Design of an Analog Computer
ENGINEERING Case LIBRARY ECL 47
SYSTRON-DONNER CORPORATION (A)
Design of an Analog Computer |
“In March 1965, Dick Dunlop, a Product Designer in the Electronic
Instrumentation section of Systron-Donner Corporation, Concord, California,
began preliminary design work on a proposed new analog computer. The com-
puter was to be a small portable model using transistors exclusively in
place of vacuum tubes,
Systron-Donner Corporation was formed in. 1960 from a merger of Systron
Corporation, makers of electronic test instruments with Donner Scientific
Company, an established manufacturer of analog computers and inertial
instrumentation. In 1965 net sales were $10 million and the firm employed
a little over 400 people. Products include various types of electronic
test instruments, counters, and gyros and servo-accelerometers for military
aircraft, in addition to computers. .
(c) 1966 by the Board of Trustees of Leland Stanford Junior University.
Prepared in the Design Division, Department. of. Mechanical Engineering,
Stanford University, by John A. Alic, under the direction of Professor
H.O.Fuchs, with financial support from the National Science Foundation.
ECL 47—
Dick Dunlop had been with Systron-Donner for a year anda half when
he began working on the new computer. Dick explained that his job as a
product designer at S-D included both mechanical design and styling of
what is basically electronic equipment. Dick arrived at this position
following varied experience after leaving art school in 1957. He worked
for two years as a technical illustrator, then took a job as a design
draftsman at Lick Observatory, where he designed small high-precision
optical instruments. Next he designed large high-precision optical
instruments, including an X-Y measuring machine for tracking guided
missiles with an accuracy of one micron* in six inches, for an Oakland
(California) firm. Dick pointed out that the design of such items --. of
which only one or two are to be produced -- is dictated primarily by
the manufacturing processes which must be employed, so that the design
seems almost to complete itself once the basic idea is stated.
' During this period Dick was acquiring valuable technical knowledge
through extension courses given by the University of California at
Berkeley and at community trade schools offering night courses in such
subjects as machine tool and foundry practices.
Previous to joining Systron-Donner Dick worked for IBM. He pointed
out that the difference between products at IBM and designing for Systron-
Donner is again a matter of manufacturing processes, as dictated by pro-
duction quantities. Dick said that while the products at IBM had been
slated for mass production, most of S-D's products are produced in
quantities such that castings, forgings, and stampings, are uneconomic.
The first project Dick worked on at S-D was the 40/80 series of
analog computers described in Exhibits 1 and 2. These two computers
were the company's first transistorized analogs and the industry's first
+ 100 volt solid state analogs. They went into production in the summer
of 1964. Dick explained that with equal absolute component accuracies
a + 100 volt computer affords accuracy superior by an order of magnitude
to the + 10 volt computers offered by S-D's competition. He also
pointed out that a vacuum tube computer built ten years previous to the
40/80 with equal capabilities would have occupied a good sized room.
The 40/80 price and quotation form which-appears in Exhibit 1 shows
that a series 40 computer with a full complement of equipment sells for
around $30,000, while a similarly equipped series 80 runs $50,000 to
$60,000. These prices put the machines beyond the reach of the smaller
companies and universities engaged in research and development activities
for which an analog computer is a valuable tool. In addition to its
R & D uses, the 40/80 also finds application in the chemical industry
for process control.
* -6
One micron equals 10° meters.
ECL 47
-S-D!s sales and engineering representatives, who market the company's
products across the country, had been reporting wide demand for a smaller,
lower cost solid state analog ever since the introduction of the 40/80.
Many prospective customers can neither afford the 40/80 nor utilize ef-
fectively its capabilities. Systron-Donner's management felt that there
was a definite demand for a small, easily portable analog selling for
around $10,000. Such a computer would be within the reach of most
universities for teaching purposes as well as for research, and would
tap another large market consisting of the smaller engineering laboratories.
It was also felt that such a computer would be purchased by many medical
centers since analogs were being used to increasing extents both in
medical research and in diagnosis.
Systron-Donner's management concluded that a computer using perhaps
7 to 10 of the computing modules from the 40/80 could be sold in quantities
of 20 to 30 per month. A computing module is shown in Exhibit 3. There
were six different modules available for the 40/80, providing various
capabilities, with more under development. The series 40 uses a total
of 21 modules; the series 80 uses 42.
The proposed new model was designated the 10/20. Discussion of the
features of the new computer in relation to the market at which it would
be aimed began in the fall of 1964. It was planned that other components
from the 40/80 series besides the modules be employed in the 10/20, for
example the coefficient potentiometers used to set the values of variables.
