Analogue Computer (Part 1)
> Steep eeneemes
PRACTICAL
ELECTRONICS
VOLUME 14 No.13 SEPTEMBER 1978
CONSTRUCTIONAL PROJECTS
ANALOGUE COMPUTER-1 byP. J. Kronis BSc*
Mathematical operations and circuits 970
COMBINATION LOCK 1 by€£.A. Parr
Double combination for added security 990
THERMOSTAT CONTROL by ™. Edmunds
For the photographer who does his own developing 1002
SOUND TRACK MONITOR by J. Schmid
Alerts the operator when the end of a recording is reached 1006
METRONOME by ™. Butt
Simple emphasised beat unit 1010
KEYBOARD by. G. Parkin BA
Provide an eight bit binary word by pressing two keys 1014
GENERAL FEATURES
WAVE ENERGY by MU. Abbott
Can the sea provide the UK with electricity? 976
STRICTLY INSTRUMENTAL by X. Lenton-Smith
Signetics TDA 1008 electronic music i.c. 988
INGENUITY UNLIMITED
Simple Clock—Beethoven’s Doorbell—Electronic Combination Lock—Distortion
Assessment—Protection for a Model Train Speed Controller—Simple Alarm 993
SEMICONDUCTOR UPDATE byf. W. Coles
A look at some recently released devices 998
NEWS AND COMMENT
EDITORIAL 969
‘READOUT
A selection of readers letters 974
BOOK REVIEWS
Selected new books we have received 980, 1013
MARKET PLACE
Interesting new products 985
INDUSTRY NOTEBOOK by Nexus
What's happening inside industry 1001
SPACEWATCH by Frank W. Hyde
Soyus-29 and Salyut-6, USSR Launchings, India and the USSR, GOES-3,-Place in Space 1005
PATENTS REVIEW
Thought provoking ideas on file at the British Patents Office 1020
Our October issue will be on sale Friday, 8 September 1978
(for details of contents see page 975)
© IPC Magazines Limited 1978. Copyright in all drawings, photographs and articles published in PRACTICAL
ELECTRONICS is fully protected, and reproduction or imitations in whole or part are expressly forbidden. All
reasonable precautions are taken by PRACTICAL ELECTRONICS to ensure that the advice and data given to
readers are reliable. We cannot, however, guarantee it, and we cannot accept legal responsibility for it. Prices
quoted are those current as we go to press.
Practical Electronics | September 1978 953
>
Fig. 1.1. A typical lissajous figure produced using the
Analogue Computer and an X-Y plotter
field the high gain d.c. amplifier or operational amplifier
which is the main element of the analogue computer, has
also come a long way since its inception. It was originally
designed for use in computers but has since found many
applications in other fields. This large market for other
applications has reduced the cost of such devices to very
low levels. Of the numerous op-amp i.c.s available on the
market the 741 was chosen for the prototype because it is
both cheap and easy to handle. More advanced op-amps are
available albeit at a higher price and constructors can
experiment with these if they wish.
By connecting an op-amp to input and feedback
components certain mathematical operations can be
performed; addition (and subtraction) integration, and
multiplication by a constant. Differentiation can also be
performed but is generally avoided due to problems
associated with noise generated by components.
Multiplication by constant coefficients between zero and one
is also performed using potentiometers with some special
circuits being employed to enable the multiplication of two
variable voltages.
THE ADDITION CIRCUIT
It is possible to add various voltages by means of a
resistance network with the output voltage being
proportional to the sum of the input voltages. The serious
drawback of this method is that this is only true if the load
resistance remains constant.
Practical Electronics September 1978
This would be an unacceptable constraint since the output
voltage may be applied to other points in the circuit which
have different values of load resistance.
To overcome this difficulty a high gain d.c. amplifier is
employed in the feedback circuit as shown in Fig. 1.2.
Ri Rf
v) ——$ WA VV
R2
V2 sap eee Sd a? ie avs
Vy O—$AA/\A—_—4
R4
Fig. 1.2. “Addition” circuit
If a voltage V, is applied via R, to the summing junction
the output voltage V, is equal to
RE
R,
The polarity of the input voltage is also changed by the
operational amplifier.
With the output voltage now independent of the load
resistance each input voltage is factored by the same ratio of
feedback resistance to input resistance.
-V,
Rf Rf Rf Rf
V=-{ V, —+4+V, — 4+, —— +M=
a ee Satie 6 Ys Ry grea
THE INTEGRATOR CIRCUIT
As with the addition circuit integration can be achieved by
using an R.C. network but this method also suffers from a
number of serious drawbacks.
The circuit in Fig. 1.3 shows how an operational amplifier
can be used to perform integration.
R1
v1 ©
Cf
R2
- R3 Bis eae
V3 ON A
Ro
Ve MAS!
Fig. 1.3. ‘Integrator’ circuit
With a capacitor connected in the feedback loop, and if
the open loop gain of the amplifier is very large, the output
voltage is given by
ae 1 1 1
=~ (—— [vat +—— [| v,dt +—— [V,dt + —— [ V, at
° az! ‘ el e cr | ‘ Rc | ‘ )
The output voltage is the sum of the integrals, with
respect to the time the voltage is applied to the inputs,
factored by — _ Ct
By choosing suitable values of Rin and Cf the factors can
be given the required values.
