- r ;

77

September 1981
U.K.60p

Electronic

barometer

the new Quad

chattering

chips?

up-to-date electronics for lab and leisure

selektor

DFM+DVM

This article combines a digital frequency counter with the 254 digit DVM
published in the February 1981 edition of Elektor. A reasonable degree
of accuracy is provided by a crystal timebase.

revolution counter

Two opto detectors and an up/down counter with a digital display form
the basis of this revolution counter. It can be used with any device that
uses turning wheels.

digital barometer

This article features ICs with a difference in a circuit that incorporates
both pressure and temperature transducers and indicates the levels
measured on a digital display.

dB converter

The converter described in this article enables the Elektor sweep generator
to display frequency response curves on an oscilloscope in levels of dB.

TV games extended

This extension board for the TV games computer almost triples the avail-
able memory space. As an added bonus, two sound effects generators are
also included on the board.

EDITOR:

UK EDITORIAL STAFF

T. Day E. Rogans
P. Williams

disco lights controller

The controller featured in this article will perform equally well
the more expensive systems available.

TECHNICAL EDITORIAL STAFF

J. Barendrecht
G.H.K. Dam

E. Krempelsauer
G. Nachbar
A. Nachtmann

K. S.M. Walraven

QUAD ESL 63

The 3-micron polyester diaphragm uses electronic conduction obtained by
doping it with one donor atom to about ten million non-conductive
atoms ... a new language when related to loudspeakers! The new QUAD
electrostatic ... in depth.

analogue LED display .

This circuit can be used
meters, thermometers and
analogue indication.

voltmeters, revolution counters, humidity
fact virtually any meter that requires an

volt/ammeter for power supplies ip. Gabier)

The circuit published here enables a DVM with an automatic range switch
to measure both current and voltage on stabilised power supplies.

chattering chips

With talking computers, washing machines and telephone exchanges just
over the horizon, it is worth taking a look at the principles involved.

transistor ignition update

soldering aluminium

missing link

market

advertisers index

EIEKTOR ROOK SERVICE

JUNIOR COMPUTER BOOK 1 - for anyone wishing to become familiar with (micro)computers, this book
gives the opportunity to build and program a personal computer at a very reasonable cost.

Price - UK £4.25 Overseas £4.45

JUNIOR COMPUTER BOOK 2 - follows in a logical continuation of Bookl, and contains a detailed app-
raisal of the software. Three major programming tools, the monitor, an assembler and an editor, are dis-
cussed together with practical proposals for input output and peripherals.

p r j ce _UK £4.75 Overseas £4.95

300 CIRCUITS for the home constructor - 300 projects ranging from the basic to the very sophisticated.

Price - UK £3.50 Overseas . £3.70

DIGIBOOK - provides a simple step-by-step introduction to the basic theory and application of digital
electronics and gives clear explanations of the fundamentals of digital circuitry, backed up by experiments
designed to reinforce this newly acquired knowledge. Supplied with an experimenter's PCB.

Price - UK £5.00 Overseas £5.20

FORMANT - complete constructional details of the Elektor Formant Synthesiser - comes with a FREE
cassette of sounds that the Formant is capable of producing together with advice on how to achieve them.

Price - UK £4.50 Overseas £4.70

SC/MPUTER (1) - describes how to build and operate your own microprocessor system - the first book of
a series - further books will show how the system may be extended to meet various requirements.

Price - UK £3.70 Overseas £3.90

SC/MPUTER (2) - the second book in the series. An updated version of the monitor program (Elbug II)
is introduced together with a number of expansion possibilities. By adding the Elekterminal to the
system described in Book 1 the microcomputer becomes even more versatile.

Price details on order card.

BOOK 75 - a selection of some of the most interesting and popular construction projects that were
originally published in Elektor issues 1 to 8.

Price - UK £3.50 Overseas £3.70

When ordering please use the Elektor Readers' Order Card in this issue (the above prices include p. & p.)

Many Elektor circuits are accompanied by printed circuit
designs. Some of these designs, but not all, are also available
as ready-etched and pre-drilled boards, which can be ordered
from any of our offices. A complete list of the available
boards is published under the heading 'EPS print service' in
every issue. Delivery time is approximately three weeks.

It should be noted however that only boards which have at
some time been published in the EPS list are available; the
fact that a design for a board is published in a particular
article does not necessarily imply that it can be supplied by
Elektor.

| Technical qucric/ I

Please enclose a stamped, self-addressed envelope; readers
outside UK please enclose an IRC instead of stamps.

Letters should be addressed to the department concerned -
TQE (Technical Queries). Although we feel that this is an
essential service to readers, we regret that certain restrictions
are necessary:

1. Questions that are not related to articles published in
Elektor cannot be answered.

2. Questions concerning the connection of Elektor designs

to other units (e.g. existing equipment) cannot normally
be answered, owing to a lack of practical experience
with those other units. An answer can only be based
on a comparison of our design specifications with those
of the other equipment.

3. Questions about suppliers for components are usually
answered on the basis of advertisements, and readers can
usually check these themselves.

4. As far as possible, answers will be on standard reply
forms.

We trust that our readers will understand the reasons for
these restrictions. On the one hand we feel that all technical
queries should be answered as quickly and completely as
possible; on the other hand this must not lead to overloading
of our technical staff as this could lead to blown fuses and
reduced quality in future issues.

Junior Computer Book 2

This, the second in the Junior Computer Book series, follows in a logical continuation of
book 1 and is an indispensible aid to users of the Junior Computer.

The book contains a detailed appraisal of the software. Three major programming tools,
the monitor, an assembler and an editor are discussed in detail together with practical
proposals for input output and peripherals. The complete source listing with relevant
comments for all described programs, and the entire contents of the EPROM, are in-
cluded.

THE JUNIOR COMPUTER BOOK 2 IS AVAILABLE NOW from
Elektor Publishers, 10 Longport, Canterbury, Kent, CT1 1PE.

An order card is enclosed in this issue for your convenience.

Price U.K. £ 4.75, overseas £ 4.95 postage and packing inclusive.

Book 2 in The
Junior Computer Series

book 2

Price — UK £4.00 inc. postage and packing

Price - Overseas surface mail £4.20 inc. postage and packing

When ordering please use the Order Card in this issue

ORDER VOUR COW NOW

INTRODUCING- from Ctektor
sc/mputer ®

build your own
microcomputer system

This is the second in a series of books describing the construction and operation of a personal
computer system that can be as large or as small as the constructor desires.

An updated version of the monitor program (Elbug II) is introduced together with a number of
expansion possibilities. By adding the Elekterminal (ASCII keyboard and video interface) to the
system described in Book One the microcomputer becomes even more versatile.

Book Two continues with a description of a BASIC interpreter (NIBL-E) and gives complete
details of a basic microcomputer card which can be either incorporated into the system described
in Book One or used separately.

The final chapter of the book gives all the information required to enable the Elektor computer
to be programmed in the BASIC language.

UK 12 - elektor September 1981

HiFi in the car - possible or not?

How is it, HiFi enthusiasts ask them-
selves, that HiFi can function in a car
where the important HiFi preconditions
have not been fulfilled. This is an im-
portant question and one worthy of an
answer.

HiFi equipment can work if the car is
big enough, if the equipment corre-
sponds to HiFi norms and if the back-
ground noise is non-existent. However,
as most cars are not as big as a living
room and do not operate noiselessly,
the answer to the question must be,
HiFi cannot function in a car. Why is it
vhen that so much money is spent on
electronics and amplifiers which dis-
tinguish between tuners, cassette decks,
equalisers and amplifiers for high
quality car music equipment and their
"normal" car radio brothers? Is this not
perhaps merely half the answer to the
question as to whether HiFi is possible
in the car? It then seems logical that the
other half of the answer is yes and that
HiFi is possible in the car. There are two
sides to the story. In fact it is not poss-
ible but nevertheless an ever-increasing
amount of car HiFi equipment is being
bought!

HiFi Criteria

In fact the solution to the problem lies
in the way in which one looks at the
subject. As far as the equipment is con-
cerned, there is no problem in repro-
ducing music of HiFi quality in the car,
i.e. the technical prerequisites have been
met. However, the interior of a car is
blatantly unsuited as a listening area for
HiFi reproduction, because it has been
constructed in an acoustically un-
favourable way and the sound is often
drowned out to an unacceptable degree
by engine and background noise. To
express this in a technical way: noises
which are not part of the music must
be below the musically desired signal so
that they do not interfere with repro-
duction. In a quiet living room this
does not constitute a problem because
the background noise is at a level of
between 20 and 30 dB and the volume
range of HiFi reproduction being above
this, at around 50 dB (the range from
the lowest to the loudest sound), means
that there is a reproduction level of at
most around 80 dB. However, even this
constitutes a volume level which might
bring neighbours to bang on the wall.

In the interior of a car the situation is
much more unfavourable. There, the un-
desired noises, caused by the moving
car, are at a level of 75 to 80 dB. If one
wanted to listen to music in accordance
with HiFi criteria, this would mean an

9

additional 50 dB and at times in the
car this would mean an almost un-
bearable volume level of over 120 dB,
somewhere between a compressed air
hammer and a jet engine. Of course, no
human being can endure that and for
that reason a HiFi compromise must be
reached between theory and practice. In
addition, every car radio listener is
familiar with the relationship between
disturbing background music and the
musically desired signal. At increased
speed one has to turn the radio up in
order to maintain the same level of
speech and music which can be achieved
at lower speeds or at a standstill.

HiFi debit and HiFi credit

The compromise had led to the present
possible optimum in high-quality sound
reproduction in the car. Since for tech-
nical reasons the car makes it impossible
to achieve the desired HiFi credit, car
radio manufacturers are concentrating
all their efforts on providing technical
equipment which corresponds to the
quality standards set by HiFi. The re-
sulting plus on the "credit side' of the
HiFi balance sheet had led to an extra-
ordinary improvement in tone quality in
contrast with the "normal" car radio
mentioned in the beginning. This is the
reason for the growing popularity of
high-quality car music equipment con-
forming to HiFi standards.

All the same, considerable attention
must be paid, not only to the technical
specifications of receiving sections,
cassette recorders, tone control sections
and power amplifiers, but also and in
particular to loudspeakers. What is the
point of having the highest quality
frequency response, if the loudspeakers
are not able to reproduce this properly?

