taken, prepared and placed an ad
in
"Engineering Education" magazine. I figured, if I landed some orders, I
would
be in business. If not, well...

Orders for thirteen came in.

It was John Edwards who made it possible. I knew him from Simulators,
Inc.
When McVey brought John in, nothing of our product showed a first sign
of
working. John made that mess of electronics into a solid analog computer.

He learned his electronics in the World War II Navy, eventually worked for
EAI
then CPI as a field service technician. He was brilliant and
difficult,
fiercely independent, wanted to work for no one but himself. Today, he
would
probably be called a hacker.

And, who is this McVey, and what is Simulators? Some of you, youngsters,
may
ask.

He was one of the many unforgettable characters who found their way into
the
so-called golden area of analog computers. How remarkable. All of a
sudden,
there appeared this magnificent electronic machine that, with
life-like
realism, could simulate the equations that, for centuries, held
virtually
inaccessible the explanations of physical behavior. And, along came the
space
program and equations needing to be solved for national honor. Money
flowed,
poured, roared into analog computer simulation, drew like bears to honey,
the
most wonderful, motley group: brilliant theoreticians; talented
engineers;
respected academics; the ubiquitous investor; slick salesman,
off-the-wall
inventors, charlatans, and downright scoundrels.

A character bigger than life, literally, was Charles J. McVey. Charlie
stood
six foot three or four, and tipped the scales at no less than 350 pounds.
He
achieved a legendary status as the salesman for CPI, EAI's embittered
spin-
off, where he won from EAI one contract after another. He made
outlandish
promises. Lab managers smiled and believed him. He lavishly wined and
dined
prospective customers. Oh, he loved to eat: Buckets of steamed
clams.
Fricassee of Hereford.

After CPI folded, he rounded up some investor cash, and started
Simulators,
Inc. I was a young EAI field sales engineer. He offered me Vice President
of
Marketing, and I jumped at the opportunity.

I soon learned that I was also designer of the entire product line.

To create the Simulators story, McVey had worked an agreement to
sell
Telefunken A.G., Germany, analog computers. The plan was to package
Telefunken
amplifiers, multipliers, etc. into an instant Simulators, Inc. line. It was
a
myth ended quickly. Not only were their components of a mediocre
performace,
they were expensive.

"We gotta build our own", said Mac, when it was pointed out that the
amplifier
price he thought to be a dual was, actually, a single.

We charged full bore ahead. From McVey selling ample quantities of
stock,
Simulators grew rapidly. Every functionary imaginable came on board: sales
and
marketing, production, quality control, inventory control,
purchasing
managers, etc., etc. Every one that was, except an electronic project
engineer
who could build an analog computer.

None worth his salt would work for McVey. John Edwards wouldn't either.
But
John needed the money. He was trying, as he would for the rest of his life,
to
make a go of his own company, Anaconic, Inc., at the time consisting of
him,
his wife, Irma, and Melana, Irma's visiting cousin from Costa Rica. The
three
worked out of their kitchen and basement. On his own, John had designed
a
quarter square multiplier which he sold for existing 231-R's and other
100-
volt systems.

Chuck made a deal with John. He came in every evening, worked till dawn. In
a
few months, he designed an excellent amplifier, an accurate multiplier,
and
just about everything else that went into the 240 Simulator.

When I was about to part from McVey and told John about my venture,
he
proposed that Anaconic design and build the electronics. That was June,
1968.
In October, Comdyna shipped the first GP-6. It went to Southern
Methodist
University, the Civil Engineering Department.

Working with John was not easy. I would take to him the mechanical GP-6
ready
for wiring and installation of p.c. boards. A few days later, I would call
and
ask if it was ready. He would always say yes, and after dinner (John
rarely
worked during the day), I would drive the hour to his house. Not once upon
my
arrival was a unit ready. I would sit with him at his basement work
bench,
hour after hour, going through the check out, Irma or Melana
finishing
assemblies, John delivering instructions in Spanish: "Punga usted, el
beecho,
a'aqui, a'aqui." If I was lucky, I would have a GP-6 by midnight, two or
three
in the morning. If not, the sun would be high in the sky before I packed it
in
my car and headed home.