By January, 1965, an electrical engineer had been named project
engineer for the 10/20 and he had written and submitted to management a
"New Product Proposal", part of which is shown in Exhibit 4. Because
Systron-Donner is not a large company, management normally keeps in touch
with the activities of the engineering departments without the aid of
many written communications. The proposal was not intended to inform
management, since the project engineer was in consultation with manage-
ment during its preparation. Its purpose was rather to provide a basis
for budgeting the development of the 10/20. Thus the product specifica-
tions in the proposal were quite flexible and changes could be made at
any time.
Dick Dunlop's involvement with the 10/20 project began in March when
he started work on the preliminary mechanical design of the package.
This was known as Phase 1 of the development -- generation of alternative
schemes for packaging the electronics specified by the electrical engineers.
Phase 1 of the electronics design was simple and straightforward, the
essentials being carried over from the 40/80 experience. During Phase 1
Dick's task was to investigate the concept of the computer with respect
to functional capabilities, styling, estimated production costs and
general feasibility. He was not in a decision making position however;
he was merely expected to propose alternatives in these areas. Specifi-
cally, he was expected to propose three or four different package designs
offering various arrangements of the components.
A-3
ECL 47
One of Dick's proposed designs was accepted by:management in June,
and work began on a hand-built model on July: lst. Ordinarily this would
have been an engineering prototype, during the construction of which all
design details, both mechanical and electrical, would have been finalized
prior to making the engineering drawings of the piece parts and assemblies.
Phase 2 for the 10/20 did not follow this usual pattern, however, because
it was decided to prepare a working model of’ the computer in time for the
1965 WESCON Show (Western Electronic Show and Convention) in August. Be-
cause of the short time available, this first model was built with bread-
boarded electronics and, while in appearance the computer was finished,
many of the details had not been finally determined. This model was
built in the engineering model shop from sketches Dick prepared. While
making these sketches he also worked on a layout of the entire computer,
After the first model had been completed and exhibited at the WESCON
Show August 24-27, work began on the Phase 2 engineering prototype which
would exactly represent the final design. During the months from July
onward Dick worked in parallel with the electrical project engineer, with
each man responsible for his own area. Since the computing modules were
in existence, the electrical engineers were concerned mainly with designing
new power supplies.
Coincidentally with work on the prototype, documentation of the design
began. In October, three draftsmen were assigned to help Dick with the
task of making drawings from which the computer could be produced by the
manufacturing department. Dick began working on a new layout of the com-
puter which would incorporate all revisions of the design -- shapes,
dimensions and tolerances, etc., which had not appeared on his first
layout. The draftsmen worked on other layouts, on subassembly drawings
and on piece part details, Parts lists also had to be prepared for trans-
-mittal to the manufacturing department, and all components of the design
were re-examined for cost and manufacturability. During Phase 2 Dick was
able to request help from the manufacturing department in reaching decisions
on such things as manufacturing and assembly processes;extrusion die costs,
dip brazing, etc. Ideally, after all these drawings had been checked and
released, Phase 3 would begin. By this time the final prototypes would
also be complete.
Phase 3 is pilot production. Normally manufacturing produces twenty
units once necessary tooling is available; the production of this pilot
lot is still under the control of the engineering department and the design
is evaluated by engineering for ease and cost of manufacturing. Design
changes are made accordingly as Phase 3 continues and the product is
_ debugged. At the conclusion of pilot production, the engineering depart-
ment formally releases the product to manufacturing and their responsibi-
lity for it ends, although they may still serve as consultants to manu-
facturing. Pilot production of the 10/20 actually began in early November,
long before documentation of the design was completed. Work began on the
ECL 47
manufacture of the piece parts for ten computers and as enough parts for
subassemblies became available, these were built up. At the same time
- work began on another lot of 40, with completion of these scheduled by
July to meet sales forecasts. Thus, although it may be convenient to
think of product development activities as divided into three consecutive
phases (design, prototype construction and development, and pilot pro-
duction), in the case of the 10/20 there was considerable overlap and the
three phases were carried out more or less concurrently.
Preliminary Design
During Phase 1 of the 10/20 computer development project, the pre-
liminary design stage, Dick Dunlop and others at Systron-Donner had to
consider, evaluate, and finally choose among many alternative design
possibilities. It had already been decided that the computing modules
would be fitted to pre-wired receptacles and that a removable problem
board would be used, similar to the 40/80 design. When a module is in-
stalled, prongs at its rear connect to terminals in the module receptacle
which are wired to the power supplies and controls. From the front of
each module protrude 84 spring prongs. In front of these prongs is
located a board with 84 corresponding. holes for patch cord plugs. Coded
directions for the various possible connections appear on the front of
each board. The individual boards for all the modules, when installed
in a rectangular frame, comprise the problem board. The shallow prong-
filled cavity in the front of the computer in which the problem board
fits is called the patch bay.