The right choice and correct installation
of loudspeakers is very important owing
to the increased difficulties in the in-
terior of a car caused by the partly
reflecting and partly absorbing surfaces.
Unfortunately, here too there are cer-
tain limits. High volume loudspeakers
cannot be installed owing to a shortage
of space. Therefore special loudspeakers
have to be used which are mounted in
or under the dashboard, in the doors or
on or in the ledge behind the back seat
to produce a similar sound quality.
This not only presupposes expert instal-
lation, it also means that the car HiFi
enthusiast has to dig deeper into his
pocket. Good loudspeakers are not
cheap and one should really only buy
first-rate ones.

The question as to which HiFi music •
sources has still to be answered. On the
one hand there are the VHF pro-
grammes of the broadcasting stations,
which in spite of interference from
background, engine and reflection noise,
often constitute a good source.

On the other hand the music lover can
always fall back on cassettes, whether
prerecorded or recorded by the indi-
vidual on his own HiFi equipment. If
the relevant criteria are observed and
the available technology for car HiFi
equipment is used, then this surely is
the best way of achieving high-quality
music reproduction in the car. In a few (
years, when the new digital mini-discs
can be played in the car music equip- I
ment as well, sound reproduction will
perhaps be even nearer perfection. <

691 S '

selektor

iber 1981 - '

Choosing the 'right' cassette

Fe-Cr-Me: It's the coating that
does it

Thirty years ago, it was a notable
achievement in tape recordings to have
attained something approaching the
of AM radio — 5000 Hz or so. Today,
by contrast, a sound quality is de-
manded from tapes which regards the
hi-fi standard of 12,500 Hz as the very
lowest limit.

In the past 15 years or so, open-reel
tapes have been largely superseded — at
least as far as domestic use is concerned
— by the handy compact cassette, which
now captures about 90 percent of the
market. However, the cassettes made at
the beginning of the 'tape revolution'
are just as impossible to compare with
those which music lovers are able to
purchase today. Their dimensions re-
main the same — having been standard-
ised at a very early stage, but their per-
formance has improved tremendously.
On the other hand, the purchasor is now
confronted with such a wide variety of
types and makes that it is wise to get to
know a few things about them to be
able to make the best choice for any
particular requirement.

Here are a few useful tips:

The most important thing about any
cassette is the magnetic tape inside it.
The original ferrous oxide tape (I) was
first joined by the chrome-dioxide tape

(II) , then came the multi-coated tape

(III) , in which a ferrous oxide coating
is combined with a chrome-dioxide
coating, followed by the super chrome-
dioxide tape (II), representing a further
improvement in quality - and the latest
addition is the so-called metal tape (IV).
The numbers shown in brackets and re-
produced on the cassettes themselves
are not indicative of any grade of
quality, but are part of an inter-
national classification system denoting
the bias adjustment recommended for
the various tape categories as elaborated
by the I EC (International Electronics
Committee). Serving on this committee
are cassette manufacturers from all the
more important producing countries,
whose aim is to solve the problems of
international interchangeability by
means of joint consultations. Now, how
do these various tape classes differ?
Ferrous oxide tapes (Fe)

These cassettes can be used with all
cassette players/recorders that comply
with the DIN Norm.

Chrome-dioxide tapes (Cr)

These are already in the 'hi-fi league',
being distinguished by an expanded re-
sponse and crystal-clear treble repro-
duction. To exploit their performance
to the full, they should be used on
cassette recorders with a chrome-
dioxide selector switch (marked Cr or
CrOj ) or a recorder which switches over
automatically.

Multi-coated tapes IFeCr)

These are cassettes of the Hi-Fi class,
which combine certain advantages of
ferrous oxide with those of chrome-
dioxide on one tape. They are designed
for universal use on all recorders, and
their chief characteristic is a deliberately
enhanced brilliance. Optimum perform-
ance is obtained on recorders with a
special FeCr switch; for Hi-Fi require-
ments with an extremely wide range of
applications.

Super chrome-dioxide tapes (Cr)

With these cassettes, a substantial in-
crease in dynamic range can be obtained
in the high register — up to 6 dB at
frequencies between 10,000 and
20,000 Hz, a specially important factor
in Hi-Fi. At the same time, the ex-
tremely low tape hiss, characteristic of
chrome-dioxide, is retained.

Metal tapes (Me)

This coating pigment stipulates the use
of special erasing and sound heads on
the recorders. The tapes can then be
even better modulated — by 3-4 dB
at 315 Hz and by some 10 dB at
20,000 Hz in comparison with the DIN
reference chrome-dioxide tape. These
cassettes can be played on conventional
recorders, but they cannot be erased
and re-recorded.

Faced with this variety of tapes, the
logical conclusion would seem to be
to go straight for the latest type of
cassette: metal tape. In reality, it is not
all that simple. First of all, a new re-
corder is required, because of the need
for new heads - and the Hi-Fi enthusiast
will not be satisfied with the cheapest.
Secondly, metal cassettes are a good
deal more expensive than oxide
cassettes, and thirdly, although it is
quite certain that some technical figures
are really better with metal tapes than,
for instance, with chrome-dioxide tapes,
some of the improvements are only
perceptible to instruments, not to the
human ear, which, after all, has the last
say in the matter. The buyer will con-
tinue here to use the tried and true
yardstick of price-to-performance
relationship, which certainly does not
disqualify oxide cassettes. This will also
probably be the reason why metal tape
cassettes have not managed to seize any
sizeable share of the UK market as yet,
although they have already been avail-
able for two years.

Finally, another point: the striking
differences in the prices of cassettes
for the user. There are, for instance.

cassettes to be had for very low prices
indeed. These cassettes have one great
advantage: they are cheap, very cheap!
But that's about all. Whether they are
good or bad is a matter of pure luck. It
can seldom be seen what kind of tape is
in the package. Sometimes the buyer
will know, sometimes he will have no
idea. When he comes to play it, he will
often get quite a surprise. Sometimes
it's a normal tape, sometimes a com-
puter tape; sometimes it's an unsuitable
studio tape cobbled together out of
waste material. It may tie itself in knots
the first time the reel revolves, or it
may wear the head out in no time in
the absence of any surface smoothness.
Any complaints? Forget it! You can
turn the carton round all ways - you
won't find the name of any manufac-

So anyone who places importance on
lastingly clean and good reproduction —
full orchestral classical music included
— would do well to bear this in mind:
choose the right one first time - a
quality cassette with a reputable trade
mark. It may cost a little more, but
it will save a great deal of disappoint-

Editorial Office,
radio tv press,

Postfach/P.O.B. 191740,

D-1 000 Berlin 19,

W. Germany.

692 S

1991 . . . and Londoners continue the
fight against the Martian invaders.
That could have been the explanation
for this photograph, but the engineers
look altogether too happy about the
prospect. In fact, it shows one of the
four microwave dish aerials erected on
the roof of the Abbey Life Building in
St Paul's churchyard for transmitting
television pictures of the Royal Wedding
ceremony to 500 million viewers
throughout the world.

British Telecom

(693 S) I

9-02 - ele

DFM + DVM

As far as test equipment and measure-
ment is concerned, the average elec-
tronics constructor has to make do with
a collection of bits and pieces that
happen to come to hand, usually second
or third-hand at that. This is, however,
exactly the type of equipment that
most enthusiasts are prepared to put up
with, for after all. their projects don't
usually have to meet very high require-
ments anyway. The line has to be drawn
somewhere for home building to be
viable (there is no tax relief on transis-
tors and ICs).

The two items of test equipment

DFM + IWM

i

digital measurement of frequency and voltage

Any additions to the range of test equipment available to the home
constructor are always welcome. This article combines an economical
digital frequency counter with the 2 'A digit DVM published in the
February 1981 edition of Elektor. Absolute accuracy was considered of
secondary importance to economy and the use of readily available
components, although the circuit does use a crystal controlled
timebase.

described in this article constitute a step
in the right direction for the home
constructor. The 214 digit DVM published
in the February issue provides a
reasonable degree of accuracy using
fairly common components. The next
stage is to combine the DVM with a
digital frequency meter that has an
acceptable 2 MHz range, a 314 digit
display and a 'real' crystal timebase.

The DFM-plus-D VM consists of two
printed circuit boards; the main board
and the display. The original display
board of the DVM has been retained
while the remainder of the DVM circuit
is combined with the frequency counter
on a new main board. This results in
two items of test equipment at very
reasonable cost.

The circuit diagram
The complete circuit is drawn in figure
1. The area inside the dotted line
represents the display section as used in
the DVM. The only changes required are
the addition of the extra display and its
driver transistor Dp4 and T4, and an
increase in the value of R8.

The timebase circuit can be seen at the
bottom of the diagram. The crystal
controlled oscillator generates a fre-
quency of 3.2768 MHz. This is then
divided by a factor of 2 W by IC3 and •
then by another 200 by IC4 and IC5a •
together. The Q4a output of IC5a then
produces the clock frequency of pre-
cisely 1 Hz. At each negative going
clock pulse the differentiators, N1 and
N2, originate a latch pulse and then a
reset pulse which are passed on to IC1.
The width of each of these pulses is
about 1 ps which is so short that the
error it causes in the reading is negligible.
Once every second, therefore, the
contents of the counter in IC1 are
passed on to the display buffer (latch)
and immediately afterwards the counter
is reset to start the count once again. It
should be noted that display of Dpi is
suppressed whenever a zero exists in this
position.

The input signal for the DFM enters the
input amplifier via the decoupling
capacitor C9 while the combination of
R15, D2 and D3 serve to protect the
gate of the FET T8 against any excessive
input voltage levels. The amplifier stages
built around T8 and T9 each amplify
the signal by a factor of ten, so that the
total amplification of the input ampli-
fier is a factor of 100. The amplified
signal is then fed to the Schmitt trigger
N4. The DC voltage at the input of this
gate can be preset by means of PI to a
level exactly half-way between the
positive and negative trigger thresholds
of the gate. The DFM input will then be
set at its maximum sensitivity.

Three decade counters (divide-by-tens)
follow the Schmitt triggers to provide a
total of four measurement ranges. The
outputs of N4 and the dividers are fed
to the range switch S2a which selects
one and passes the signal, via S3a (selects
either DFM or DVM), to the clock input -
of IC1.