Well, I could not take a lifetime of that production schedule. Gradually,
I
phased John out.

I think John viewed those basement sessions as social occasions, ones in
which
he could pass on to me bits of his wisdom, express his views on
the
achievement of individual initiative. I especially remember one of his
adages,
the one, I would say, that personified him most: "If a design works the
first
time, you have taken too long to do it."

John died a few years ago, penniless. His fertile, inventive mind could
never
quite carry through to a profitable conclusion. What happened to Chuck
McVey?
It is my understanding that he still resides in a Vancouver jail,
fighting
extradition to stand trial in California for high-tech espionage, or
some
charge to that effect. When Simulators went under, McVey became an exporter
of
sorts. One evening in 1982, John Edwards called to say he had just seen
McVey
on the evening news, arrested by the Customs Department for attempting
to
illegally export a Memorex computer. Later, a reporter wrote that, by
change,
sitting next to McVey on a flight to Moscow, McVey bragged about how he was
on
a first name basis with Leonid Brezhnev, and that he was transporting
high-
tech goods into Soviet Russia via Zurich, Switzerland. After the arrest,
he
flew to Malta, where he lived as a fugitive. In 1987, he was again
arrested,
this time in the Yukon by a Canadian Mountie. (He loved Alaska; many
summers,
he flew his airplane there.) According to "Time" magazine, Charlie headed
a
ring of 'technobandits' that had stolen designs for a new computer
developed
by the Saxby Computer Corp., one, the article stated: "that could
track
satellites and missiles."

As far as the GP-6 is concerned, it simply kept going year after
year,
essentially unchanged operationally and physically from those first units.
One
of its virtues is that new GP-6 units can be purchased to operate
side-by-side
with ones five, ten, twenty years old. Internally, the old and new bear
little
resemblance. At first crammed with circuit boards, they now seem empty.

The GP-6 was originally, and still is, primarily an educational
product,
inexpensive enough for schools to afford the multiple stations needed to
run
laboratory programs. The EAI sales force established the market. Bids
would
come out for one, maybe two, EAI TR-20's. I would offer five or six
GP-6's,
and end up with the orders.

If all worked nicely. Then EAI disappeared. The GP-6 was on its own.

The presentiments grew louder. The analog computer is dead, finished,
buried,
the last nail is in the coffin, those old voices said. Indeed, it seemed
they
might be right. It was the mid to late seventies, and engineering
enrollment
was down. Space age achievements were old hat, considered redundant
and
expensive. Technology was seen as the cause, not the cure, of world
problems.
Control labs were being dropped from curriculums.

If we learn anything in growing older, it is the impossibility of
predicting
the future. About this time, hirers of engineers, so-called industry,
sounded
an alarm. They did not like the new graduates, their deficiencies
of
laboratory experience. Pressure was brought to bear for accreditation
changes,
an upgrading of laboratory programs.

The controls laboratory became, once again, an object of
attention.
Ironically, it was the microcomputer that stimulated the new GP-6
purchases.
Even if the focus is on digital controller design, control theory remains
a
subject of continuous variables. Running simulations on the analog computer
is
the way students first learn the fundamentals, experiment with
continuous
systems.

Faculty who had disposed of their last 'obsolete' analog computer,
found
themselves buying new GP-6's. One of the first rejuvenations occurred
at
M.I.T. in 1979. To this day, Dr. Paul Haupt, then head of the
Mechanical
Engineering controls program, credits the GP-6 for resurrecting
his
laboratory.

Of course, many programs remained strong, then as they are today. Ohio
State
University, Purdue University, Virginia Polytechnic Institute, and many
others
offer good examples. All three academies (Naval, Military, and Air Force)
made
major purchases of the GP-6 during the seventies. They are still using
them.
Visit the Astrolab at the Air Force Academy, and you will find
GP-6's
controlling every conceivable piece of apparatus.