’ Among those things that had not been decided upon were the
following: © oo:
1) The number of computing modules to specify. The six different
modules available for the 40/80 would be available to purchasers
of the 10/20 along with two new modules being designed for use
in both computers. Since the modules are all interchangeable
they must have a common size; this size is 2-1/4" x 16" x 6-3/8".
The size of each coded module board is 2-1/4" x 6-3/8" x 3/8".
2) The number. of coefficient potentiometers to specify -- also
whether the number installed should be a customer option as with
the 40/80.
3) Where to place the variable diode function generator (VDFG) cards
(or boards) and how many to specify. The VDFG cards and their
installations on the 40/80 are described in Exhibit 5. The size
of each VDFG card is 1" x 3-1/2" x 6",
ECL 47:
4) Whether it should be possible to mount the computer in a
standard electronic equipment rack. If so, its width would be
limited to 19 inches. Electronic equipment is often rack
mounted so that related or inter-connected pieces can be
neatly grouped together. For instance, an analog computer
might receive its input from a digital/analog converter with
the readout being shown on an oscilloscope. It might then be
desirable to mount all three together in a rack.
5) How the following components of the computer should be arranged:
--modules, potentiometers and VDFG cards.
--the two power supplies, one of size 6" x 15" x 8", the other
8" x 10" x 6".
--the fan and motor for cooling the power supplies, a purchased
assembly of size 6" x 5-1/2" x 2-1/2".
--the controls: 8 pushbuttons (each 7/8" x 3/8" x 3-1/2" deep),
4 concentric rotary switches (each 1-1/2" dia. x 3" deep),
3 single pole double throw switches (each 3/4" dia. x 2-1/2" deep)
and a voltmeter (3-3/8" x 2" x 2" deep). Dick was free to
lay out the panel and design knobs, etc., as he wished for
good human engineering.
Some thought would have to be given to air flow from the fan
through the electronic components. The amplifiers in the
modules produce the most heat.
6) Styling of the package -- consistent with the engineering and
suited to the anticipated production volume of 30 units per month.
A more specific problem which faced Dick Dunlop was the design of the
problem board and its latching mechanism. On the 40/80 coded module
boards are held in a rectangular frame which is latched to the front of the
computer after the proper connections between and within the modules have
been made with patch cords and patch plugs. : The patch plugs and patch cords
plug into the holes in the front of the module boards and contact the spring
prongs, as shown in the sketch of Exhibit 6. The manufacturer of the prongs
told Dick that when installed correctly they should be loaded by the plugs to.
a deflection at their ends of about .060 inch. The 40/80 problem boards are
programmed first and then latched to the computer. During the latching
operation the board is lifted vertically upwards to load the prongs.
To be able to design a latch arrangement -- or problem board re-
ceiving mechanism -- Dick felt that he should know how much force would
be required to load the prongs. He filled a module board with patch cords
so that he could check for the worst case -- 84 plugs loading 84 prongs.
Using this board and a single module he found that it took a force of about
28 lbs to raise all the prongs .060 inch. He used a platform scale in the
shipping department to measure the load.
ECL 47-é
Exhibit
Page 1
TRANSISTORIZED +100V ANALOG COMPUTER
PRICE and QUOTATION | FORM.
NUE ATE abate?
Tee eee ee es Function
id a Generator
ROR OSE EE Group
- SHORE HCE HEH
~ . Digital
a Teeeeeeeen Voltmeter
Seeeeetece
"VECCt cc ee Computing Modules
bALALL ES So
bb
| @eeee,
eee ewes
Problem Board
Potentiometer Group
SD 40 Computer Cabinet, Controls, and Receptacle
. 4 SD 40 COMPUTER OVERALL DIMENSIONS:
a Cabinetry, Control and Receptacle including: 52"L x 26"H x 23”D
: . «Desk top cabinetry with cooling fans . , : ee to oo “*
Complete control wing with mode and test controls, rep-op, voltmeter, APPROX. WEIGHT: 450 lbs.
Null reference system and digital address selector.
Potentiometer wing for mounting up to 6’potentiometer groups. : POWER CONSUMPTION: 440 watts
Pre-wired module receptacle unit for up to 21 computing modules.