The S2b wafer of the range switch
ensures that the decimal point is

Tech

for the DFM

+ DVM

Number of digits

Input impedance

Accuracy

Max. input voltage

DFM

314

10 Hz... 2 MHz
<15mV Rm> (f < 1 MHz)
< 30 mV Rms (f > 1 MHz)
1 MO

± 0.05% (full scale)
100V Rms

DVM

2%

. . . 200 V —

positioned in the correct place. The
decimal point also doubles as an over-
range indicator. As in the DVM this uses
the carry-out output of IC1 which will
produce pulses at the rate of one per
second when the maximum count (1999)
has been exceeded. Since IC1 operates
at 5 V, the pulses have to be 'translated'
into 12 V and then rectified to provide
a logic 1 level for switching N3. A
square wave produced by another output

of IC5a in the timebase then becomes
the controlling signal (at the other input
of N3) that, via T7, causes the decimal
point to flash. It will continue to do so
until either switch S2 is correctly
positioned or the offending input is
removed.

A full description of the DVM circuit
was published in the February issue.
The circuit remains unchanged with the
exception of a few modified component

values to accommodate the new time-
base. As previously mentioned, switch
S3 selects either DVM or DFM. In the
DVM position the display Dp4 is
switched off as due to the accuracy of
the a/d converter, only three figures are
needed.

The power supply has been kept simple
with two regulators, IC2 and IC8, to
produce the 5 and 12 volts required. It
will be as well to provide IC2 with a

substantial heat sink if it is to be kept in
one piece.

It should be noted that S2 acts as a
range switch for both the voltmeter and
the frequency counter and will there-
fore require two scale legends. For
instance, switch position (a) will have
10 mV . . . 2 V and 10 Hz . . . 2 kHz as
its two ranges. One suggestion would be
to place a range vertically either side of
the control knob. A further set of
contacts on S3 could then be used to
light a LED above each set of ranges to
show which was selected. However, we
are sure that readers will have their own
ideas.

Construction

The printed circuit board layout and the
component overlay for the main board
and the display board are shown in
figures 2 and 3.

The transformer will have to be a little
larger than that required by the original
DVM circuit due to the 12 V required.
The power supply section of the display
board can be assembled in a slightly
unorthodox manner to improve the
space situation behind a front panel.
Figure 4 illustrates the component
positions. Capacitors C2a and C2b may
be replaced by a single 1 000 p/25 V
type and be fitted on the underside of
the board as shown. To improve the
clearance problem even further, the
regulator IC2 and its heat sink may also
be fitted under the board but take care
not to be confused with the wiring
connections. If you're still stuck for
space, the bridge rectifier can also be
mounted on the track side of the board.
A handy little trick for raising the
displays is to place two 1C sockets under
each display as shown. After assembly
of the two boards is complete, the wire
connections between the boards can be

Parts list for the display board:

Cl " 10 p/10 V tantalum
C2a,C2b - 470 p/25 V

Dpi . . . Dpi = 7760
(common cathode)
B = B40C1000
T1 . . . T4 = BC 141
IC1 = 74C928

fitted, DS, R, LE, CO, DP, +5 V, ++ and
1. Since it is possible that many readers
have built the DVM from the original
article, the changes in wiring to the
present design could become somewhat
chaotic. For this reason the following
point to point wiring guide has been
included.

S2a: a . . . d to S2a: a . . . d main
board

S2a wiper toS3a:b

S2b: a . . . d to decimal points on
display board

S2b wiper to display side of R8
S2c: a . . . c to R29 . . . 33
S2c wiper to S2c main board

S3a:a to S3a main board

S3a wiper to CL main board

S3b:a to 1 display board

S3b wiper to pin 11 IC1 display

board

The resistors of the DVM input voltage
divider (R29...R33) are soldered
directly to the switch terminals. The
+ point marked on the main board next
to S2c is first taken to R29 on S2c and
from there to the DVM + input socket

on the front panel. The DVM - input
socket is connected to the a terminal
of S2c. The DFM input and O can now
be connected to DFM and 1 respectively
on the main board. Readers may prefer
to use a BNC socket for the DFM input.
Finally, the transformer connections
can be made.

Calibration

All that remains (after checking all the
wiring again) is the calibration pro-
cedure. The timebase is the first section
to be adjusted. If an accurate 6 digit
frequency counter is available simply
adjust C8 so that the frequency at test
point TP 204,800 Hz. Otherwise, it is
sufficiently accurate to set C8 roughly
to the central position.

A frequency of about 1 kHz is then fed
to the DFM input. With S2 switched to
a and PI left in the mid position, the
amplitude of the input signal is increased
until the display indicates the input
frequency. The signal amplitude is then
decreased until the display just stops

reading. PI is now readjusted until the
reading returns. This procedure is
repeated until the input level can no
longer be reduced without losing the
display altogether (that is, adjustment
of PI is unable to return it). Care is
required during this stage of the cali-
bration since it sets the maximum
sensitivity of the input amplifier.

To calibrate the DVM section of the
circuit the DVM inputs must first be
shorted together. Switch S3 must be
set to the lowest range (position a) and
P3 is turned fully clockwise. Now 'back
off' P3 until • 00 appears on the display.
If a reference voltage is not available,
connect a DC voltage of about 1 V
(a battery!) and adjust P3 while
comparing the reading with an accurate

It is as well to remember that all future
readings of the equipment in use will
depend on the accuracy of the initial
calibration.

Blektor September 1981 — 9-07

revolution

counter

counts both up and down

An up/down counter and two opto detectors with an LED display
form a relatively easy to build and highly accurate counter. Since
the cost of components is kept to an absolute minimum, the device
has infinite possibilities for hobbyists ranging from amateur sound
technicians to photographers. Recently, a prototype model was
exhibited at an electronics fair in Germany and was found to be
surprisingly popular.

Judging by people's reactions at the
electronics fair, this revolution counter
can serve a variety of purposes. One
favourite 'application' involves trying
to "trick" the device into giving a wrong
indication by simply turning it back-
wards and forwards for a time. Readers
are assured, however, that that gets to
be pretty frustrating pastime after a
while, for provided the counter is con-
structed in the proper manner, it is
very difficult to 'lead it up the garden
path'. However slowly or quickly you
turn it backwards and forwards, it
will imperturbably give the correct
indication.

Thus, any device that is based on
turning wheels and that can be con-
nected by mechanical means to the
counter can be checked with it.

The up/down counter

What does the rev counter look like? To
start with, the design is surprisingly
straightforward. This should put pessi-
mistic readers' minds at rest who
always fear the worst and who almost
always underestimate their own capa-
bilities.

The simplified block diagram in figure 1
shows that there is really nothing to
worry about. All it consists of are two
opto detectors, a simple logic circuit, an
(integrated) up/down counter and an
LED display.

Photo 1 illustrates the small vaned or
slotted wheel that is turned by the
device of which the revolutions are
to be counted. The vanes of the wheel
move into the slots of the opto
detectors and so periodically interrupt
the light beam between the LEDs
'and the phototransistor. One opto
detector is already enough to give a
correct pulse count. If two are used,
the direction can be indicated ("up"
or "down"). In other words the counter
counts either up or down, depending
on the direction in which the counter
wheel is turning. The device counts
the number of transitions from
light-to-dark and vice versa. After one
revolution the display will indicate
twice the number of slots in the

Figure 1. The block diagram of the revolution counter. The direction of rotation and number
of turns are detected by two opto detectors. The information is then decoded and passed on to

The circuit diagram
The full circuit diagram is shown in
figure 2. As you can see very few com-
ponents are required. This is partly due
to the fact that a lot of what is needed
is already integrated in the MK50398N
1C. This 1C has been used in Elektor
before in the 'mini counter' and the
'% GHz counter' (June '78) and contains
an up/down counter, full display
control and a preset facility. The latter
is achieved by connecting diodes
D1 . . . D24 with the aid of thumb-
wheel switches (or wire links) to points
DTI ... DT6. A certain number can
then be programmed which will appear
on the six digit LED display after the
'load' switch S2 is pressed.

The control information for the circuit
is derived by the opto detectors IC6 and
IC7 and the decode logic following
them. With the circuit illustrated in
figure 3, this takes care of the
'count and 'up/down' signals. Figure 4
shows how the two sensors have to be
mounted. They are moved with respect
to each other until there is a phase angle
' of 90° between the pulse signals pro-
duced. This means that when opto
detector IC6 is completely covered. IC7
will be in a transition period. The
detectors register the light-to-dark tran-
sitions, which brings us to the pulse
flow chart in figure 5.

Let's look at figure 5a first. The signals
which are 90° phase-shifted are shown
at the top of the diagram. The up/down
information is taken from these with
the aid of N12. If the output states of
N8 and N9 are equal, the output of N12
will be high, if they aren't equal, it will
be low.

Now for the count pulses. These should
not reach the counter 1C until the up/
down information has established
whether they should be added to or
subtracted from the pulse count. For
this reason, the beginning of each pulse

4

Figure 4. The positioning of the opto
detectors is fairly critical. When one is in a
dark-to-light transition (IC7I the other should
be fully covered (IC6).

is delayed slightly by way of R12/C9.
As figure 5a shows, the negative going
edge of the counter pulse (output N13)
is always just a little late with respect to
the edges in the output signals of N9
and N12. The pulse duration is deter-
mined by the difference between the
RC networks R12/C9 and R11/C8. As
soon as C8 is charged, both inputs of
N13 will have the same logic level, so
that the output will go high again.

Figure 5b, on the other hand, shows
what happens when the wheel is turned
in the opposite direction. Again, there is
a 90° phase shift between N8 and N9,
only this time in a different direction.
The counter pulses (output N13) there-
fore do not coincide with the positive
but with the negative half periods in

5

Figure 5. The waveforms of the decoder in
figure 3. Illustrations a and b each correspond
to a different direction.

the up/down signal produced by N12.

In addition, the rev counter is provided
with a zero stop. The carry signal pro-
duced by IC1 offers this facility, so it
might as well be used and the circuit
is shown in figure 6. When the display
passes from count '999999' to '000000'
or vica versa, a pulse is produced at the
carry output which operates a relay via
flipflop N10/N11. If the system is used
in a cassette recorder for instance, the
recorder motor can be switched off with
a relay. Pressing S3 will reset the device
and the cassette recorder can be started
again.

Construction

First a few general remarks. Although it
is extremely handy that the MK 50398

9-10 - elektor September 1981

6

Figure 6. The circuit of the automatic switching facility provided by the 'carry' output of the
up/down counter. N15 is an unusual gate.

1C can control an LED display directly,
it should be taken into account that this
will produce considerably heat. To
prevent this from getting the better of
the 1C, it is advisable to use good
quality displays like the type rec-
ommended. The segment current is set
at 20 mA. This level should not be ex-
ceeded; in fact, preferably it should be
lower than that, so choose slightly
higher values for R1 . . . R7.