Although the GP-6 took a renewed and positive turn, it was no means a
rebirth
of the analog computer. Is is the growth of control laboratories (industry
as
well as university) that has created new prospects. In the controls lab,
the
analog computer never died.

If one could go back to the Sixties to the height of the Mercury Program,
for
instance, and visit McDonnell Aircraft, upstairs where the engineering
design
took place, there would be seen rack after rack of analog computers
solving
the simulation problems of the day. Downstairs in the laboratory would be
the
hardware: apparatus; actuators; controllers; AD/DA converters; digital
and
analog computers working together in testing control system designs. Go
back
today. Upstairs, the racks of analog computers are long gone. But,
downstairs
in the laboratory, analog computers are still hanging on to
apparatus,
controlling actuators, and simulating parts of systems as they were,
back
then.

The analog computer patch panel, then and now, offers the only means
of
programming linear circuits. Patch a controller, filter, etc. on the
analog
panel, and it is up and running in minutes. The alternative: Lay out a
circuit
of amplifiers, resistors, capacitors, etc. Buy the components, insert them
in
a breadboard, and interconnect the wiring. Test and troubleshoot.

What, in reality, is the GP-6 analog computer? It is, above all else,
a
programmable, easy-to-use collection of high-quality linear
circuit
components: Stable drift-free amplifiers. Precision summing resistors.
Low
leakage integrating capacitors. Potentiometers, voltage dividers that can
be
easily set with a digital voltmeter. Accurate multipliers/dividers.
Diode
function generators. Precision 10.000 volt reference.

The roots of linear circuitry are analog computer programs. It was the
analog
computer where the operational amplifier was perfected. It was programming
the
analog computer where engineers concocted the innovations that led to
today's
circuit designs. It was the hybrid computer where logic and analog
functions
were first combined, where early digital computers operated in parallel
with
analog systems.

What is the future of the analog computer, the GP-6? Probably a continuance
of
the present. As long as there is a need for linear circuits, a need for
high-
speed in-the-loop simulations, there will be reasons for buying the GP-6
and
other Comdyna analog computers. Consider information as bounded by
discrete
and continuous, where analog is to continuous as digital is to discrete,
and
where digital speed, ever increasing, shrinks a distance between the two:
No
matter how much of the spectrum is viewed as digital, approaching
the
continuous fringe, the world becomes analog.

Speed is digital computing's limitation. It seems inevitable that some form
of
analog processing will break the bottlenecks of present and future
information
handling: Two current areas that offer prospects are fuzzy logic and
neural
networks. In both, one finds an awareness of analog steepest ascents,
and
linear programming. How fascinating to witness a digging up and dusting off
of
these 'ancient' Sixties concepts, and envisioning them as part of some
new
data processing.

And, the record setting GP-6? Well, Ken McLeod, my associate and good
friend
of many years, and I will wait and see. Maybe we will go ahead with
that
production line that will turn out a thousand or more GP-6's every hour, to
be
ready for the twenty-first century, when every scientist and
businessman,
every mother, and dad and child, will be patching their own neural
network,
fuzzy logic, or state-of-the-art breakthrough speed-of-light program.
Just
maybe.


- "The Comdyna GP-6 Analog Computer: Alive, but not exactly kicking"
Ray Spiess
"IEEE Control Systems" magazine, June 2005, pp.68-73

(Retyped by Emmanuel ROCHE.)

General-purpose consoles filled whole rooms, and ran simulations that laid
the
foundation of our space program. These machines have disappeared, except
as
science-fiction movie-set props. Keep a watchful eye when you are
viewing
"Star Trek" or similar show reruns, and you may recognize a Comcor, Reeves,
or
Beckman with added banks of flashing lights for communicating with aliens.

Fate has been kinder to the small analog computer. At least to our GP-6.
The
Comdyna GP-6 continues its timeless, engineering/educational role,
primarily
in the controls laboratory.