Computer Power supplies and ++ 100V d.c., ¥2 ampere reference supply. PRICE . . . . . . . . $ 8,585.00
“SCCee eee
i Function
4666668666 Ei Generator
SECC EEC OEE + Group
€ECCOC ECO CE 5 .
Power are Digital
S8seesesee: , voltmeter
Computing Medules
Problem Board
Potentiometer Group
: pikes tece ll
Fd
i
pe" Stetewewes pies
‘SD 80 Computer Cabinet, Controls, and Receptacle
GM SD 80 COMPUTER
Cabinetry, Control and Receptacle including: OVERALL DIMENSIONS:
Desk top cabinetry with cooling fans 68"L x 26”H x 23”D
Complete control wing with mode and test controls, rep-op, voltmeter,
Null reference system and digital address selector. APPROX. WEIGHT: 600 Ibs.
Potentiometer wing for mounting up to 6 potentiometer groups. POWER CONSUMPTION: 650 watts
Pre-wired module receptacle unit for up to 42 computing modules. ,
Computer Power supplies and + 100V d.c., ¥2 ampere reference supply. PRICE . . . . . . . . $10,500.00
Exhibit 1: The 40/80 Series of Systron-Donner Analog Computers.
REV.5-65 - ;
ECL 47-A
Exhibit’l
Page 2
UNIT PRICE QTY. TOTAL PRICE
El Basic COMPUTER: $D/40...$8,585.00(] SD/80...$10,500.00C] - $$
FA] POTENTIOMETER GROUP:
Mode! 3370 Pot panel, 20 ten-turn wire-wound pots with counting dials $ 810.00 $
Potentiometer TOTAL (Including 5 on control wing):
FUNCTION GENERATOR GROUP:
Mode! 3350 Function Card Receptacle $ 270.00 $
Model 3351 Variable Diode Function Generator card 215.00
Model 3341 Function Generator Setup Unit 165.00
Function Generator TOTAL
24 COMPUTING MODULES:
Model 3320 Dual Integrator Amplifier $ 700.00 $.
Model 3321 Dual Summer Amplifier 650.00
Mode! 3322 Dual Inverter Amplifier and Dual Operational Relay 540.00
Model 3323 Dual Inverter Amplifier and Dual Electronic Multiplier 945.00 . _-
Mode} 3324 Dual Inverter Amplifier and Quad Electronic Switch 835.00
Model 3325 Quad Summer . $1,000.00
Computing Module TOTAL: including Amplifiers, _______ Multipliers,
Integrators, ______ Relay Comparators, Electronic Comparators
EY picitat Logic CONTROL MODULES:
S-D Quotation No:
Model 3326 Flip-Flops $ 500.00 $
Model 3327 Logic Gates 500.00 $.
Model 3328 Time/Event Control 950.00 $
[Al Removable Problem Board: $D40...$ 270.00] SD80... $450.00 (4 $
Patch Cord and Shunt Plug Assortment (200 items) 200.00
FA Four-place Digital Voltmeter 2,200.00
E) Universal Module Extender . 100.00.
, Instruction Manual, A.C. Power Cords and Spare Fuse Kit . 1 each N/C
Special Instructions or assembly requirements:
Price for special work (if required) . $
TOTAL PRICE, F.0.B. Concord, California . $
(Please reference this number on all correspondence)
Delivery:_..-=~===s—s days after receipt of Purchase Order
Terms: net 30 days
NOTE: Budgetary prices are subject to change with-
SYSTRON-DONNER CORPORATION
out notice. Signed quotations are firm for a
period of 30 days.
Signature Date
Printed in U.S.A.
Converting a SD 40 to a SD 80 computer is accomplished merely by inserting
an additional receptacle which doubles the computing module capacity from 42 to
82 amplifiers, and exchanging problem boards and top and bottom cabinet plates.