971
nape.
acs be Boys ui Ce
BS verso sa
ats
THE COEFFICIENT MULTIPLIER
The coefficient multiplier is used to multiply a voltage by a
constant between zero and one. This is the only
mathematical operation that is usually performed without
the use of an op-amp. A potentiometer is connected as
shown in Fig. 1.4.
At one extreme of the slider’s travel Vo=Vin, i.e. Vin is
multipled by 1, whereas at the other extreme Vo=0 i.e. Vin is
multiplied by zero.
Any intermediate value can be set up by moving the slider.
The dial of the potentiometer can be calibrated to facilitate
this. However, it is not normal practice to set up a value on
Vin
Yo
Fig. 1.4. Coefficient Multiplier
the dial of the potentiometer because this circuit also suffers
from the effects of load resistance. :
An op-amp employed as a voltage follower could be
connected as a buffer to isolate the effects of the load
resistance, but this is an unnecessary addition because the
problem can be overcome by measuring the output of the
potentiometer using a voltmeter, after the circuit has been
connected, i.e. in the presence of the real load to be applied
in the particular problem being examined. The value desired
is then set by adjusting the potentiometer and ignoring the
graduations on the dial.
The circuits described so far form the fundamental
building blocks of the analogue computer. Various special
circuits have been developed over the years for other
operations. The most important of which is the formation of
the product of two variables. One of the early methods
developed was the cumbersome servo multiplier. This
involved the control of potentiometers using servos.
Nowadays this operation can be achieved electronically
using four-quadrant multiplier integrated circuits.
INTEGRATION
Addition, subtraction and multiplication are concepts that
are easily understood; integration, however, is not so easily
grasped. by the non-mathematically minded and so a simple
explanation may be useful at this point.
If for example a motor car is cruising on a motorway at 50
miles per hour this can be represented by a graph of speed
against time (Fig. 1.5). Since the speed is constant the
distance travelled will increase by equal amounts in equal
SPEED
m.p.h.
50
» 1!ME
hrs.
+ 4 ‘4 ..
+ 25, +
i) 1 2 3 4
Fig. 1.5. Graph of speed against time
972
DISTANCE
miles
TIME
hrs.
Fig. 1.6. Graph of distance against time
time intervals. These distances are shown plotted on a graph
of distance against time for intervals of one hour (Fig. 1.6).
It can be seen from Fig. 1.5 that the distance travelled
during a period of time is represented by the area shown
shaded on the velocity-time graph. (Velocity x time
representing the height x base of the shaded rectangle.)
Now if the results of all these intervals were added up, the
result would be the total distance travelled in a period of
time.
The mathematical way of saying this is that the distance
travelled is the integral of velocity with respect to time
between two time limits. In the above example since the
speed was constant one could have arrived at the required
result by multiplying the total period of 5 hours say, by the
velocity of 50 m.p.h. to obtain 250 miles travelled, without
going into the trivial process of integrating, by considering
small time intervals. =
In reality the velocity may vary as shown in Fig. 1.7, i.e. in
a random manner. To obtain the required result then, the
velocity would have to be integrated over the required period
of time by considering small time intervals. This is how a
digital computer would be programmed to solve the
problem. The accuracy in that case would depend on how
small the time intervals were made. This is left to the
discretion of the programmer. If the intervals were made too
big, then the result would be inaccurate. On the other hand
too small a time interval would mean that the computer
would take longer to solve the problem and involve the
programmer in unnecessary expense. The analogue
computer programmer need not worry about this since the
computer integrates continuously, i.e. it deals with
SPEED
TIME
Fig. 1.7. Graph showing variations in velocity
Practical Electronics © September 1978
infinitesimally small time intervals and does this at high
speed. ;
Each of the circuits that have been described so far
constitutes a computing element. When the computer is
programmed to solve a problem, systems of equations can
be set up by connecting together combinations of computing
elements, and the results can be obtained by measurements
taken at various points in the system.
The computer will of course be required to solve many
different problems and the computing elements will have to
be rewired every time. To facilitate this a patch panel is used,
with sockets connected to each computing element in the
computer. By using wire leads the computing elements can
be connected in any order.
INTEGRATOR
Ri uy
R2
v2 Oo——-AA\A——+
=a A oO
v3 0—_\/\/\-———_?
Ro
i ——
“Compute”
Ra Ct
WV ae
2
V2 PEROT CORO
R3 q pO 0
V3 Qe AAS
Ré Fe
“Hold”
Ric Ric
Vic jr A\—
Ri ct
“1 © Ae -—_it+—_+
R2
V2 Omen /\ emma 1
R3 : § bp Vy
3, Oo
RL
vy Ome Nye
“*Reset”’
At the beginning of a computation the variables of the
problem will have certain values, not all of which need be
zero. The requirement here is that is should be possible, if
desired, to give the output of integrators a value, before the
computation commences. This facility is called “Initial
Conditions”. ;_ %
Fig. 4:7 shows how the “Initial Conditions” for—the
“Compute”, “Hold” and “Reset” facilities are achieved for
summers and integrators. In the case of the summers no
change in the circuit is necessary. For the integrators, the
“Hold” mode requires that the input resistors are
disconnected from the op-amp and grounded. In this way
the charging or discharging of the capacitor stops and the
op-amp maintains the charge at a constant level.