How Comdyna started
-------------------

It was spring of 1968. Chuck McVey had just fired me. Three years earlier,
I
helped him found Simulators, Inc. The venture was going nowhere, spending
gobs
of money, and we were not getting along. Out of work, I had two choices:
I
could either start my own company like everyone else in those days, or I
could
seek employment. In the spirit of the times, I fabricated a sheet-metal
model,
named it the GP-6, snapped a couple of pictures, placed an add in
"Engineering
Education", and waited to see what would happen. If the ad produced
some
orders, I would go into business. If not, I would start interviewing.

Thirteen orders came in.

The GP-6 shipped that October was not exactly the GP-6 of today, but
close,
both conceptually and operationally. All the changes in all the years
since,
except for a subsequent digital microcomputer interface, have been
internal.
Originally crammed full of parts, the chassis evolved to become
virtually
empty.

What is remarkable, at least to my thinking, is that the GP-6, for its
36
years of production, essentially looks, operates, and performs the same as
the
GP-6 advertised by that sheet-metal mock-up. It would seem that 36 years
of
production, of teaching the same concepts, should set some kind of records.

The GP-6 keeps going without change because the analog computer's
application
fundamentals, physical system simulation and linear circuitry, have not
and
will not change. Students of today learn these fundamentals from
patching
simulations on the GP-6 exactly as the students of 36 (and more) years ago.

The analog computer retains its advocates because certain educators
recognize
that there is a limit to the use of shortcuts and clever devices in
teaching
the basics of a subject.

There was a time when the analog computer's future was supposed to lie
in
electronic patching. "Eliminate the patch panel", was the cry. So we did.
Back
in the 1980s, Comdyna designed Micropatch to be programmed by one of the
new
microcomputers. It worked pretty well. Did not cost all that much. No
one
wanted it.

It turned out that the patch panel, much maligned over the years, has been
the
analog computer's strength, its source of durability.

So, where is the progress? Progress has come from, and can continue to
come
from, students learning math modeling and simulation,
inductive/deductive
analysis, and the application of linear circuits... from the ground up.


In the controls laboratory
--------------------------

"What's that?", a student asks, showing up for the first control lab,
and
seeing the GP-6 and oscilloscope sitting on the bench. "Didn't those things
go
out with the dinosaurs?" The assignment: the forever-first design of a
servo-
motor control of an inertial load. As it has been, from even before my day,
a
control design is proposed; its math model derived; parameters
estimated,
measured, or taken from manufacturer's data; and the mathematics *solved*
to
determine the best controller settings. Working with the analog computer,
the
math model is transformed into a program schematic. Parameters (that is
to
say: the system constants) are gathered from known data or
estimates;
dependent variables are amplitude scaled; the independent variable, time,
is
scaled; the program is patched; parameters are entered as
potentiometer
settings; and the program is run as the model's electronic analog, tested
as
if it were the real hardware. Where available, the real hardware, the
servo-
motor and load, replaces the patch-panel counterpart, and is controlled by
the
patch-panel controller. When the experiment extends into digital
control,
analog/digital converters and a digital computer program replace the
patch-
panel controller, and the analog computer's job is done.

By conducting the design in such phases, the student compares step-by-step
the
theoretical model with its electrical analog, and the electrical analog
with
the real system. He evaluates the simulation and hardware with identical
test
methods/instruments and, in so doing, compares the design, operation,
and
performance of digital and analog control.

Here, please allow me to brag about what students gain from the
analog
computer. First, classical control theory assumes continuous variables, so
it
makes sense for the first design to be a continuous controller. The
analog
computer enables the system to be controlled, and the controller design to
be
electrically simulated.

How else would you build a real system simulator? And, what better way
to
quickly understand system subtleties than to experiment with a real
system
simulation of the system's mathematical model? Right away, it becomes
apparent
that physical laws and their math expressions are one and the same.
(Obvious
in definition but not necessarily in perception, especially for the
young
student.) It has been, over the years, one of those experiences where
light
bulbs turn on. *I get it!* The common denominator: mathematics, that is
to
say: physical laws expressed as mathematical functions.