GENERAL DESCRIPTION
The SD 40/80 series of general purpose
analog computers are constructed of fully
transistorized circuitry and operate over a
full +100 volt computing range. General
configuration is a desk-top design, pre-
wired, with removable problem board,
‘modular computing elements, and movable
control and potentiometer wings. _
The entire series is designed to solve
ordinary and partial differential equations,
and other engineering, design, and control
equations by solution, simulation or logic
analysis. The SD 40/80 Computers include
operational program check circuitry, a stor-
able program set-up system, and a com-
pletely short circuit proof design that pro-
tects computing components as well as the
power supply against errors in patching
and accidental shorts to ground. :
COMPUTING EQUIPMENT
+100 volt Operational Amplifier is a removable,
identical, dual-channel unit which is included in
each computing module. (Specifications as meas-
ured at the Problem Board. )
Maximum Output Voltage +105 v.(1 ma)
Output Voltage (at +25ma) +100v
Maximum Output Current (at +100v) +25ma
Overall DC Gain => 10’
Summing Junction Offset/8 hours 100 xv
Summing Junction Offset due to
+10% line variation <204¥
Short-term Stability (referred to
Summing Junction) * 20 ay
Noise (referred to Summing Junction) <2.5 mv P-P
SD 40 SD 80
14 Integrating Amplifiers 28
14 Summer Amplifiers 28
6 Multiplier-Dividers with 12
6 Inverting Amplifiers 12
4-15 Variable Diode Function Generators 8-15
with Inverting Amplifiers
8 Electronic Switches with 16
4 Inverting Amplifiers 8
4 Operational Relays with 8
4 Inverting Amplifiers 8
up to 65 Coefficient Potentiometers up to 125
5 Function Switches 5
80 Trunk Lines 160
1764 Problem Board Terminals 3528
52"L x 26"H x 23”"D Overall Dimensions 68"L x 26”H x 23”D
300 Ibs. Approximate Weight 370 Ibs.
440 watts Power Consumption 650 watts
self-contained Cooling self-contained
Control Wing (on left) and Po-
tentiometer Wing (on right) are
hinged. They swing to any conven-
ient angle to give operator total
visibility and control.
Modular, plug-in computing ele-
ments form patchbay to provide
high dynamic accuracy. Gold con-
tacts, with double wiping action,
assure positive connection.
A centralized, fully expanded pow-
er supply system is comprised of
four supplies, each individually
fused and fully short circuit proof.
# APG iGifar aie ven’
Removable problem board, made
up of color-coded patch panels,
couples directly into:computing
elements which can be arranged
in any convenient order.
Exhibit 2: Systron-Donner 40/80 Series Computer.
VEqTUXg
"49 Toa
Plug-in Computing Modules
- * Front panel of module forms patch bay - no lengthy interconnecting cables
Exhibit 3: Computing Module,
€ AtAyuxg
Vel9 Tod
Exhibit 4:
Part of the Proposal for the 10/20 Computer.
APPROVALS* ECL 47-A
Exhibit 4
A.B. Blessing Page lL
Q. James ;
O. Reese
F.L. Kazabowski
G. Washington
Copy only:
O.P. Henry
F. Virgil
SYSTRON-DONNER CORPORATION
Electronic Instrumentation
Date; 12 January 1965
New Product Proposal
Proposal: EI-5B
Name of development: Desk Top Analog Computer, Model SD 10/20
Description:
A) Control Center
1.
The five modes of operation: Reset, Compute, Hold, Balance
and Potset shall be controlled by a rotary switch.
An optional plug-in unit to provide rep-op operation. Both
Compute and Reset modes of the Rep-Op operation shall be
continuously adjustable from 5 msec to 1.0 sec.
An address and meter select system shall be of rotary type
switches. The address system shall be capable of monitoring
20 amplifier outputs, and 30 potentiometers through the pot.
bus, All power supply voltages shall be available for moni-
toring either on the problem board or control center.
A voltmeter read-out device of the taut-band type shall have
an overall accuracy of + 2% of full scale.
A differential voltmeter read-out system for pot setting shall
have an overall accuracy of +0.05%.
A visual master overload indicator of.the latching type shall
respond to any momentary or sustained overload condition from
any amplifier in the computer system.
A slave option to make control of the computer possible from
a remote source such as another computer through the trunk
lines of the integrator network module. -
Three function switches of TIDP type shall be available./
Visual indicators to indicate modes of operation, overload,
power-on shall be available.
* Fictionalized names (Case Writer).
Description:
ECL 47-A
Exhibit 4
‘Page .2
(continued)
B) Power Supplies
1. All power supplies shall be fully protected from accidental
shorting to ground with no consequences.
2. All components of each power supply shall be capable of with-
standing a cabinet temperature of 50°C (122°F).
3. The power supplies shall have the following specifications:
+112V +100V +28V
Output current 500 ma 150 ma 500 ma
Load regulations (0 to full load) 112 mv - 10 mv 28 mv
+10% line-change regulation 112 mv 10 mv 28 mv
Ripple with full load 20 mv 5 mv 10 mv
Tracking , - 3 mv -
-4, The +100 V reference supply shall have an 8 hour stability
of 10 mv under constant temperature, line-voltage, and
loading conditions.
5. The temperature stability of the +100 V reference supply
over the temperature range of 0°C (32°F) to 43°C (100°F)
shall be better than 75 mv (less than 1 mv/°F).