The values of resistors R13...R15
may also need modifying. For the
detectors IC6 and IC7 to produce
clearly defined logic zeros and ones,
it is important to make sure that the
counter wheel is constructed correctly.
Sometimes it may be necessary to try
out various resistor values and so empiri-
cally establish the correct setting for the
two opto detectors. Not more than
50 mA should pass through the detector
LEDsI

Figure 7 shows the printed circuit board
for the rev counter. This consists of two
sections which can be seperated if re-
quired. This allows the display and
switch section to be mounted at right
angles to the rest of the circuit, see the
illustration in photo 2. Points 1 ... 6,
a . . . g, SI... 3, +12 V anj ground
in both sections should be connected
to each other with wire links.

Readers who do not intend to 'pre-
program' a certain count, can omit
diodes D1 . . . D24 and their associated
thumb-wheel switches. This saves quite
a bit of money! The relay may also be
left out, if it isn't needed. As far as the
displays are concerned, the circuit
works equally well using only one!

The counter wheel

The wheel may be made of virtually
any material: cardboard, wood, metal.

O

Photo 2. A suggested method of mounting
the printed circuit boards is illustrated here.

3

Photo 3. Two common types of opto
detector, the reflective type is shown on
the right and the slotted, as used in this
article, is on the left.

plastic, etc. The wheel in figure 2 was
made of Veroboard.

The wheels in both photo 1 and figure 4
were provided with slots rather than
holes. However, readers are welcome to
drill holes, if they prefer. The best
method of making a good wheel is to
carefully draw it on paper first and then
print a reduced photo of it on copper
clad board. Cutting or sawing out the
apertures should then be quite straight-
forward. You can also use a negative
and glue it onto a piece of transparent
perspex, as this can also be very accu-

It does not matter how wide the holes
or slots are, as long as they are all the
same size. The opto detectors' position,
however, is rather critical, although not
to the extent shown in figure 4, where
there only seems to be about half a
vane between IC6 and IC7. What must
be ensured is that when IC7 is in a state
of transistion, IC6 is completely covered
by a vane — it doesn't matter which
one. IC6 may also be mounted at the
other side of the wheel, if this simplifies
matters. This improves the device's ac-
curacy considerably and there may even
be 254 or 354 vanes between the detec-
tors. The vanes should not be narrower
than that of the light beam between
the two halves of the opto detector.

Let's take another look at the opto
detectors. These may be bought ready-
made or may be home-built. Readers
who wish to try their hand at con-
structing them themselves, need to
make sure the photo transistor is well
screened from the surrounding light —
which is easier said than done. 'Manufac-
tured' opto detectors fall into two
categories: 'real' and 'reflective' types.
Photo 3 shows both kinds. Elektor's
design is based on the first type (there
are a few which incorporate two detec-
tors in a single package easel), but the
reflective alternative is equally suitable.
This has the advantage that the counter
wheel does not have to have slots or
holes cut into it. All that has to be done
is to draw a black and white chess-
board pattern along the edge. This
method also has the disadvantage,
that it is not as reliable as it is very
sensitive to any changes in ambient
light.

The type of opto detector used here was
the MCT81 from Monsanto. Make sure
the opto detector radiates infrared
light . . . this is invisible!

Finally, it should be noted that if the
counter is to count 'up' when the wheel
turns in a certain direction and it counts
'down' instead, the opto detector con-
nections merely need to be changed
round. It's as simple as that ... H

9-12 -ele

September 1981

digital barometer

digital

barometer

an electronic weather station

However hard we try, there's not much we can do about the weather,
except maybe draw up a short-term forecast of the rain showers ahead
and repair the roof. Recently, National Semiconductor introduced a
set of transducers to lend amateur meteorologists a helping hand when
building their own barometer. This article describes a circuit that
incorporates both pressure and temperature transducers and indicates
the levels measured on a digital display.

Although the idea to design a barometer
for home construction has been around
for some time, Elektor has had to put
the project 'on ice' until only quite
recently, due to a number of problems
attached to it. For one thing, the
pressure transducers required were so
expensive that it was cheaper to buy a
ready-made instrument. What is more,
any attempt to build the device by less
sophisticated means was doomed with
failure, as its calibration called for an
almost superhumanly steady hand. So it
was a nice idea, but it stayed on the
'future projects' shelf to gather dust.
However, recently the price of pressure
transducers suddenly dropped! Now
nothing, not even the difficult cali-
bration procedure, could stand in the
barometer’s way. And here is the result.

An electronic pressure gauge

As any control engineer will tell you,
measuring atmospheric pressure is really
a straightforward job. But then that's
the point of view of an expert who
thinks nothing of converting mechanical
values, such as stress and force, into
electrical signals with the aid of highly
sophisticated extensometers and the
like. Anyone who tries to do the same
without technical experience and
armed only with a home-made inductive
transducer is bound to encounter some
difficulties. Fortunately, pressure trans-
ducers are now widely available on a
single silicon chip, so that mechanical
precision and RF know-how are no
longer essential requirements. The trans-
ducers are simply soldered onto a
printed circuit board and once 'the
chips are down' the rest of the oper-
ation can be left to them, for their
output voltage gives a direct indication
of the pressure value.

The barometer described here is based
on two tiny silicon strips separated by a
vacuum. According to the amount of
pressure exerted, the upper strip bearing
four piezo-resistive resistors in the form
of a Wheatstone bridge (see figure 1)
will start to cave in. The resistors act as
extensometers and alter in resistance
when the strip bends. As a result, the
voltage across the bridge will also
change.

Monolithic pressure transducers contain
a silicon chip bearing the four bridge
resistors and sometimes an additional
circuit to ensure temperature compen-
sation. There are also 'hybrid' types
which include a complete hybrid 1C to
look after temperature compensation,
amplification and voltage control. Such
ICs have the advantage that they are
already fully calibrated when they leave
the factory, which makes life a lot easier
for the constructor. Not surprisingly,
this type of luxury considerably affects
the price, but since the alternative is to
spend hours building a rough and ready
calibration circuit, it's a question of
choosing the lesser of two evils. Figure 2
provides the block diagram of a hybrid
pressure transducer.

strips are separated by a vacuum (A). The
upper membrane bears four piezo-
resistive strain gauges which together form
a Wheatstone bridge. When air pressure causes
the upper membrane to buckle, the bridge
becomes unbalanced.

The transducers

The illustration in figure 3 shows the
internal structure of the monolithic
pressure transducer used in this circuit,
the LX 0503 from National Semicon-
ductor. As well as a resistor bridge the
device contains a 'transistor diode' to
provide temperature compensation. The
voltage across this 'diode', between
pins 3 and 7, decreases by about 10 mV
per degree centigrade. Using the re-
commended supply voltage of 7.5 V (at
pin 3), the voltage level to the bridge at
pin 7 will be about 3.5 V and a current
of approximately 2 mA will pass
through the bridge.

The circuit will be in a state of equilib-
rium in a vacuum, that is to say, 0 mb
(millibar) of pressure. Ideally, the
voltage level at both outputs of the
bridge (pins 5 and 6) would be equal at
around 1.7 V. In practice, however,
there is a discrepancy of some 100 mV
(offset) between the two outputs.

With a rise in pressure, the bridge loses
its balance and pin 6 will become more
positive and pin 5 more negative. The
pressure-dependent output voltage level
is very low, some 29 ... 1 16 /iV per mb.
The precise value varies according to the
type of transducer used.

The internal transistor stage compensates
the pressure-dependent voltage in
temperature but cannot correct the
offset voltage which also appears at the
output. The offset voltage changes by
between 30 and 120 jtV per degree
centigrade and may rise or fall with
changes in temperature . , . again, de-
pending on the particular transducer
used. Since the offset voltage is as
sensitive to temperature as the trans-
ducer is to pressure, the offset voltage
must have effective temperature
compensation. A suitable temperature-
dependent voltage is produced at pin 7
of the 1C. This output, together with
the bridge output at pins 5 and 6 will
only supply a few milliamps of current.
For readers who are interested, the
technical specifications for the trans-
ducer are provided in table 1.

Designing the circuit
A few further requirements will have to
be met if the transducer output voltages

September

digital bare

are to be displayed in the form of
comprehensive pressure and temperature
values. Figure 4 gives a general idea of
what the final concept should look like.
The pressure transducer output voltage
is amplified by a factor of 41 and is
then compensated for both temperature
and offset. With the aid of a voltage
divider at the output, the scale division
is set to 1 mV per mb. This is of course
just an example and other units, such as
1 mV per mm Hg (mercury), or 10 mV
per metre (height) are equally suitable.
The transducer's temperature voltage Ut
is fed to the high impedance input of an
opamp which has a gain of 3 and which
removes the offset. Since the offset
voltage of the transducer may feature
both a positive and a negative 'reaction'
to temperature, the polarity indicator in
front of the value can be inverted if
necessary.

The temperature voltage produced by
the transducer is also used to display
temperatures. An output stage allows
the calibration point to be adjusted
(T adjust) and the scale division to be
adapted (scale adjust) to 10 mV per
degree centigrade.

A 3% digit DVM is connected to the
output of the circuit to display pressure
and temperature values.

The circuit

Figure 5 shows the complete circuit
diagram. IC1 represents the LX 0503A
pressure transducer. The bridge voltage
that is proportional to pressure is
derived across pins 5 and 6. This will
rise in proportion to any increase in
pressure.

Opamps IC2 and IC3 constitute a
differential amplifier which provides a
gain of 41 and converts the floating
bridge voltage into a grounded output
voltage. Using two opamps has the
advantage that the bridge receives a
small and evenly distributed load due to
the non-inverting inputs of the amplifier
being connected to the two outputs.
Capacitor C2 shorts the bridge output
voltage as far as AC is concerned to
suppress any low frequency signals.
After all, it's a semiconductor baro-
meter we're after - not a microphone!
The dotted capacitors Cl and C2 are
only required, if the opamps used for
IC2 and IC3 are not internally fre-
quency-compensated types, like the
ones recommended here (LM 301 or
LM 309).

Since the initial stage amplifies the
offset further, the output of IC3
produces a voltage which is derived

elektor September 1981 — 9-15

from both the pressure voltage and the
temperature-dependent offset voltage.
Pin 7 of the LX 0503A acts as the
temperature output. A1 buffers the
temperature voltage so that it has a high
impedance. This opamp also serves to
smooth, the DC voltage at pin 7 (about
3.5 V). Since A1 amplifies the voltage
by a factor of 3, its output voltage will
be 0 V +30 mV per degree centigrade.
After this initial preparation, the
temperature voltage passes through a
unity gain amplifier, opamp A2, which
also inverts the polarity. If the non-
inverting input of A2 is grounded by
way of the dotted wire link in the
diagram, it will act as an inverting
amplifier. If it is not grounded, the
input signal will not be inverted.