Indulge me also to put in a plug for one of the analog computer's
often
forgotten virtues. The rigors of programming (system simplification,
amplitude
scaling, and time scaling), so maligned during the analog computer's
golden
era, were frequently instrumental in solving problems, even before
the
simulation was run. Shortcuts offered with digital languages have not
always
been to the problem solver's advantage. Bypassing simplification,
amplitude
scaling, and time scaling encouraged confused, unwieldy simulations where
the
problem-solving suffered essentially by not seeing the forest for the
trees.
Those who have struggled with analog computer programs know inherently
the
importance of dynamic range, the benefits of working with properly
scaled
variables, and the weighing of internal loop frequencies to judge
when
variables can be disregarded because of response time differences.
Engineers
who have not struggled have had to learn such concepts another way.

Not to be overlooked, while students wrestle with analog computer
programming
and patching, they are learning the fundamentals of linear circuits. There
is
no avoiding it. Programming the analog computer requires an understanding
of
operational amplifiers; the not-so-obvious qualities of high gain,
negative
feedback, high input, and low output impedance in applying Kirchhoff's
Laws;
the converting of operational amplifiers into summers and
integrators,
multipliers/dividers; and many other elements of program schematics.
What
better way to learn about linear circuits than to be forced to create one?

Way back, in ancient history, when I was an Electronics Associates, Inc.
(EAI)
sales engineer, prospective customers would take a doubtful look at the
TR-48
or one of the other EAI models, and ask how long it would take to learn how
to
program one of those things. The answer: "Hours when a problem is to
be
solved. Days, or even weeks, without one."

Linear circuits owe their development to the analog computer. A
most
difficult, unstable device, the operational amplifier was perfected in the
due
course of analog computer research, development, and design. Designers
also
had to come up with schemes for analog voltage multiplication,
division,
polar/Cartesian resolution, logarithm and arbitrary function generation,
and
other inventive simulation tools. And, the schemes had to work. Our
space
programs, Mercury and Apollo, depended on it. The analog computer, at
the
time, was the only means for solving the complex, non-linear
simulations
critical to the space program's success.

Motivation? Consider the operational amplifier that, today, can be bought
for
pennies, costing a US[imath]1,000, US[/imath]1,200, or US$1,500 apiece, a system of
them
going for a US$1 million and more. It was the major players, EAI,
Applied
Dynamics, Boeing, Reeves, and others chasing big bucks that laid
the
foundation, first in vacuum tibes, then transistors, that evolved into
the
linear, integrated circuits of today.

The analog computer, dead? No way. Every circuit is an analog computer.


Why are they still around?
--------------------------

There is no denying the attrition of analog computers in the engineering
labs.
I can offer no argument to counter the contention that analog computers
have
limited use, are clumsy apparatus that take up valuable bench space, and
are
often taught by laboratory assistants who have never seen one before. It is
a
wonder that any are around, that certain faculty stick with the old
analog
computer, no matter how persuasive the reasons not to. Why? I asked.

The following are brief, first-hand answers from faculty who were kind
enough
to respond. The first is from Umit Ozguner, Electrical Engineering, Ohio
State
University. Prof. Ozguner's controls laboratory has been, and continues to
be,
distinguished by highly imaginative and creative projects. Umit
Ozguner
writes:

Hi Ray,
It was nice to hear from you. I was going to respond immediately
but
wanted to see if my students could use the GP-6 for a new
application
that I had. That worked out fine, and now I can indeed report we
have
a recent application.
For a number of years, I have been involved in developing
autonomous
vehicles (mostly self-driving cars), and we now have in our hands
a
large truck to automate. We intend to use a fairly sophisticated
real-
time software operating system with digital feed-back control, and
I
wanted to try out the software in the lab before deploying it on
a
huge, heavy vehicle.
So we have now simulated the longitudinal dynamics of the truck
(not
very complex) on the GP-6 with the A/D and D/A to be used on the
truck
and the real-time computer control to be implemented. The students
who
were scared about analog computing first seemed happy enough and
want
to try lateral control (more complex) now.
Best wishes,
Umit

Thanks, Umit.