SUMMER
RI Rf
Vj: Onno Amenieey —W—
R2
ot
” R3 A en V0
R4
v oe
“Compute”
Rt Rf
Vv, Om AAA, AA
R2
V2 mA A Amn
R3 IN, “OVO
V3 Om A Nn
Ro
vs oO MAA
“Hold”
RI Rf
R2
oo NAA
v2 or Be OVo
V3 VY ae
R4
““Reset”’
Fig. 1.8. ‘‘Initial Condition’ circuits for Integrators and Summers _
MODE CONTROL AND INITIAL CONDITIONS
The main modes of operation are compute, hold and reset.
When in the compute mode the computer proceeds to solve
the problem. As it is sometimes desirable to stop the
computation after a certain period of time this is achieved by
putting the computer into the ‘!Hold” mode. The “Reset”
mode is used to make the output of all computing elements
take their initial value. Sometimes this mode is called
“problem check”.
Practical Electronics September 1978
The calculation is therefore frozen and the results can
then be observed at leisure. This, however, should not be
practised literally, since electronic components, like
everything else, are not perfect and some drift will always
affect the results. These should therefore be noted as soon
as the ‘‘Hold” mode has been selected.
The “Reset’’ mode for the integrators has two resistors R;
in the circuit. These are the “Initial Conditions” resistors an
when an initial condition voltage,V,. is applied as shown, the
973
=e
ee
INPUT | ouTPUT
[
A s
COMPUTING Move
ELEMENTS CONTROL
OVERLOAD WARNING
Fig. 1.9. Block diagram of the Analogue Computer
feedback capacitor charges up to this value. When
“Compute” is selected these resistors are disconnected and
the output of the amplifier, i.e. the voltage across the
feedback capacitor, may vary above or below the initial
condition value. When “Reset” is reselected the feedback
capacitor discharges or charges, through R,. to V,, and the
computer is again ready for a repeat of the calculation.
THE OVERLOAD WARNING FACILITY
This facility, usually employed in analogue computers, is
necessary because the voltage range over which operational
amplifiers operate linearly, is limited to approximately +13V
for readily available i.c.s. In the course of the solution of a
problem, all computing elements must operate within this
range, otherwise the wrong results will be obtained. The
overload warning circuit warns the programmer of any
amplifiers that have saturated. Measures can then be taken
to scale down the values of the variables.
It is now possible to imagine the general arrangement of
an analogue computer and this is depicted by Fig. 1.9 ina
block diagram form.
To summarise, input signals are fed to the computing
elements via the patch panel and are processed. The results
are fed back through the patch panel to the output, which
may be an ordinary voltmeter, a CRO or an X-Y recorder. The
operation of the computing elements is controlled by the
Mode Control and the overload warning circuit monitors
the output of the computing amplifiers and warns the
programmer of any saturating amplifiers.
NEXT MONTH: CONSTRUCTION DETAILS
POCCOs
... a Selection from our posthag
Readers requiring a reply to any letter must include a stamped addressed envelope. ;
Opinions expressed in Readout are not necessarily endorsed by the publishers of Practical Electronics.
Champ Waves - \
Sir—I hope you can clear up the confusion
COMPLEMENT
ADDRESS
BINARY
ADDRESS
Y Too Powerful
Sir.—Working as Product Marketing
—40 to —48V
that has arisen about your EPROM ou
programmer in the CHAMP series.
When purchasing INTEL 1702A
EPROMS I was sent a data sheet, which
detailed the programming voltages as ~ 48 to ~48V
ov
PULSED VDD \ /
Engineer for the UK’s largest distributor of
National Semiconductor products I was
highly amused by the letter which appeared in
the July issue of P.E. from reader R. G. Silson.
I can only assume from reading his letter
completely different from those produced by
CHAMP-PROG. Since you said that INTEL
had supplied the basic circuit for your project,
and use it in their “Intellec” development
systems, it has resulted in much head
PULSED VGG
-35 to -40V
ov
that he must be extremely well versed in the
world of microprocessors—indeed he must
know far more than the vast majority of
industry’s electronics engineers.
scratching on my part.
The waveforms given on the data sheet are
PROGRAMMING PULSE
Dealing with engineers every day from all
fields of the electronics world I quite naturally
get a very good indication of their thoughts
and feelings towards various projects.
The number of times I have spoken to
DATA customers about the Pace microprocessor,
pide ue Le: only to be told “Not interested—it’s too
as shown. ~46 to -48V
Any. clarification you can give will be ov
greatly appreciated.