If the polarity indicator was correctly
chosen, the temperature drift of the
pressure voltage may be cancelled out
with the compensation voltage provided
by A2. that is, if P5 is correctly adjusted.
The compensation voltage is added to
the pressure voltage at the input of A3.
Similarly, a DC voltage that can be
preset with P6 compensates the offset
voltage of the pressure transducer.

At the output of this adder circuit there
is another potentiometer (PI) which
presets the scale division factor. Also

>r September 1981 - 9-17

needed is the wire link shown in its
vicinity, but more about this when we
come to the calibration procedure.

To return to the temperature section of
the circuit, another opamp, A4, is
connected to the output of A1 to
prepare the temperature voltage for
display on the thermometer. P3 is
linked to the negative reference voltage
-7.5 V and allows the calibration point
to be set, whereas P4 at A4's output sets
the scale division for the temperature
display at 10 mV per degree centigrade.
Now for the power supply. The opamps
are provided with a symmetrical supply
voltage of about ±15V. A mains
transformer with a centre tap and a
bridge rectifier produces a raw DC
voltage of 18 ... 22 V across C4 and
C5; the rest is taken care of by two
integrated 15 V fixed voltage regulators.
Auxiliary circuits such as an LED UAA
1 70 voltmeter can be powered by way
of the 18 V (non-stable) output. Further
to this, a 9 V voltage regulator was
included in the prototype power supply
circuit to power a DVM (digital
voltmeter module) and its liquid crystal

In addition, the circuit also produces a
fairly stable positive voltage of 7.5 V for
the pressure transducer and a negative
voltage which acts as a reference source
when calibrating the offset of the
pressure and temperature signal. The
positive reference voltage is produced
by a voltage regulator 1C. the well-
known 723 type in standard format,
and a 741 opamp takes care of the
negative voltage.

Construction

The differential amplifier around IC2
and IC3 must be able to amplify micro-
volts by a factor of 40. If the instrument
is to be used for domestic purposes the
opamps may be standard 741 or
LM 301 A types and R1 . . . R4 may be
carbon resistors. Although these com-
ponents feature greater temperature
drift, this can be remedied by careful
calibration of P5. In any case, 'room
temperature' inside a house should be
relatively constant.

If, on the other hand, the temperature
measurement must be absolutely stable,
metal foil resistors with not more than
1% tolerance are an absolute must.
Their values should be: R1,R4 = 200k,
R2,R3 = 5 k. Opamps IC2 and IC3 may
be selected with the aid of table 2.
There are three different types:
standard, low power and low drift. Low
power types have the advantage that
they distribute temperature evenly very
soon after the device is switched on and
so help to counteract the slow reaction
to temperature changes by the trans-
ducer. This enables the total temperature
drift of the pressure test circuit to be
compensated for with P5.

The printed circuit board in figure 6 is
designed to cater for both ordinary
preset potentiometers and Cermet

trimmer potentiometers. In practice,
however, it is best to use the Cermet
type for PI and P6. P3 and P4 may be
ordinary presets, although it stands to
reason that multi-turn types are a lot
easier to set up. As for P2, P5 and P7,
there is no need for these to be special
types.

First of all, mount the resistors and
capacitors on the board. It is rec-
ommended that 1C sockets are used for
all the ICs. The ICs themselves are not
mounted until the circuit has been
tested and calibrated. Holes are pro-
vided in the board for solder pins and
fitting wire links after testing. Wire links
A . . . C (at P4 and PI ) are only required
during the calibration and it is a good
idea to use solder pins here.

Test equipment

The main requirement is a digital
multimeter, but if one is not available,
use an ordinary multimeter to test and
set the supply voltages and a 3'/j digit
DVM to calibrate the temperature
compensation and the pressure and
temperature outputs. The latter is
needed in any case to display pressure
and temperature values. The DVM
module does not have to be calibrated
with great precision, as it is adjusted
along with the pressure and temperature
display. Thus, it only needs to be set at
2 V full scale, which can be done with
an ordinary multimeter.

When calibrating the pressure display,
have a fairly reliable barometer on hand
by way of comparison. The ideal
solution would be a barometer or an
altimeter with a tube connection, for
then the meter could be constructed in

the manner shown in the figure and so
could measure various values very
quickly. Plastic tubes, valves and T
connectors, etc. can be obtained at very
reasonable cost at the local petshop
(they are used in aquariums). The
petshop may also be able to supply an
electric air pump, although an ordinary
bicycle pump will also be suitable. Using
the air pump, produce a pressure about
30 ... 40 mb above the existing level.
Then connect the valve to preserve the
pressure level and the electronic
barometer can be adjusted at leisure.
Fitting the valve is a simple operation.

Testing and calibrating the circuit

Follow these instructions carefully and
there will be nothing to worry about.

To start with, the transformer is con-
nected to the circuit. Don't forget to
connect the centre tap to the ground of
the board! Now check the supply
voltages: +15 V at the positive pole of
C8, —15 V at the negative pole of C9
and +9 V at the output. These voltages
must not vary by more than 5%.

Next insert IC6 (type 723) and adjust
the voltage across C16 to 7.5 V with P7.
Now mount IC5 (type 741) and check
the output voltage at pin 6, this should
be —7.5 V. It is time to fit the pressure
transducer IC1. Its pins must be bent
slightly to allow it to fit into the DIL
holes on the board. Make sure the
polarity is correct! The wires should not
be too short, as this would cause the 1C
to be overheated when it is soldered.
The pressure opening of the sensor
should be sealed with sellotape before it

7

pressure levels to be measured and so compare the DVM and barometer displays. By closing
the valve, the pressure can be maintained at a constant level for some time in the tube.

Figure 8. How to wire the digital barometer and a digital voltmeter using a 2 V measurement range. Switch S converts the display from pressure
to temperature, and vice versa. Figure 8b shows how the range of a 200 mV DVM can be extended to 2 V.

is soldered to prevent any harmful
moisture from entering the 1C.

After this, IC2 and IC3 are installed.
The voltage at IC3's output (pin 6),
which may be negative, is primarily
determined by the offset voltage of the
sensor. The transducer can now be
tested. Depress the tube of the sensor
slightly with the tip of your finger and
the pressure should increase causing the
output voltage of IC3 to rise by several
millivolts. Once the last 1C (IC4 = type
LM 324) is mounted, the actual cali-
bration procedure may begin.

First of all, the output voltage of A2
(pin 8 of IC4) is set at 0 V with P2
(T offset). Then the DVM module or
the digital multimeter, whichever is
available, is connected to the output of
A3 (pin 14 of IC4, wire link C) and the
voltage displayed is adjusted to about
1 V with P6 (P offset). P5 (T compen-
sation) is turned fully clockwise so that
its wiper will be grounded. Use an
electric hairdryer to warm up the
transducer and the output voltage will
change. The question is now: will it rise
or drop? If it drops, the 'A' wire link
will have to be fitted; if, on the other
hand, it rises, the link may be omitted.
Now P5 (temperature compensation) is
adjusted until the DVM display does not
alter during changes in temperature. The
hairdryer method allows P5 to be
correctly positioned in no time at all. If
the preset range 'runs out', select a
smaller value for R6 and a larger one if
the range is too wide. The hairdryer
method is not particularly suitable for
adjusting the compensation finely, as
the heat is not evenly distributed
throughout the circuit. Give the circuit
'a change of air' and test it under
different temperature conditions to be
absolutely sure it works properly.

A word about P5. If it is turned fully
clockwise, the temperature compen-
sation will be totally ineffective. What is
measured at the pressure output is the
temperature drift of the transducer.
Slowly turn P5 in an anti-clockwise
direction and the compensation voltage
will be added to the pressure voltage
and its polarity will be inverted. As a
result the drift of the display will be
reduced and when P5 is in the correct
position (the compensation voltage is
equal to the drift, so that this is can-
celled out) the display will not alter. If
P5 is turned up too far, the drift will be
over-compensated and inverted. Once
the right temperature compensation has
been found, we may proceed to adjust
the scale division factor. Depending on
the type of 'control' barometer available,
the scale division may be established in

1. Compare the circuit with another
barometer.

Read the air pressure shown on the
barometer. Make a note of it and adjust
the DVM (DMM) display to the same
value with P6. Now wait until the
pressure level alters due to a change in
weather (by at least 10 mb). Make a
note of this new value on the barometer
and compare it to that shown on the
DVM/DMM. Now a small calculation is
called for. By relating the change in
pressure the alteration on the display,
the temporary (not yet calibrated) scale
division factor of the circuit will consti-
tute a mV/mb ratio.

Let's take an example:

Supposing the air pressure (barometer)
has changed from 1003 mb to 1018 mb,
whereas the DVM display has altered
from 1003 mV to 1033 mV. Thus, the
scale division factor is 30 mV divided by
1 5 mb, which is 2 mV/mb.

Now assuming you wish to adjust the
device to a 1 mV per mb ratio. This is
done by reducing the output voltage by
means of PI . The division ratio is
calculated by dividing the required scale
division factor (1 mV/mb) by the factor
obtained previously. Thus:

Divider factor D equals 1 mV/mb div-
ided by 2 mV/mb, so D = 0.5.

D = 1 mV/mb + 2 mV/mb = 0.5
To adjust PI as accurately as possible to
this division factor, PI is linked to 7.5 V
by way of wire link 'B' and the voltage
at Pi's wiper (P output) is adjusted to
7.5 VxD. In this example (D = 0.5) the
voltage is set at 7.5 V x 0.5 = 3.75 V,
Now remove wire link 'B' and connect
PI to the output of A3 by way of the
link 'C' and the scale division factor will
be calibrated to 1 mV/mb.

Once P6 has been adjusted so that the
display shows the value of the 'control'
barometer, the pressure section of the
circuit will have been completely
calibrated. The electronic barometer |
may then be considered ready for use.

2. The pressure meter comparison
(using a barometer with a tube
connection)

PI ind P6 are calibrated in exactly the
same manner as described above, the
only difference being that there is no
need to wait until the air pressure has
changed by more than 10 mb, as differ-
ent values can be obtained quite simply
by pumping air into the system. This |
method enables various pressure values
to be compared to those shown on the
pressure meter.