The University of Wisconsin has been using analog computers at their
Madison,
Milwaukee, and Platteville campuses. Prof. Dan Cobb, who has taught at
both
Milwaukee and Madison, writes:

I have been using the GP-6 analog computer in my instructional lab
on
automatic control at the University of Wisconsin since I created
the
course in 1996. Although I do not use all of the analog
simulation
features, I have found the amplifier patch bay to be indispensable
for
quickly assembling circuits. With half of my course devoted to
analog
control, students are frequently required to build analog
compensators
to drive electromechanical hardware. As long as analog
methods
continue to be used in control and signal processing, systems like
the
GP-6 will be advantageous for instruction and prototyping.
Students find the GP-6 far more convenient to work with
than
breadboarding techniques. The patch bay offers banana
plug
connections, potentiometers with a digital readout, a
built-in
voltmeter, power-supply voltages, an overload indicator, and
short-
circuit protection. Its size also adds to its
functionality.
Furthermore, the units are virtually impervious to abuse. With the
GP-
6, students only have to worry about broken wires. Two of our
units
were purchased more than 20 years ago and are still fully
operational,
so I expect that the GP-6 will outlive most of the other equipment
in
the lab.

You can visit Prof. Cobb's lab at his Web site [Ref.2].

Prof. Peter Meckl, Mechanical Engineering, Purdue University:

Ray,
Yes, we still use them, primarily to illustrate continuous
domain
controller design. We want students to make a connection
between
controllers (PID and lead/lag) and op-amp circuits, which
analog
computers illustrate very nicely. We have the students develop
simple
controllers for a servo table (motor and amplifier with
position
feedback) using the GP-6. At the end of the course, we use
Quanser
equipment, which has the WinCon interface with MATLAB Simulink so
we
can do a purely digital controller. But we haven't yet given up on
the
analog computer, since it's the best way to implement a
continuous
controller.

Dr. Robert Paz, Klipsch School of ECE, New Mexico State University:

Ray,
The analog computer continues to be a valuable teaching tool
for
students studying linear systems. It provides a simple way to
realize
analog transfer functions. Even in our modern
"almost-all-digital"
world, there are settings in which true analog devices are
necessary.
One of the applications I like to teach students about in my
course
"Computer Control Systems" is the tracking of various signals.
It
turns out that for tracking almost anything but a constant input,
it
is impossible to get a true "ripple-free" response using a
digital
controller alone. for example, if a motor is to track a sinusoid,
the
input to the motor must be a pure sinusoid in order for the output
to
be a pure sinusoid. A digital controller alone will generally
give
intersample "ripple" because the output of a zero-order hold
(D/A
converter) cannot faithfully represent a true sinusoid. However,
using
a continuous-time (analog) model of the reference signal
in
conjunction with the digital controller can achieve a smooth,
ripple-
free result. The analog computer is thus the perfect companion to
the
digital computer for such a setting.

Prof. Ian Petersen, School of Electrical Engineering, University of New
South
Wales at the Australian Defense Force Academy:

Dear Ray,
As you know, I have for quite a few years used the GP-6
analog
computer in the control systems laboratory in the School of
Electrical
Engineering at the Australian Defense Force Academy. The primary
users
of the laboratory are undergraduate students who are training to
be
officers in the Australian Defense Forces (Army, Navy, and
Airforce).
When I first arrived at the school in 1985, the control
systems
laboratory had a small amount of equipment including two old
EAI
analog computers and some "feedback" servo systems. The
available
equipment was inadequate for the increasing numbers of students in
the
control course. The GP-6 combined with the Comdyna servo system
was
chosen as a suitable platform for building a practical
control
laboratory. Over the years, the laboratory built up to
five
workstations. Initially, data capture, printing, and digital
control
were done using the board and software available with the GP-6,
but
later we switched to dSPACE hardware and software.
The GP-6 provided a valuable tool for modeling physical systems
and
implementing analog controllers. In my opinion, the GP-6
provided
students with a valuable educational experience in connecting
the
theory in lectures with physical hardware.