T. G. Keslake
Romford ~46 to -48V
Essex
powerful for what we need”, is more than
ample evidence for myself that Mr Silson is
completely out of touch with the amount of
I can understand your confusion over the
difference between the 1702A data sheet and
the operation of the CHAMP-PROG board,
but really it is quite simple. You will notice in
the data sheet that all voltages are related to
GND or 0 Volts, and this means that all chip
voltages are related to the Vcc pins. In
CHAMP-PROG the voltages appear to be
positive going, but if you look at the Vcc
reference pins you will find that they rise to
+47V during programming, and this means
974
that the program pulse is a 3ms —47V pulse
as required. As with many things in
electronics, the secret lies in viewing the circuit.
operation with one’s feet firmly on the ground
(or in this case, the ceiling!). If you check the
other supplies with this new perspective, you
will find that they are substantially as dictated
in the data sheet.
Once again, I quite understand your initial
confusion!
R. W. COLES
knowledge possessed by the average amateur
actively engaged in microprocessors. Further
proof of this is the vast amount of 8 bit
SC/MP chips sold related to the relatively
slow moving Pace.
P. V. Hodson,
Melton Mowbray,
Leicestershire.
Practical Electronics September 1978
Australia 85c South Africa 80c New Zealand 85c Malaysia $2.25
PRACTICAL
OCTOBER 1978 45p
From pracT
iEcTRON= =
Road
ayh
in Le cept sercheoarl Seenonts Kent
7 <
ts
<
:
y
7
%
PRACTICAL
ELECTRONICS
VOLUME 14 No.14 OCTOBER 1978
CONSTRUCTIONAL PROJECTS
P.E. V.D.U.SYSTEM—1 byA.A. Berk, B.Sc., Ph.D.
Up to date ‘‘one chip” memory mapped system 1054
FUEL CONSUMPTION METER ByJ. McCarthy
An aid to economy, suitable for most cars 1060
HIGH PERFORMANCE POWER SUPPLY UNIT by A. Lawrence, B.Sc.
Voltage control down to zero, plus current limiting 1070
ANALOGUE COMPUTER—2 by P.J. Kronis, B.Sc.
Construction details 1074
TWO RANGE TIMER by/J. D. Jardine
A portable, inexpensive general purpose unit 1088
GENERAL FEATURES
IMPEDANCE by Joby Bailey and Bob Whitaker
Do you understand it? 1066
MICROBUS byD.WJ.D.
A bi-monthly focus on micro’s for the home constructor 1098
INGENUITY UNLIMITED
Capacitor Continuity Tester—Synthesiser Repetitive Waveform Generator
Simple Fuzz—Accenting Metronome—Stereo Indicator
External Input Unit for Synthesisers 1080
NEWS AND COMMENT
EDITORIAL 1049
MARKET PLACE
New products 1050
SPACEWATCH by Frank W. Hyde
Pluto, More from the USSR, Copernicus discovers Black Hole 1053
POINTS ARISING
Linear Capacitance Meter, Dimwit, Kiln Controller 1058
BOOK REVIEWS 1084
NEWS BRIEFS
Strain Gauge—Big Brother Check—Micro Power Pack—Computers Galore Club 1087
Underground Cameras—Steam Advice—Here’s To Progress 1090
Teletext Course 1094
Disc Full of Holes 1102
HOW TO USE YOUR FREE STICKIES 1093
INDUSTRY NOTEBOOK by Nexus
What's happening inside industry 1097
PATENTS REVIEW
Thought provoking ideas on file at the British Patents Office 1100
READOUT
A selection of readers’ letters 1102
Our November issue will be on sale Friday, 13 October 1978, price 50p
(for details of contents see page 1059)
© IPC Magazines Limited 1978. Copyright in all drawings, photographs and articles published in PRACTICAL
ELECTRONICS is fully protected, and reproduction or imitations in whole or part are expressly forbidden. All
reasonable precautions are taken by PRACTICAL ELECTRONICS to ensure that the advice and data given to
readers are reliable. We cannot, however, guarantee it, and we cannot accept legal responsibility for it. Prices
quoted are those current as we go to press.
Practical Electronics October 1978 1033
Te Se UTI Ta
bey AVING formed a general picture of the workings of the
analogue computer, the complete circuit of a computing
element can now be described. This is shown in Fig. 2.1. The
basic circuits of input and feedback, components connected
around the op-amp can be readily recognised. The input
comprises four resistors, R, to R,, which are connected to
sockets in the patch panel and to the inverting input of the
op-amp, via switches RLA2, Sic, and S1b. The feedback
circuit consists of R5, C1 and C2, which can be selected by
means of switch S1a and sockets (C7, B6 and C6) on the
patch panel.
Consider switch S1a set so that R5 is selected in the
feedback loop. The computing element now becomes a
summer. By recalling the equation for the addition circuit
that was described last month and by substituting the values
for R5, R1, R2, R3, and R4 it can be seen that a voltage
applied at inputs 1 and 2 will be multiplied by unity,
R5 ae SFr
REOERZ 71.
whereas inputs 3 and 4 will multiply an input voltage by 10.