6. All power supplies shall be capable of 115V or 230V
(50 cps to 60 cps) operation.
C) Computing Modules
1.
All modules shall have the same physical dimensions, number
of terminals on front panel as those of the SD 40/80.
There shall be three separate types of modules, namely:
quad-operational amplifier, quad-integrator network, and
dual multiplier/comparator.
Quad-operational amplifier module shall contain two dual-
operational amplifiers, Model 3310, four precision resistor
plug-in type of P.C. boards, each shall have five wire wound,
.01% resisters. No additional patching shall be required
for amplifier balance. Balance adjustments shall be avail-
able on the front panel. Optional choice of the number of
dual-amplifier boards and precision resistor boards can
conveniently convert the quad-operational amplifier module
to a dual operational amplifier, dual-summer or quad-summer
module.
ECL 47-A
Exhibit 4
‘Page 3
Description: (continued)
4. There shall be two types of quad-integrator network, one
for real-time another for rep-op operation. When either
integrator network is used in conjunction with the quad-
summer, they become two separate dual integrators capable
of real-time or rep-op type of operation depending on the
type integrator network used.
a) Quad integrator network for real-time operations shall
contain two separate plug-in type P.C. boards each with
two 1.0 MF, 0.05% and 0.1 MF, 0.05% polystyrene capa-
citors, four 50K, 0.01% resistors, and two 2-C mechani-
cal relays to provide logic control for two channels of
integrator network.
b) Quad integrator network for. rep-op operation shall con-
tain two separate plug-in type P.C. boards each with two
1.0 MF, 0.05%, 0.1 MF, 0.05% and 0.01 MF, 1% polystyrene
capacitors, four 50 K, 0.01% resistors, two 2-C relays
(one shall be high speed type) to provide logic control
for two channels of integrator network.
5. Dual multiplier/comparator module shall contain two channels
of 1/4 square type of multiplier as used in the SD 40/80 and
two channels of comparator that require no external opera-
tional amplifier. Each comparator channel shall contain
an unstabilized amplifier which output shall be connected
to a 2-C relay capable of fast operation. Each comparator
_ channel shall occupy a P.C. board of the plug-in type.
Optional choice of multiplier board and comparator board
can conveniently convert the dual.multiplier/comparator
module to dual multiplier, dual comparator, or dual
multiplier/comparator module.
6. Function generator board shall be same as those in SD 40/80.
Their termination shall be available on the integrator net-
work, multiplier/comparator module problem boards.
7. There shall be two types of potentiometers available, the
10 turn wire wound and single turn molded carbon. With the
10 turn wire-wound potentiometers, plain black plastic knobs
or duo-dial helipot knobs shall be made available. Pot.
panel shall have groupings of ten potentiometers of either
type. A push-button type of switch shall be used with each
pot. to perform the functions of pot. setting and monitoring.
D) Cabinet Assembly
1. The construction of the removable problem board, problem
board receiving mechanism and module receptacle are essentially
the same as the SD 40/80. Attempt to reduce the cost of the
existing design shall be made.
-ECL 47-A
Exhibit 4
Page 4
Description: (continued)
2.
The function generator receptacle unit shall be located: on
the bottom level of the computer. The receptacle unit shall
‘accept 5 function generator boards.
The pot. panels shall be mounted next to the problem board
area on the right side of the computer.
The control center shall be located next to the problem board
area on the left side of the computer.
The front side of the computer shall be a single plane con-
struction. The cabinet shall be a simple box-type shape
which may be purchased.
All power supplies shall be either behind the control center
or the pot. panels.
Effect on Established Systron-Donner Products: None, but does extend
the SD 40/80 Analog Computer series into a similar market area of the
Donner 3400 which has proven to be a widely accepted computer model
over the years. The SD 10/20 modules will provide new additions to
the module family of the SD 40/80.
ECL 47A
Exhibit
Function Generator
4 fo poteof
Input Voltage +100 v maximum
Output Voltage © Arbitrary function of input voltage; within
the range of =100v
Frequency Response _ 1 kc
Input Impedance | Greater than 45 k® (depends on function)
Output Impedance —_ Equal to the output Z of the output amplifier
Function Simulation Straight-line approximation of 12 line
segments
Line Segments 12 breakpoints total
(6 adjustable between 0 and +100 v)
(6 adjustable between 0 and —100 v)
Slopes _ Each segment has a maximum adjustable
slope of 2.5 v/v input. (Larger stopes are ob-
tainable by adding individual line segments.)