Prof. Petersen has placed a complete, first control design manual on his
Web
site, and it can be downloaded in Acrobat format [Ref.3].

Prof. Dan Block, University of Illinois, writes:

Ray
We have four laboratory assignments that use the Comdyna
analog
computers. All these labs are part of control systems theory
courses
we teach here at the University of Illinois.
1) One lab introduces the students to analog computers demonstrating
the
summer, differentiator, and integrator. These students know
very
little about op amps so this is their first introduction to them.
2) Then we have three sets of labs that use the analog computer
to
control the position of a dc motor with dc tachometer feedback
for
speed and a pot for angular position. The students implement a
PD
controller with tachometer feedback for D, a PD controller
with
differentiation of the pot as D, and finally a lab that implements
a
lead controller design.
3) We also use the analog computer to perform a PI speed controller
for
the same dc motor with tach feedback.
I have no plans of removing the analog computers from my 18
benches
here in the Control Systems Lab. The only lab that I see being
removed
in the near future is the lead controller design. The wiring for
that
controller can be very confusing for a first time analog
computer
user. We did use them more in the past and have moved many of
the
controller implementations to Simulink auto code
generation
implementations. But we plan on keeping at least one lab for a
class
using the analog computers. I feel it is very important to
show
students that these controllers can also be implemented with
analog
circuitry.
We also use the analog computers for simple summing circuits
to
combine signals for digital filtering labs.
Dan

Prof. Julio C. Mandojana, Electrical Engineering, Mankato State University:

I have used analog computers (Comdyna) as part of a
junior-level
laboratory for at least ten years. Our department runs two programs
in
parallel, one for electrical engineering and one for
electronic
engineering technology. Students in both programs are required to
take
a class on control systems. Engineering students take the
control
class in their junior year, while the technology students take
an
equivalent class in their senior year.
In both classes, the experiments are:
1) simulation of a positioning system; we have the actual plant.
The
results of the analog simulation are then compared to a
digital
simulation using Scilab or MATLAB.
2) repeat the previous experiment adding velocity feedback.
3) simulation of the open-loop plant to be controlled digitally from
a
PC.
We purchased the computers sometime in 1990-1991, and their
operation
has been flawless since then, except for the occasional cleaning of
a
potentiometer. The analog computer has been a great tool for us
for
two reasons: it provides a quick way of implementing a
control
algorithm and it gives the student a "picture" of the solution to
a
differential equation. It is difficult to overemphasize the
insight
into a control problem given by an analog simulation. The students
do
get a better understanding of dynamic problems when they see
how
solutions change under the variation of coefficients.

Prof. YangQuan Chen, Electrical and Computer Engineering, Utah
State
University:

Dear Ray
The GP-6 is an asset for control and mechatronics education. We
have
three GP-6s actively being used in ECE5320 Mechatronics. The GP-6
has
been used as a plant emulator for rapid real-time control
system
prototyping labs. The role of "analog computing" of the GP-6
should
now be shifted to applications such as acting as a flexible
bread
board, a plant emulator, an analog signal conditioning unit,
and
analog controller implementation among many other possibilities.
From my observation, GP-6 can stay alive forever in control
education.

Prof. Rachid Manseur, Electrical and Computer Engineering, University of
West
Florida:

Ray,
An important practical step in the design and development of a
control
system, after digital computer simulations, is laboratory
test-bench
verification and testing. Analog computers are used in our
linear
controls lab to teach undergraduate students about analog
simulation
of systems described by differential equations. Our first
two
laboratory experiments are about the analog simulation of first-
and
second-order differential equations.