R5 tise
R3 or )- My cpa Sct
With capacitor C1 selected in the feedback loop, the
computing element is converted to an integrator and if
values are substituted in the equation for the integrator, it
can again be shown that inputs 1, 2 and 3, 4 give a gain of 1
and 10 respectively. The selection of C2 in the feedback loop
increases the gain of all inputs by a factor of 10. This is
usually referred to as a nose gain of 10. The symbols used to
denote adders and integrators with the relevant gain values
are shown in Fig. 2.2.
The “Initial Condition” resistors R6 and R7 are brought
into the circuit by means of switches RLA2 and S$1d. VR1 is
a 10kQ potentiometer, which provides the op-amp with
external offset nulling. This is connected across pins 1 and 5,
1074
| MPUTER P. J. KRONIS s.se.
PART 2
with the pot slider taken to the negative supply rail. The non-
inverting input of the op-amp is grounded via R8. The value
of this resistor should be chosen for good thermal drift
performance. The optimum resistance would be equal to the
parallel value of the input and feedback resistances. Since in
this case there are two values of input resistances, a
compromise solution is necessary.
The circuit of Fig. 2.1 represents just one computing
element and analogue computers may have many such
elements. The prototype has ten computing amplifiers which
is an adequate number for the solution of fairly complex
problems.
R?7 R6
2 > AAA
100%2 ] 100kN
Sta RS
]
RLB2 1 1Mn.
Sid
1 2 2
—_ c7 ah
= C1
86 >—{F-
R1 wr
ce “VV c2
a Sic c6 IR
aS a cease | Or +15V
co AA /A+ ' 2 ped: —C 83
mn 6
¢ RLA2 Ici > —
R3 2 é
«AA h-* ! cs VR31 <4
100k. Re
i 270kn 10K |
Ro
cs yom AL 1SV
100kN y
TOICK &IC13
PINS 3
Fig. 2.1. Circuit diagram showing one of the ten com-
puting elements of the Analogue Computer
Practical Electronics October 1978
Te Se UTI Ta
bey AVING formed a general picture of the workings of the
analogue computer, the complete circuit of a computing
element can now be described. This is shown in Fig. 2.1. The
basic circuits of input and feedback, components connected
around the op-amp can be readily recognised. The input
comprises four resistors, R, to R,, which are connected to
sockets in the patch panel and to the inverting input of the
op-amp, via switches RLA2, Sic, and S1b. The feedback
circuit consists of R5, C1 and C2, which can be selected by
means of switch S1a and sockets (C7, B6 and C6) on the
patch panel.
Consider switch S1a set so that R5 is selected in the
feedback loop. The computing element now becomes a
summer. By recalling the equation for the addition circuit
that was described last month and by substituting the values
for R5, R1, R2, R3, and R4 it can be seen that a voltage
applied at inputs 1 and 2 will be multiplied by unity,
R5 ae SFr
REOERZ 71.
whereas inputs 3 and 4 will multiply an input voltage by 10.
R5 tise
R3 or )- My cpa Sct
With capacitor C1 selected in the feedback loop, the
computing element is converted to an integrator and if
values are substituted in the equation for the integrator, it
can again be shown that inputs 1, 2 and 3, 4 give a gain of 1
and 10 respectively. The selection of C2 in the feedback loop
increases the gain of all inputs by a factor of 10. This is
usually referred to as a nose gain of 10. The symbols used to
denote adders and integrators with the relevant gain values
are shown in Fig. 2.2.
The “Initial Condition” resistors R6 and R7 are brought
into the circuit by means of switches RLA2 and S$1d. VR1 is
a 10kQ potentiometer, which provides the op-amp with
external offset nulling. This is connected across pins 1 and 5,
1074
| MPUTER P. J. KRONIS s.se.
PART 2
with the pot slider taken to the negative supply rail. The non-
inverting input of the op-amp is grounded via R8. The value
of this resistor should be chosen for good thermal drift
performance. The optimum resistance would be equal to the
parallel value of the input and feedback resistances. Since in
this case there are two values of input resistances, a
compromise solution is necessary.
The circuit of Fig. 2.1 represents just one computing
element and analogue computers may have many such
elements. The prototype has ten computing amplifiers which
is an adequate number for the solution of fairly complex
problems.
R?7 R6
2 > AAA
100%2 ] 100kN
Sta RS
]
RLB2 1 1Mn.
Sid
1 2 2
—_ c7 ah
= C1
86 >—{F-
R1 wr
ce “VV c2
a Sic c6 IR
aS a cease | Or +15V
co AA /A+ ' 2 ped: —C 83
mn 6
¢ RLA2 Ici > —
R3 2 é
«AA h-* ! cs VR31 <4
100k. Re
i 270kn 10K |
Ro
cs yom AL 1SV
100kN y
TOICK &IC13
PINS 3
Fig. 2.1. Circuit diagram showing one of the ten com-
puting elements of the Analogue Computer
Practical Electronics October 1978
erpameay bE SS}
‘_Oo—4 x1 —
O
o——4}x10 o——4X10 (>
X10
(on ° x10
Fig. 2.2. Symbols used to denote adders and
integrators
Mode Control is achieved by means of relay contacts
RLA2 and RLB2. Relays are necessary because all ten
amplifiers need to be controlled simultaneously. Table 1
shows the positions of relay and other switches for mode
control of summers and integrators.