Noise 150 mv P-P
Power Requirements +100v, +6 ma
mo '
a in :
a : nN
; ‘ . . , i .
, 3, : a :
, : Aa
ca aa -
. % eat A, i tees (4 y kee |
The Function Generator Receptacle mounts up to 15 cards and ter-
minates each FG channel at the Dual Summer, Model 3321, or at
the Inverter/Operational Relay, Model 3322. Each FG card contains
12 independent segments, all have screw driver pot adjustment for
break point and slope. This flexible FG card set-up permits “stack-
ing” of breakpoints for increased slope, and 24-segment function
generation by paralleling two FG cards.
Exhibit 5: VDFG Cards and their Mounting on the 40/80,
SPRING
],
«
MODULE
ee
PRONG
PATCH CORD
PATCH CORD PLUG
CODED MODULE BOARD
Exhibit 6: Spring Prong and Patch Cord.
ECL 47=A
Exhibit
ECL 47
SYSTRON-DONNER CORPORATION (B)
Design of an. Analog Computer
Dick Dunlop began Phase 1 of the development of the new model 10/20
analog computer during March 1965. He was expected to propose three or
four alternative package designs.
One of Dick's preliminary design sketches, made in April, is shown
in Exhibit 1. This computer incorporates ten computing modules and is
suitable for rack mounting; however, Dick did not like the upright
appearance of this design nor of the several others he sketched that
were limited in width to 19 inches so that they could be rack-mounted.
He noted that the principal competition in the small portable analog
computer field would be from an upright + 10 volt machine which could
be rack mounted. However, Dick felt that this machine looked more like
a slot machine than a computer. He began to feel that it would be
impossible to design an aesthetically satisfactory computer that could
be rack-mounted and turned his attention to designs with more pleasing
proportions.
When Dick first decided to try a design with all the computing
modules in a single horizontal row,, he made the sketch shown full size
in Exhibit 2. .He showed this sketch to several people concerned with the
project and it met with favorable response. Then he went ahead and made
a number of larger sketches based on this idea, one of which is shown
(reduced in size) in Exhibit 3. This design contains only seven modules
ECL. 47
and was made to investigate rack mounting of a computer having a single
row of modules. Dick became quite enthusiastic about the basic idea
common to Exhibits 2 and 3, so much so that he felt it was the only design
worth pursuing for the 10/20, and he made a number of further sketches
investigating modifications and variations in the location of components
such as the VDFG cards. He also sketched computers with differing numbers
of modules, He said he arrived at this basic design by a deductive
process -- by considering designs and rejecting those he didn't like until
he finally came up with one that he did like. Although he was supposed
to present several alternative designs to management for their choice,
Dick felt that it would be a good idea to present his preferred design
in as forceful a manner as possible. With the sketches having served their |
purpose, he then had the model shop make a full-size model of the cabinet
and front panel of a nine module computer, one virtually identical to his
original sketch of Exhibit 2, and thus too:-wide for the rack mounting.
This model incorporated the sheet metal cover, trim molding, dummy knobs
and. switches, and a problem board. Work on this model started in the
middle of May. Its design received management approval with no signifi-
cant changes and work, as previously mentioned, started on the engineering.
prototype on July lst, with hopes of completion before the WESCON show
in August.
The final design has room for nine computing modules, up to 24
coefficient potentiometers, and four VDFG cards. .A description of the
computer appears in: Exhibit 4. Nine modules were finally decided upon
because it was felt that eight would be too few but ten too many.
The number of coefficient potentiometers is about the same ratio of
modules to pots as for the 40/80 series. .A 10/20 price list appears in
Exhibit 5.
One of the sketches Dick made as an aid to estimating manufacturing
cost of the computer appears in Exhibit 6. Dick explained that the cost
of the tooling for the frame and cover of the computer totalled about
$300, with the extrusion die for the trim molding being about. $75 of this.
Photographs of the chassis and cabinet of the 10/20 appear in
Exhibit 7. The chassis is shown from the rear in both pictures. The
fan blows air along one side of the computer; then the air hits the
front panel and is deflected upwards through the modules. There are
outlets at the top rear.of the cabinet. This design allowed only
filtered air to enter and also allowed natural convection to assist in
cooling.
Another of Dick's blow-up sketches appears in Exhibit 8. In this
sketch and that of Exhibit 6 the problem board and part of the latch mech-
anism can be seen. Photographs of a problem board with patch cards, patch
plugs, and overload lights appear in Exhibit 9. The overload lights indicate
non-linearity in an amplifier. The nine module boards are held from the
back of the problem board against flanges on the top and bottom of the
' B-2
ECL 47
frame. Each end of the frame is a channel-section aluminum extrusion. When
installing the problem board, it is first pushed back into the patch bay.