Finally, the thermometer section still
has to be calibrated. For this a 'control'
thermometer is required and two
different environmental temperatures, |
for example the circuit may be tested [
indoors at room temperature and then j

1981 - 9-19

Figure 9. The connections for a digital barlolthermlolhygrometer. The humidity sensor board
was published in the 1981 Summer Circuits issue.

out of doors or in the garage/cellar, etc.
Be sure there is a real difference be-
tween these temperatures! The DVM or
DMM is linked up to the T output and
P4 is adjusted so that the output voltage
shows a significant difference of at least
10mV/°C (scale division) between the
two temperatures. Then the display is
adjusted with P3 (T calibration) to show
the same value as that on the 'control'
thermometer.

Now the printed circuit board is
, complete, but that does not mean the
device itself is ready for use. Local
electronics dealers are bound to have a
suitable case the circuit can be built into.

1 Once this has been done, the wire links
can be finally fitted. A toggle switch
may be fitted to switch the display
from pressure to temperature and vice
versa. Make sure the case is well venti-
lated, as an airtight box might lead to
errors.

The barometer may be extended into a
hygrometer with the addition of the
printed circuit board published in the
1981 Summer Circuits' Issue: the
'humidity sensor' (circuit number 77).
This circuit can be supplied with the
+9 V output produced by the barometer
board. The DVM display will then also
indicate relative humidity levels.

Figure 8 shows how to wire the digital
'weather station'. If the user wishes to
display several values at once, several
displays will have to be included, but
to add a few more DVMs would be
rather costly. A cheaper solution is to
use an LED voltmeter scale incor-
porating the well-known UAA 170 1C.
This will allow both temperature and
humidity readings to be displayed
simultaneously.

A similar idea was published in
December 1980 in Elektor in the form
of a bath thermometer. Elsewhere in
this issue an article describes how to use
the bath thermometer circuit as an LED
voltmeter. Figure 10 shows the proto-
type of a weather station using a DVM
pressure display and two LED volt-
meters to indicate temperature and
humidity levels.

What else can the circuit be used
for?

Air pressure levels are mostly measured
in terms of millibars, but may also be
indicated as millimeters of mercury as
so many mmHg. One mmHg corre-
sponds to 1.333 mb. If the control
barometer used has an mmHg display,
the electronic version may be calibrated
in terms of mmHg to a scale division
factor of 1 mV/mmHg. This has the
advantage that a 3 digit DVM (like the
universal digital meter published in
Elektor, January 1979) can be used.
Since a pressure level of 1050 mb corre-
sponds to 788 mmHg, the range of the
meter may cover 1 V (999 mV).
Practically every barometer may be used
as an altimeter. Since pressure drops by
1 mb per 8 metres, the barometer display
may be calibrated to a scale division of

10

Figure 10. A suggestion for an 'electronic
weather station’. The two LED scales indicate
temperature and relative humidity levels,
whereas the digital display allows all three,
pressure, temperature and humidity, values

1 mV per 1 0 metres.

In addition, the circuit may be adapted
to transducers in the LX series from
National Semiconductor which have
different measurement ranges including
types which allow liquid pressure to be
measured.

Important!

Some digital voltmeter ICs have a
negative measurement input which must
not be connected to the meter's negative
supply voltage (ground). Such meters
have a separate power supply. Examples
of this are the Intersil types IC1 7106
and 7107. LED DVMs must also have
their own power supply due to their
relatively high current consumption.

In a forthcoming issue a 3V4 digit DVM
with an LCD display will be described
which can be powered by means of the
barometer board.

M

Data sources:

National Semiconductor: Monolithic
Pressure Transducers LX05XXA,
LX06XXD , LX06XXGB Series,

DA-FL25M80, August 1980
National Semiconductor: Pressure
Transducer Handbook.

dB converter

The dB converter circuit

The complete circuit of the dB con-
verter will be seen in figure 1 and can
be discussed in detail stage by stage.

The input preamplifier employs low
noise FETs, T1 and T2. Since the
slightest amount of ripple from the
power supply at this point will affect
the lower threshold in the dynamic,
range, the supply voltage is filtered by
C6 and R10/C5. Diodes D1 and D2 are
included to protect the gate of T1
against any overload. Components Cl,
C2/R4 and C3, C4/R7 constitute
filters with a 20 dB per decade (that
is 6dB per octave) slope. These are
ideal for use with microphones and for
frequencies above 100 Hz.

If, on the other hand, C4 is in the
circuit the frequency response will be
1 dB down at 100 Hz. These filters will
not be necessary when testing hum-free
circuits. This means that the El input
should be used for microphones if
possible. Low frequency noise (rumble)
will be filtered out during measurements
above 100 Hz. At 100 Hz the error
indication will be —2 dB, but will decay
rapidly at higher frequencies.

Below 100 Hz amplification will de-
crease by 40 dB per decade (12 dB
per octave). Capacitors Cl 5 and Cl 6
control the response of the meter.

Auxiliary circuits

The two circuits AC1 and AC2 are used

dB converter

a useful addition to the AF sweep generator

The May 1979 edition of Elektor
included a design for a sweep
generator, which enabled
frequency response curves to be
displayed on an oscilloscope
screen 'at one sweep'. This saves a
lot of work when measuring the
frequency response of amplifiers
or filters. The sweep generator
really becomes an invaluable asset
when it is combined with a
logarithmic rectifier, as the latter
'translates' the frequency response
curves into levels of dB. This type
of rectifier is known as a dB
converter.

This stage is followed by an amplifier,
IC1, which is extremely fast due to
under-compensation. This ensures that
the precision full-wave rectifier receives
a low impedance drive. IC1 has a gain
of 11 and 100% feedback for DC
voltage levels. As a result, its drift is
very slight.

Both IC2 and IC3 in the precision
rectifier are also under-compensated.
Furthermore, there is no feedback
around IC2 until either D5 or D6 starts
to conduct. This could well cause the
circuit to oscillate unless due care is
taken during its construction. If faced
with this problem avoid increasing the
value of C11 and Cl 2 as this may cause
a lowering of the upper frequency
threshold.

The following stage, around IC4, is a
logarithmiser circuit. In the absence of
an input signal, resistors R36 and R37
serve to prevent the voltage at pin 6
from reaching +15 V and causing an
excessive current to pass through diode
D7. Capacitor C17 is included to
stabilise the output.

The inverter amplifier built around
IC5 has a relatively high input im-
pedance and therefore only presents a
slight load on the logarithmiser. The
inclusion of R44 ensures that the
output of the converter is short circuit

The filters

The lew frequency threshold of the
converter depends, as mentioned earlier,
on the high pass filters Cl, C2. R4 and
C3, C4, R7. When switches SI and S2
are operated, in other words input E2
and C3 are in circuit, the frequency
response drops by 1 dB at 10 Hz.

for calibration purposes only. Setting
the output levels of AC1 is simply a case
of adjusting the two potentiometers to
provide 100 mV with switch S in
position 1, and 1 mV in position 2.

Calibration

The main items of equipment required
for calibration are an oscilloscope
(switched to DC) with a sensitivity of
50 mV per cm, and a 20 kfl/V multi-
meter.

All the potentiometers should be set
initially in the mid position. Calibration
starts at the last stage as this can then be
used as a temporary measuring amplifier.
Before calibration begins allow the
circuit to 'warm up' for at least 5
minutes. The procedure for calibration
can then be carried out in the following
order.

la. Separate W7-W8 and ground W8.
Connect the oscilloscope and the

multimeter to point A. With the multi-
meter switched to a suitable current
range set the DC compensation by
adjusting P7 for a minimum reading
on the meter. Switch the meter to its
most sensitive range (50 /iA or so) and
carefully readjust P7 with greater
precision. Then switch the meter to
5 V DC.

lb. Disconnect W8 from ground and
reconnect it to the auxiliary circuit

AC1 with S in position 1 . Now adjust
P8 to obtain an output voltage level of
—3.33 V at output A.

2. Link W7 and W8 together and
separate W5 and W6. Connect AC1
to W6.

(i) With S in position 1, adjust P5 to
obtain 0 V at A.

(ii) With S in position 2, adjust P6 to
obtain -4 V at A.

(iii) Repeat (i) and (ii) until no improve-
ments can be made.

(iv) Connect W6 to +10 V and measure
+4 V at A.

3a. Disconnect W6 altogether. Separate
W7-W8 and W3-W4. Link W5 and
W8 with a short length of wire. Remove
Cl 5 and C16 from the circuit. Connect
W4 to the auxiliary circuit AC2.

3b. Adjust P2 . . . P4 to achieve a
symmetrical full-wave rectified sine
wave on the oscilloscope at output A
for all voltage levels between 1 and 10
mV. The rectified sine wave will be
inverted by IC5 so that its sharp 'spikes'
will touch the zero line. Initially. 100 K
can be used for R21 and R27. If in
doubt, reverse the polarity of P2 and/or
P4 that is, either apply +15 V or —15 V!
Replace C 1 5 and C 1 6.

4a. Connect W8 to W3 and ground
point D of T2.

4b. Set PI so that the zero line appears
on the oscilloscope.

5. Replace the links W1-W2 .. .W7-W8
as drawn on the circuit diagram.
Disconnect point D of T2 from ground.
The trimmers are now fully calibrated.
6a. Connect the oscilloscope to TP2
and provide an audio frequency
signal at the E2 input. Switching SI and
S2 should show a 1.2 amplification
factor.

6b. Switch SI off and ground the El
input. The zero line should now
appear on the screen, slightly noisy
(hiss) but without any signal.

6c. Disconnect the El input from
ground and reconnect it to not
more than 10 mV. Check whether the
amplification factor is equal to about
960.

The power supply of the existing sweep
generator can be used to power this
additional circuit. *<

Technical specifications

frequency range 10 Hz (—1 dB) ... 100 kHz

dynamic range 80 dB

response range -40 dB . . . +40 dB

input El 160 pV = 0 dB

input E2 130 mV = 0 dB

The maximum definition on the s
0.5 dB = 1 cm where the sensitive
oscilloscope is 50 mV/cm (DC).

TV games extended

The extended version

This extension to the TV games com-
puter was designed, quite literally, in
response to repeated requests from
readers who wanted more 'elbow

The best way to illustrate the new fea-
tures is by means of a memory map as
shown in figure 1. In the basic version,
the monitor occupied the first 2K of
memory and this was followed by 2K of
RAM (1% R for the user). The address
range from 1 000 was virtually unused —
it only contained a few input/output
lines and the PVI. Now, a further
3K RAM is added, from 1000 to 1BFF,
and two programmable Sound Gener-
ators (PSGs) are located at addresses
1D00 to 1D1F.