SUMMER
INTEGRATOR
SWITCH
COMPUTE
HOLD
RESET
COMPUTE
HOLD
RESET
RLA2
RLB2
Sta
Sib
Sic
Std
1
1
2
1
OPEN
OPEN
CLOSED
2
2
1
2
1
CLOSED
CLOSED
OPEN
2
2
2
TABLE 1
Fig. 2.5 shows how the ten computing amplifiers are
arranged on a printed circuit board with the component
overlay shown in Fig. 2.7. At the extreme ends of the board
the two four-quadrant multiplier i.c.s are accommodated.
This main p.c.b. is connected to other points in the computer
by means of edge connectors. *
The Four-Quadrant Multipliers
So far it has been shown how to multiply a variable
voltage by a constant. This is easily done, using the
coefficient multiplier, in conjunction with the amplifier gain.
The formation of the product of two variables is much more
difficult to obtain. Of the many methods that have been
devised, most have involved the use of devices with certain
characteristics, e.g. a diode function generator can be set up
to provide a square law action, or a log-antilog action. Op-
amps are usually employed with these circuits.
For the sake of simplicity and compactness it was decided
to use two four-quadrant multiplier i.c.s in the prototype. As
their name implies these can multiply in four quadrants,
mA VR3
22k
= Gy. VV itm +15V
ti 22k
| &
"1 12 9
2 4
1Ci1
AD 533JD 1
°
Ly
Fig. 2.3. Circuit diagram of the Four Quadrant
Multiplier
Practical Electronics October 1978
b—(_ PATCH PANEL.
Ab AS
which means that either or both voltages can be positive or
negative. This dispenses with the need to have an absolute
value circuit preceding the multiplier, as is the case with
other methods.
The particular device chosen for the prototype was the
AD533JD integrated circuit (shown in Fig. 2.3). This is not
the cheapest four-quadrant multiplier on the market, but it
has the advantage of being simple to operate, with the
minimum of external components. The i.c. comprises a
transconductance multiplying element, a stable reference,
and an output operational amplifier on a single monolithic
silicon chip.
The AD533JD multiplies with a transfer function of x
The division by 10 should not worry the programmer but it
should always be borne in mind when solving a problem. The
op-amp output provides +10V at 5mA, and is fully protected
against short circuits to ground or either supply voltage. The
inputs are fully protected against overvoltage transients.
The Overload Warning Circuit
The operation of the overload warning circuit is very
simple. The output of every computing amplifier is sampled
and compared with a positive and a negative reference
voltage. If the amplifier output goes higher than the positive
reference voltage, an |.e.d. is switched on, to indicate that
the amplifier is saturating in the positive sense. Similarly, if
the amplifier output falls below the negative reference
voltage another I.e.d. is switched on to indicate saturation in
the negative sense. The prototype uses +11V as the
reference voltages. An overload warning circuit is shown in
Fig. 2.4. Only one pair of comparators and l|.e.d.s are shown
but ten pairs are necessary to serve the ten computing
. amplifiers. This circuit is arranged on a separate p.c.b. shown
in Fig. 2.6 with the component overlay shown in Fig. 2.8.
+15V
Vin FROM IC1 (PIN 6)
TO PIN 2 ICIS 2
Yen +15V
a 2 fy
D2
3 i‘ gas ee TIL 209 =
R10
aut =v TO PIN 2 IC16 oe
+15V
7
2
8 : 22kn 7a
IC14
vRi0 OP VR12 ot soma
22kn 3 TIL 209 =
~15V
-15V
Fig. 2.4. Circuit diagram of the Overload Warning
system required for each computing element
Resistor RQ and potentiometers VR11 and VR12 are
connected across the positive and negative supply rails to
form a potential divider that generates the positive and
negative reference voltages of +11V and —11V. These
voltages are applied to the inverting inputs of the twenty
comparators as shown. The output of each computing
amplifier is applied to the non-inverting inputs of the
corresponding pair of comparators. The comparators drive
the warning l.e.d.s, the brightness of which is set by preset
potentiometers. The 741 op-amp was also used here as a
comparator. Experience with the prototype has shown that
the 741 is capable of driving the |.e.d.s with reasonable
brightness without overheating.
1075
afavalalalalatalafaleataalalalal
arning p.c
ig P &
Practical Electronics October 1978
Fig. 2.8. Component layout for the Overload Warning Circuit
The Relay Mode Control P.C.B.
With ten amplifiers and two relay contacts per amplifier
there is a need for twenty relay contacts. Complete mode
control could be achieved with two ten-pole relays, one
operating the RLA and C switches and the other the RLB and
D switches. Ten-pole relays are difficult to find however and
the prototype uses four six-pole relays operating in pairs.
(The coil connections for the four relays are shown in Fig.
2.11.) This arrangement leaves four unused poles, which
may become useful if it is decided to extend the computer.
50
ath nN Ren
Fig. 2.11. Coil diagram for relays
The p.c.b. which accommodates the four relays is shown in
Fig. 2.9. Connections to and from this board are also made
via edge connectors.