_Four spring loaded ball plungers, .or ball .detents, hold the problem board in
place when the rear legs of the channels on the ends of the frame are pushed
past them. Two ball plungers are pressed into holes at each end of the cab-
inet as shown in the layout of Exhibit 10 and the photographs of Exhibit 11.
Only part of the drawing Dick made is shown in Exhibit 10 and explanatory notes
have been added. Dick placed two hard wear pads on the inside of each rear
channel leg (Exhibit 9) when he found that the aluminum channel would be
severely scored by the ball plungers. When the problem board is snapped into
place past the ball plungers there is clearance between the patch cord plugs
and the spring prongs in the patch bay. a
Dick used a camshaft to move the problem board up 1/8 inch and load
the spring prongs. The cams are actually flats, 1/8 inch deep at their center,
milled in a 3/4 inch diameter stainless steel shaft, as can be seen in
Exhibits 10 and 11. On the layout, the shaft is shown in the up position with
the stepped oilite foot on the bottom of the problem board (Exhibit 9) resting
on the circumference beyond the slot. This corresponds to the up (vertical)
position of the lever at the side of the problem board. With the handle and
shaft rotated 90° clockwise (on the drawing) the oilite pads sit in the milled
slots and the problem board can be removed or installed. Part of the handle
appears on the drawing in phantom view for this position. When the problem
board is raised it is positively locked; the ball plungers no longer hold it
alone. The four plungers actually serve only as a convenience during loading.
to hold the problem board in place while it is being lifted.
The stainless steel camshaft sits in slots milled in aluminum blocks. At
first Dick had planned to mount it in bronze bushings, but then he decided that
wear of the aluminum would probably be, negligible, :since-the shaft would be
turned only through 90° and then probably only a few times a week.
Dick had also designed a camshaft to lift and lock the problem
board on the 40/80 computer and it was a natural step to use a similar
arrangement for the 10/20. He recalled that the camshaft idea had seemed
to have a self-generating origin, that after toying for awhile with all the
requirements of the problem and the parts that already existed, the only
feasible way to move the problem board upwards seemed to be with cams.
Using a handle turned through 90° to rotate a shaft with two eccentric
supports then followed logically. During the 40/80 design period Dick had
considered replacing the handle by a torque motor with a worm reducer to
drive the camshaft. He investigated available motors and reduction units
but found that a very high reduction would be needed to give sufficient
torque output to raise the problem board and that the best combination he
could assemble would take about five seconds to raise the board. Also,
the cost was too high.
B-3
ECL 47
The 40/80 camshaft is machined from 7/8 dia. shafting with an
eccentric 1/2 dia. x 1" W. at each end. At first Dick planned on a similar
arrangement for the 10/20. He decided to check a 3/8 inch diameter shaft
for the handle force needed to lift the problem board and for the accompanying
shaft deflection and twist. His calculation sheet appears in Exhibit 12.
On the basis of these results he felt a more rigid shaft was necessary and
for the first prototype specified a 3/4 inch diameter shaft with a flat
milled its entire length to provide the 1/8 inch lift. Later he realized
it would be much simpler to cut only two flats in the shaft, each 5/8 " wide
x 1/8 '" deep.
The Phase 2 engineering prototype, begun after the completion of the
WESCON model in August, was finished in January 1966. No significant
changes were made to the design during the construction of this prototype,
or later in the project; however, Dick pointed out that a number of dimen-
sional errors in the drawings were caught at this stage and that it was a
lot better to find them before putting the computer into production. With .
Dick and three others still working on the documentation, it was about
90% complete by March 1966, and Dick expected it to be finished by August.
The first batch of ten computers, all of which had already been sold, were
completed and ready to be shipped by the end of March.
B-4
COOLING
Boty Sipés
Not USED Op eAck
Mobet)
'INTakE
B-Donloe
SYSTRON HonneR_
Exhibit ls One of Dick Dunlop's Early Sketches,
\ Po ver Seer ly Sticks ox 4
Cow oo.
AERA
7
CGowre ot CEwreez t PotsS
ow ONE Pawar '
SD So StTanpace
MopneL
SHow\n Covere. Removed
For S apne an
ower Supe ly in REAR
Conteor Centar $ Pot Panett
SWitag OoT A)
Reduced in Size
atqtuxg
“19 log
ECL 47=B
Exhibit 2
1%
{f
WD NLAe aces |< eo
Exhibit 2: Dick Dunlop's First Sketch (Original Size) of a Co