The area from 1 800 on gets rather
confused, owing in part to a minor error
in the original monitor ROM. The Read
Cassette address is located at 19BF,
instead of 1DBF where it belongs. Since
the addresses in this range were not
fully decoded, this made no difference
at the time. Now, it complicates
matters. A facility must be added to

TV games extended!

another 3K of memory, and sound effects!

The basic TV games computer (Elektor, April 1979) has nearly 2K of
program memory. This has proved to be adequate for a lot of
interesting games, without the programs getting completely out of hand
- bearing Parkinson's Law in mind: 'Computer programs grow until all
available memory space is filled'.

However, several of the more dedicated TV games computer fans have
reached the point where they can put a greater memory range to good
use. Witness the majority of the 30 (!) new games to be released shortly
on ESS tapes. For this reason, we feel that it is now time to release an
extension board that almost triples the available memory. Also included
on the board are two 'Programmable Sound Effects Generators' (PSGs).
In this article, we will only give a brief description of 'how it works'
and 'how to build it'; fuller details, plus a lot more interesting
information about the TV games computer, are contained in a book
that will be available in the very near future.

The extension board contains one further interesting feature: provision
was made for plugging in games cartridges that are intended for some of
the commercial machines!

disable the RAM at 199C . . . 199F and
19BC...19BF. This is selected by
setting S3 to position V and S5 to
position 'B'.

The total 'memory map 'from 1 800 on is
shown in greater detail in the right-hand
half of figure 1. First, the RAM with the
two 'input' ranges mentioned above;
then a gap; then the sound generators,
at 1D00, followed by some input and
output ranges (which include cassette
read and write). At 1 E8E, the keyboard;
and, finally, from 1 F00 on, the PVI,
Readers who feel so inclined can, of
course, add other features. A hardware
random numbers generator at 1D20?
Other custom-designed input/output
devices in the 1 Dxx range? A commer-
cial-style explosion sound generator at
1E80? They're all possible, under one
proviso: they musn't be used in pro-
grams submitted to us for including in
the ESS service!

The extension board

A block diagram of the extension circuit
is given in figure 2. It is fairly straight-
forward. To cater for the additional
load, the address and data lines from the
main board must be buffered. The
actual extensions are shown in the
lower half of the figure: ROM con-
nectors, for commercial game car-
tridges; RAM extension; Programmable
Sound Generators, with their associated
audio outputs. Above these are the
control circuits: the address block
decoder, with switches for selecting
commercial ROM game cartridges as
required; the decoder for addresses

elektor September 1981 — 9-23

i==4

corresponding to the (erroneous) input
blocks; the PSG address decoder and
control circuit. A further connector to
the main board, drawn centrally in the
figure, carries a few 'odd' signals like
'clock' and 'PVI audio output'.

The PSGs

The programmable Sound Generators
are the most noteworthy part of the
extension board. These ICs are quite
complicated, as illustrated in the block
diagram given in figure 3. Each 1C con-
tains 16 registers, corresponding
to addresses 1D00...1D0F or
1D10 . . . 1D1F, as shown. These regis-
ters are controlled by an address/data
demultiplexer in the 1C — unfortunately,
in our case, since it means adding an
external address/data multiplexer as
shown in figure 2!

The data stored in the registers controls
three tone generators, a noise generator
and associated mixers; an envelope
generator, with associated D/A con-
verters; and two input/output ports. To
start with the latter; in the normal case,
each input/output port is 8 bits wide,
and can be used at will — independently
from the rest of the PSG, provided it is
enabled by the corresponding bit in R7.
In the application described here,
however, they can only be used as
outputs. Furthermore, the four least
significant bits of port A in PSG 1
(corresponding to address 1D0E) are
used for amplitude control of the PVI
sound output.

The basic frequency of each tone
generator can be set anywhere in the
whole audio range — the TV games
computer book includes a table for
selecting any note in an eight-octave
range. The basic noise generator 'fre-
quency' can be determined. The desired
outputs can be enabled individually.
The output amplitude for each channel

can be set, or else controlled by the or noise generator by means of R7
'envelope generator'; in either case, the (address 1D07 or 1D17) is not sufficient
result is determined by a 4-bit D/A to completely 'kill' it. For complete
converter, corresponding to 16 ampli- silence the 'amplitude level' of all
tude levels. outputs must be set to zero, by storing

The envelope generator can be set for 00 in registers R8 . . . RA. This we have
single-shot attack or decay effects, like discovered after a lot of frustration . . .
explosions, or periodic amplitude All in all, a phenomenal range of sound
effects like tremelo or machine-gun fire. effects are available. The original
It should be noted that disabling a tone application note gives an 'explosion'.

'gun shot', 'European siren', 'laser
effect', 'whistling bomb', 'wolf whistle',
and 'racing car'. In no time at all, we
added a rattling chain and several
polyphonic wedding marches. Great
fun!

Construction details

The complete circuit is given in figure 4.
Virtually the only point worth nothing
is that it conforms to the block diagram
given in figure 2 . .. Some further details
are given in the TV game computer

For the purposes of this article, we can
pass on to the printed circuit board,
given in figure 5 — reduced to 70%, for
reasons of space - and, more import-
antly, to the wiring diagram given in
figure 6. As can be seen, there are two
connection points to the main board:
the main 31 -pin connector and a 14-pin
DIL connector which runs to a 16-pin (!)
version in the original IC6 position on
the main board. This 1C, a 74LS139, is
now located on the extension board
(IC15). More precise details are given
adjacent to figure 6.

Two new audio outputs are provided,
one from each PSG; the PVI sound is
mixed into each. Although this offers
the possibility of 'stereo' sound effects,
we found it easier to combine the whole
lot into a single audio output by means
of a few 4k7 mixer resistors. Crazy —
we offer an option, and don't use it
ourselves! However, we must cater for

1WS3PH1

lifflSSi

KMSEIE

•SHssSlfMaiii

ins

A final word on the power supply. The
original version, unfortunately, may
sometimes prove rather 'under-powered'.
The complete extension board gets
quite hot — and heat comes from
watts . . . The supply must be capable
of delivering nearly 2 A, without getting
too hot under the collar. If you made
the same mistake that we did — making
the original supply too light-weight —
some modifications are now in order.
The power transistor needs an adequate
heat sink, preferably mounted outside
the case. Also, it may prove necessary
to provide the bridge rectifier with a
heatsink. The easiest way to check these
components is to touch them with a
moist finger. If it sizzles, they're too
hot!

A plea to programmers

Now that extended memory is avail-
able, some ardent programmers may
feel that they would like to submit
programs to Elektor for publication.
Fine, but before you do we would like
to stress some urgent requests. In the
first place, please bear in mind that
other TV games computer owners may
not (yet) have extended memory avail-
able. if at all possible, arrange the game
so that a basic version will run on a
basic computer and the fully enhanced
version only becomes operative if ex-
tended memory is actually available.
The 'Maze adventure' and 'Memory'
games on the new ESS tape are ex-
amples. In both cases, a branch to
'initialise for extended memory' is
included as follows:

0C1000 LODA, R0
BAFC BSFR, N, 1000
If memory extension is not available,
the data at 1000 will be FF (negative), so
nothing happens; in the extended case,
a subroutine at 1000 modifies a few
instructions in the basic program to
incorporate the program extensions.
Owners of a basic version will merely
notice that the load routine breaks off
at 'AD = 1000'; they can start the pro-
gram, however, at PC = 0900.

The other request concerns the joy-
sticks. Please remember that no two
joysticks are alike! A calibration pro-
cedure is essential. M

H LOT OF
COOd

LSI 39) .at

lisco lights controller

1981 - 9-29

It would appear to some that the
essential ingredients of enjoyment for
the younger generation consist mainly
of company, noise . . . and coloured
lights, with plenty of each. This is not
entirely true, of course, but a disco
light system goes a long way towards
enabling the junior members of the
family to entertain their friends on
those nights that ten teenagers manage
to sound like fifty.

The main features of the system de-
scribed here are its safety, and its high
performance at relatively low cost. A
further advantage lies in the fact that,
unlike many designs for disco lights
controllers, there are none of those
fearful coils to wind. With economy in

disco lights
controller

It seems that parties are out and disco's are in with the younger set and
it goes without saying that the most successful disco's are those that
copy the 'real thing' in terms of presentation. This means that a 'light
show' is an absolute must. The disco lights controller featured in this
article will perform equally well as some of the more expensive systems
available, while still remaining reasonably economical in terms of cost
and components.

mind, the present design has just the
usual three channels, although there is
little to prevent the enterprising
constructor from expanding this to
double the number or even more . . .
simply by building two (or more)
complete systems.

For convenience, the signal input can be
taken from either the preamp output
(preferably the tape output) of the
amplifier or direct from the speaker

terminals. In the latter case, the limiter
shown in figure 2 can be placed in series
with the input of the controller to
reduce the effect of the volume control
of the audio system. P2 should be set at
minimum resistance.

The important safety aspect of the
circuit is that the mains voltage section
is separated from the rest of the
electronics by means of opto couplers.

The circuit

The input signal amplification is taken
care of by opamp A1 , seen to the left of
the circuit diagram in figure 1 . The
preset potentiometer P2 is used to
adjust the amplification factor of this
stage between 1 ... 20. The sensitivity
is varied by PI and is a maximum of
100 mV RMS when both PI and P2 are
turned fully up.

The output of the preamp feeds twin T
filters (A2 . . . A4) of the three channels.
This form of filter design is easy to
construct and maintains good selectivity.
At the same time however, these filters
do tend to be 'peaky' for this appli-
cation, in other words, a little too
selective at 5 Hz, 1 kHz and 5 kHz. To
counter this effect to a certain extent,
the three resistors R5, R9 and R13 are
included to give some measure of
'attenuation' to the filters.

The filter outputs are rectified and
smoothed by diodes D1 . . . D6 and
their associated capacitors before being
fed to the three comparators A5 . . . A7.
The switching thresholds of these are set
by potentiometers P2 . . . P5 which can
vary the voltage level at the inverting
input of each comparator from about
IV to 8 V. The resistor R20 has been
included to prevent any input level from
reaching the 1 2 V of the supply voltage,
as this would completely confuse the
comparators.

At this stage it should be pointed out
that potentiometer P3 controls the high
frequency channel, P4 the middle range
and P5 the low frequency channel.

With no input signal being fed to the
circuit, the comparator outputs will
be low. This will cause the LED in the
opto couplers to light (via the 1 k
resistors R22, R24, R26) thus turning
the photo-transistors hard on to hold
the thyristor gates down to zero volts.
No disco light will be lit yet.