Case Construction
The front panel requires a large surface area to
accommodate the patch panel, potentiometers, switches,
l.e.d.s etc. Because of this it will be difficult to obtain the
right shaped case off the shelf. The prototype case was
constructed from aluminium sheet. Two square panels form
the front and the back of the case and the sides, top and
bottom are cut and shaped as shown in Fig. 2.10, using the
100
a —— 2HOLES 035
| a | | 20 |
(oS ee a a ie
v —$ * © |
4 a
|
ey io eae a —¢- — 6 — 6 —-o
1 26 12 220
HOLES 08S Bet G0O0000 ABABA DADBARDRAARAAHRA Cesar | |
| DCOESG66O0 DO® DO GOSS SO00 ee a ”
HORIZONTAL MO VERTICAL | P0COOOGHHHHSOO666000000
PITCH 00 $969906999900666066006
99SGSS909909S96990O69000G6000-
$0600609600000000000
a 9999090600999 069006000
> et +++ oe eo + a di aii aie alin aie, ai an
n §600605000880008 888 a
Yi (een i ie a i ln Sn a lS 2 HOLES 06
20 SLOTS 105 ro Pe Oe tom
|
Sle es Nie <i, ili in a im as, ow aareee ei
ee BS |
20 HOLES @5 ee ee et ee ea ee
DIMENSIONS IN mm [| i
Fig. 2.10. Case cutting and drilling details
1078
\—3 HOLES 09.5
Practical Electronics October 1978
a ee ae aes
‘
same gauge aluminium sheet. A bench vice, folding bar, and
a sheet metal mallet are useful for this purpose. Fig. 2.10
shows the positions and dimensions of the holes required in
the front panel. A lot of patience is required for the process
of drilling, due to the large number of holes and the fact that
a badly positioned hole will be detrimental to the
appearance of the layout. This is particularly true in the case
of the patch panel holes. A pitch of 12mm in both directions
is enough to give a reasonable tolerance for positioning
errors and at the same time avoid excessive gaps between
the sockets. For the larger holes the use of sheet metal
punches is recommended. Having drilled or punched all the
holes, the front panel should then be labelled using dry letter
transfers and sprayed with a clear lacquer fixative. The
suggested labelling is shown in the photograph of the front
panel.
The Patch Panel
The patch panel is constructed using 3:2mm sockets
arranged in a matrix and packed together as closely as
possible. There are 148 of these sockets and because
identifying each one is difficult a colour coding system was
used. Fig. 2.12 shows the arrangement of the sockets for
one amplifier, one coefficient multiplier and one four-
quadrant multiplier.
The pattern for the amplifier and coefficient multiplier
shown, in Fig. 2.12 is repeated ten times for the ten
computing elements. The eight coefficient multipliers use 16
sockets on the top row. Two of the remaining four sockets
are connected to the two panel meters and the other two are
connected to batteries to provide reference voltages. Both
positive and negative reference voltages will be needed for
the solution of certain problems.
e)
je)
GREEN
Ke) RED
WHITE
Eo
YELLOW
BLACK
w<~ £720
yO
pe)
'Y
gO G
et
Fig. 2.12. Patch panel layout for one computing ele-
ment (rows B and C) and one Four Quadrant Multiplier
(row A)
[om Ke)
Es) Es)
0 fo
Es}
[om Ke)
Practical Electronics October 1978
For the four-quadrant multipliers four sockets are needed
per multiplier and these are positioned on the extreme left
and right of the patch panel.
Resistors :
-R1,R2,R5 1MQ 4W 2% metal oxide (30 off)
R3, R4 100kQ {W 2% metal oxide (20 off)
R8 270kQ 4W 5% carbon (10 off)
oo Res 100kQ 4W 5% carbon | Pe
R10. 7-5kQ $W 5% carbon (2 off)
—-VR1,VR8—4-7kQ (2 off)
~VR2-VR7 —s-.22kQ. (6 off)
- VR9-VR30_ = 22kS7 (22 off)
~ VR31-VR40_ 10k 0-5W Lin (10 off)
All horizontal min. presets except where stated
‘Cl ~ 1pF 160V (10 off)
C2 O-1pF 160V (10 off)
1C1-IC30 741 op amp (30 off)
1C31-1C32 AD533JD (2 off)
D1-D20 TIL 209 (20 off)
Miscellaneous
4 off 6 way changeover relays oo
4 off mounting sockets for relays o
holders for i.c.s (if req.)
CONSTRUCTOR'’S NOTE: The AD533JD
Four Quadrant Multiplier is available from
Analog Devices Ltd., Central Avenue, cant
Molesey, Surrey.
Stage by Stage Construction
The computer has been designed so that it can be built in
stages. At this point in the construction, ie, with the
aluminium case and the p.c.b.s constructed and drilled, the
constructor has to take a decision, as to whether he wants to
opt for a stage construction. His choice can be very flexible.
For example, one may decide that initially, all ten computing
amplifiers are not absolutely necessary for the solution of
simple problems with which the inexperienced programmer
will be involved. Four amplifiers are enough to carry out
fairly interesting experiments. Later, when more experience
is gained, more computing amplifiers can be added as
necessary. The same applies to the coefficient multipliers
and the panel meters.