And why they are not
--------------------

Locating schools that discontinued the use of their analog computers was
more
difficult in that the individuals responsible for the analog computers
were
invariably no longer with the institutions. One exception, the U.S.
Naval
Academy, was once a leading proponent of analog computers. Prof.
Robert
DeMoyer writes:

Ray,
We discontinued the use of general-purpose analog computers about
ten
years ago. The digital simulation package Vissim initially became
the
simulator of choice, and then Simulink. The reasons for the
change
were what you might expect. The change was made because
digital
simulators are faster to program and easier to use, more
accurate,
more capable, and require virtually no maintenance. Furthermore,
we
felt that we should provide the students with experience using
a
simulator that they are likely to encounter in engineering
practice.
What has been lost is a tool to teach troubleshooting. This loss is
by
far overshadowed by the gains previously mentioned.

A testament to the analog computer's invincibility: Rather than junking
unused
units, laboratories put them up for sale. Check eBay. Every so often, you
will
find (how do they say it?) a previously owned Comdyna GP-6 for sale.

A few years ago, I received a telephone call. The man introduced himself
as
Dan Slater. He had a personal laboratory with a few GP-6s purchased
from
surplus listings. An engineer by occupation, one of his side interests
was
music synthesis. He had just published a paper in "Computer Music", an
MIT
publication, on two analog circuits for enhancing the performance of a
Moog
synthesizer. As brought out in the paper, the circuits were patched and run
on
one of his GP-6s. The reason he called, other than to introduce himself
and
make known his application, was that musicians were contacting him and
he
wished to refer them to Comdyna.

I thought, what a nice application. Maybe Comdyna should be in the
music
business. I offered on our Web site (it is still there [Ref.4]) one of our
GP-
10s stripped down to handle his circuits at my lowest possible price. I
did
not sell any. Even at this reduced price, the special music analog
computer
could not compete with the old GP-6s offered on eBay.

Useful as they are, virtually no analog computers today are purchased for
(nor
are many old analogs retained in) the laboratory as programmable
circuit
signal processors. Too bad. Like Dan Slater's circuits, a low-cost
analog
computer like the Comdyna panel-mounted GP-10 offers an easy and
reliable
means to quickly have a prototype circuit up and running.


If just for a different name
----------------------------

The roots of linear circuitry are analog computer programs. It was the
analog
computer where the operational amplifier was perfected. It was in
programming
the analog computer that engineers concocted the innovations that led
to
today's linear circuit designs. It was the hybrid computer where logic
and
analog functions were first combined, where early digital computers
operated
in parallel with analog systems. And so on. It has long been my opinion
that
the analog computer's misfortune has been to share with digital the
name
"computer". Given some other name, there may have been no analog
versus
digital battle waged successfully by digital computer advocates, and the
two
may have coexisted on the basis of shared strengths and weaknesses. Maybe
if,
from the beginning, the analog computer were called, let's say, an
analog
simulator...

...ah, mellowing into old age, now one of the old analogers (*old*
the
acknowledged redundant modifier), I am waxing sentimental.


References
----------

1 "An interdisciplinary, interdepartmental control systems laboratory"
A.G. Alleyne, D.J. Block, S.P. Meyn, W.R. Perkins, and M.W. Wong
IEEE Control Syst. mag, Vol.25, No.1, pp.50-55, Feb. 2005

2 "ECE Feedback Control Laboratory"
D. Cobb
http://eceserv0.ece.wisc.edu/~cobb/lab/lab.html

3 [Online]
I.R. Petersen
http://routh.ee.adfa.edu.au/~irp/P1.html

4 "GP-10s"
Ray Spiess
http://www.comdyna.com/music.htm

5 "The GP-6 Analog Computer: Twenty-Five Years... and still counting"
Ray Spiess
"Simulation" mag. Vol.59, No.5, pp.323-326, Nov. 1992


EOF