It should be mentioned that if four 6-pole relays are used
for the mode control, as is the case with the prototype, at
least two of these will be necessary even if only one or two
amplifiers are used initially. Two 6-pole relays can provide
mode control for six amplifiers.
Another area in which stage by stage construction can be
applied, concerns the overload warning circuit. Here, the
comparators and the l.e.d.s can be added following the
addition of more amplifiers. Alternatively it may be decided
to leave the warning circuit out altogether initially. This will
make life difficult for the programmer, but it will not affect
the operation of the computer.
NEXT MONTH: WIRING AND TESTING
1079
Australia 85c South Africa 80c New Zealand 85c Malaysia $2.25
PRACTICAL
ELECTRONICS
NOVEMBER 1978
MOON
LANDIN
wap aim es Rg bp Becrad Gas
, ae 7 <i Z,
i hae ae. eae be pant ee Sr ny
A ee ag , eorer * eee,
Poa OT Fe, _ ERE
PRACTICAL
ELECTRONICS
VOLUME 14 No.15 NOVEMBER 1978
CONSTRUCTIONAL PROJECTS
MOON LANDING GAME by A. Russell
Sixty seconds to avoid a negative altitude situation 1138
P.E.V.D.U. SYSTEM—2 byA.A. Berk, B.Sc., Ph.D.
Construction and setting up 1146
PROXIMITY SWITCH byA.K. Langford
A variable sensitivity touch and proximity switch 1160
ANALOGUE COMPUTER—3 by P. J. Kronis, B.Sc.
Wiring, testing and programming 1165
WIDE RANGE C/R BRIDGE by W. English
1000uF/10MQ 1180
GENERAL FEATURES
POWER FETs by David Shortland
The latest developments in field effect technology 1154
INGENUITY UNLIMITED
741 Supply—Soldering Iron Simmer Control—Auto-Tune Generator—Touch Tuner 1173
NEWS AND COMMENT
EDITORIAL 1137
EXCLUSIVE OFFER
Two irons at special prices 1143
SPACEWATCH by Frank W. Hyde
German Satellite, U.S. Satellite, TDRSS, Pioneer Venus 2, TRS, Intelsat Terminal, Spacelab 2 1144
NEWS BRIEFS
Starlight Vision 1153
Bubbling with Bits—Blinking Good—On the Levell 1170
Data Encryption Unit 1182
MARKET PLACE
New products
SEMICONDUCTOR UPDATE bDyAf. W. Coles
A look at some recently released devices 1179
STATESIDE SCENE by Dave Coutts
New York News 1182
SPECIAL SUBSCRIPTION OFFER ‘1185
INDUSTRY NOTEBOOK by Nexus
What's happening inside industry 1186
PATENTS REVIEW
Thought provoking ideas on file at the British Patents Office 1187
READOUT
A selection of readers’ letters 1188
POINTS ARISING 1188
SPECIAL 8-PAGE SUPPLEMENT
CAR DEVICES Auto-Light—Vari-Wipe—Assisted Ignition System—tIntruder Alarm—Battery State
Indicator—Polarity Inverter between 1164 and 1165
Our December issue will be on sale Friday, 10 November 1978
(for details of contents see page 1145)
© IPC Magazines Limited 1978. Copyright in all drawings, photographs and articles published in PRACTICAL
ELECTRONICS is fully protected, and reproduction or imitations in whole or part are expressly forbidden. All reasonable
precautions are taken by PRACTICAL ELECTRONICS to ensure that the advice and data given to readers are reliable.
We cannot, however, guarantee it, and we cannot accept legal responsibility for it. Prices quoted are those current
as we go to press.
Practical Electronics | November 1978 1121
FTER the p.c.b.s have been assembled and checked, all
the components should be fitted into the case with the
overload warning p.c.b. mounted on the base of the box
using 6BA screws. With all the components mounted in the
case the coefficient multipliers and the two panel meters
should be wired first, following the wiring diagram shown in
Fig. 3.1. Resistors R6, R7 and the links shown in Fig. 3.2
should be wired to each of the ten computing amplifiers. The
wiring to the relay board and main p.c.b. is via eight edge
connectors and to ease the problem of wiring these
connectors a wiring schedule is given in Fig. 3.3. The
numbering and layout arrangement of the patch panel and
switches is given in Fig. 3.4.
The main p.c.b. is mounted above the patch panel and the
relay board above the offset null potentiometers as shown in
the photograph. After the computer wiring has been
completed and checked a +15V power supply should be
connected to the unit and the following test procedures
followed.
BATTERY
REFERENCE
VOLTAGES
Ir
\
Fig. 3.1. Wiring diagram for the coefficient multipliers
and panel meters
Practical Electronics © November 1978
PUTER P. J. KRONIS .s-.
* Wiring
* Testing
* Programming
THE OFFSET NULL TRIM PROCEDURE
Set all the amplifiers to “add” by pushing all the slide
switches down. Put the computer into the ‘‘compute”’ mo