Upon the arrival of an input signal large
enough to switch the comparator of any
channel (Adam and the Ants and
colleagues will ensure that it does!) the
opto coupler LED will go out switching
the photo-transistor off. The thyristor
gate will now be taken high by the 47 k
resistor and there will be light . . . from
the bulb of that particular channel.

The reason why thyristors were chosen
in preference to triacs is due to the fact
that thyristors are far more sensitive. In
this circuit for instance, a gate current
of only 300 pA is sufficient to activate
the thyristor. Furthermore, the gate
control current can be mains powered
via a straightforward circuit, without

the need for large capacitors or 'heavy'
resistors. The problem is that thyristors
only allow half of the mains waveform
to pass. This can be remedied by
rectifying the mains voltage with the aid
of diodes (D9...D12). The 15V DC
supply for driving the gates is derived
from the mains voltage by means of
R28, D7, D8 and Cl 7. A simple 12 V
power supply is included to provide the
electronic section of the circuit with
power. Except for the transformer, all
the components are situated on the
board. A 220 V mains voltage that is
derived from the mains switch and a
separate fuse is linked to points X and Y
on the board. Since the power supply is
fed to the lamps via diodes, the latter
determine the peak current flow, or
rather, the power rating of the bulbs.
This must not exceed 200 W per channel
and so the thyristors will not require
heat sinks.

Figure 2. The input limiter circuit. If this
circuit is used to connect the disco lights to
an output amplifier, the volume control will
not affect the lights' operation to such an

UiL*

R22, R24 and R26 are mounted a little
higher up on the board. The anode of
each LED is connected to the output of
the comparator (IC2) by way of a 1 k
resistor. The cathode of each LED is
grounded. Be careful to earth the case if
it is made of metal!

During AM reception on a radio in close
proximity to the controller there may
be some interference caused via the
mains. This can be filtered out by con-
necting the filter in figure 3 in series
with the mains supply to the controller.
The coils are obtainable ready-wound
from good electronics dealers. Other
filters may be used, provided they are
suitable for 3 A. M

Figure 3. The noise filter. The coils can be
obtained ’ready-wound’.

The construction

An LED may be inserted in the front
panel for every channel. For this resistors

QUAD ESL63

9-32 - elektor September 1981

An opportunity to examine the
circuit diagram and to listen to
Peter Walker's new loudspeaker
was too good to miss . . .

QIM)

It has never been our policy to
publish review articles on audio
products - and we are not going
to start now. This does not mean
that we, as critically-eared
engineers, do not have our
opinions on the standard of
engineering demonstrated in any
particular product. Therefore
when the manufacturer provided
us with a copy of the circuit
diagram and an opportunity to
listen to a Quad ESL 63 we did
not put up any resistance . . .

Having said that, we believe that the
following article, about the rather
intriguing innards of this unusual
loudspeaker, will interest many of our
readers. After a more general introduc-
tion on electrostatic loudspeakers, to set
the stage, we shall go into more detail
on the ESL 63. Finally we shall discuss
the way the ESL 63 behaves as a
doublet and how it interfaces with
listening rooms.

The electrostatic loudspeaker

An electrostatic loudspeaker is in several
ways the counterpart of the ubiquitous
moving-coil loudspeaker. Firstly in the
strictly theoretical sense: the force

patterned electrodes to be used. (See
figure 2.)

One more interesting aspect is that of
the damping of the fundamental
resonance. In an electrostatic drive unit
this is the resonance between the
diaphragm compliance (due to the
earlier-mentioned restoring force) and
the mass of air in the immediate vicinity
(plus of course the very small mass of
the diaphragm itself). Musically oriented
readers may note that this air-mass-load
Cinertance' in the acoustical scheme of
things) is the same mechanism as that
requiring end-correction in the tuning of
an organ pipe. Electrical damping of this
resonance can be achieved by regulating

ESL 63

exerted on an ESL diaphragm is
proportional to the applied voltage
rather than to the current, whereas it is
the current flowing in the ESL motional
impedance rather than the voltage
across it that is proportional to the net
diaphragm velocity. Another aspect is
that an ESL diaphragm, carrying only a
more or less static charge (where else
could the name come from?), can be
very large in size and yet light - much
lighter in fact than the mass of air that
it moves. Figure 1 shows a cross-section
of a modern electrostatic drive-unit. The
diaphragm is a very thin flexible thermo-
plastic film, stretched tightly to provide
a restoring force directed towards the
central equilibrium position. It must be
possible to charge the diaphragm
reliably without the individual charges
being able to distribute themselves
unevenly during excursions — otherwise
the force, which of course acts on the
charges would be unevenly distributed
as well. This calls for an extremely high
surface resistivity. The 3-micron
polyester diaphragm in the ESL 63 uses
electronic conduction, obtained by
doping it with one donor atom to about
ten million non-conductive atoms. (This
is incidentally something of a techno-
logical breakthrough and it was in fact
Quad's toughest development problem).
The fixed plates, perforated to allow the
passage of air, actually form a parallel-
plate capacitor. The AC voltage applied
across this capacitor sets up the signal
field and the driving force is simply the
total charge times this field strength.
When the diaphragm moves it will cause
a charge-displacement in the plate-
circuit and we see that the rate of
diaphragm movement (velocity!) must
be proportional to the rate of charge-
displacement (current). In practice the
perforated plates are also of a thermo-
plastic material, with a conductive layer
printed on the outside; this greatly
simplifies high-voltage insulation
problems and in the ESL 63 also enables

1

Figure 1. Cross section of a typical
electrostatic drive unit with external circuit.
The apparent simplicity is misleading.

the motional current (in the moving-coil
case, the voltage). In the ESL 63 there is
also internal acoustical damping by
means of an airflow resistance.

Cabinets

Loudspeaker cabinets as we know them
exist mainly to achieve useful bass out-
put from the moving-coil driver, with its

QUAD ESL 63

1981 -9-33

elektor September

relatively small diaphragm area. They
evolved from typical open-backed radio-
set cabinets, when it was realised that
destructive interference by the phase-
inverted backwave was reducing the net
bass pressure in the room. Not counting
bass-horns — which are acoustical
matching-transformers and tend to be
enormous — there are two basic ideas:

either absorb the backwave in a closed
box, or use it to drive a phase-shifting
resonant system that will provide extra
bass output (bass-reflex and some
'transmission-line' - labyrinth - systems).
As we see it, two properties of electro-
static drivers make it unnecessary and
undesirable to mount them in cabinets.
Unnecessary because they can easily be

made so large in radiating area that the
destructive interference problem is only
significant at the very lowest musical
frequencies. There it can be overcome
by bass-boosting — without distorsion —
since the drive system itself is linear.
(The system must of course be able to
make large enough excursions: cubic-
metres-per-second of volume-velocity
remain cubic-metres-per-second.) Unde-
sirable because boxes, by their nature,
operate with considerable internal
pressures. Such pressures would tend to
come straight out through the electro-
static diaphragm.

Delay-line matching

One of the problems with electrostatic
drive is the reactive current flowing in
the plate-to-plate capacitance at high
working frequencies. Peter Walker
pointed out, as long ago as 1954, that
this problem can be solved by using
several loudspeaker sections as the
shunt-elements in an LC delay-line.
British patent number 1 228 775, pub-
lished in 1971, explained how this
matching arrangement could be deliber-
ately used to control the radiation
pattern of a large-surface electrostatic
loudspeaker. In May 1979 an AES paper
finally made it clear what was coming.

A section of the delay-line used in the
ESL 63 is shown in figure 3. It is rather
intriguing. With the diagonal capacitors
it looks like a first-order all-pass net-
work. It probably is, although at least
part of the capacitor-current will null
that through the winding stray-capaci-
tance (shown dashed). The apparently-
shorted secondaries in fact apply
damping to the inductors, either to
make the all-pass section behave itself
on transients or to provide amplitude
taper along the line (or both). The delay
per section is 24psec, which corresponds
to a path-length difference in air of just

The essential point about this method
of matching is that the unwanted
acoustical reflection at the edge of the
finite diaphragm will appear as an
electrical reflection on the line. It can
therefore be eliminated by a simple
electrical modification to the line.

High voltage audio

The signal voltage applied to the ESL 63
drive system can peak at over ten
kilovolts. This is necessary to achieve
field-strengths near the ionisation (flash-
over) limit, across a total airgap wide
enough to permit sufficient diaphragm
movement at low working frequencies.
Designing an audio transformer to
actually do this, over the full frequency
range and at low distortion, must have
been an interesting exercise . . .

In fact the ESL 63 has two identical
quite hefty transformers, wired with
their secondaries in series. Apart from
enabling the bulky things to be fitted
neatly inside a shallow base, this offers
an elegant way of reducing the leakage
inductance and stray capacitances that

9-34 - elektor

QUAD ESL63

set the upper limit to a transformer's
bandwidth.

A point worth noting here is that there
is no reason why an iron-cored audio
transformer should in any way degrade
the performance of the circuit in which
it is used. On the contrary, use of a
transformer is often the best way - if
not, as here, virtually the only way — of
doing the job.

Figure 4 plots the modulus of the input
impedance. Perhaps surprisingly, it does
not look too different from that of a
typical conventional loudspeaker.
Maintaining the diaphragm charge in
spite of leakage - and extra losses due
to localised airgap ionisation — requires
the application of EHT to the (semi)
conductive surface. The voltage should
be so high that it will produce a polar-
isation field strength in the two airgaps
(i.e. between the diaphragm in rest and
each fixed plate) equal to half of the
ionisation threshold. In the ESL 63 this
is about 5.25 kV, corresponding to
about 2 kV/mm. The diaphragm charge
is of course proportional to this polar-
isation field strength.

The EHT generator is shown in figure 5.
It is a classic Cockcroft-Walton cascade
rectifier with one little addition: the AC
feed is roughly stabilised by VDRs to
make the EHT more or less independent
of mains voltage fluctuations. A neat
further detail is that the charge is
delivered via a capacitor-bridged neon
lamp. This, together with the 1 OM-
ohm resistor and the very much higher
leakage resistances, forms another
classic circuit: the flashing-neon relax-
ation oscillator. The number of flashes
per second is proportional to the rate at
which charge is delivered to the
diaphragm. Presumably this originated
with Peter Walker's need to keep an eye
on what was happening. One can almost
hear him say: '. . . extremely sensitive
and much better than clumsy meters
with all those kilovolts around.’

Protection

An electrostatic loudspeaker is funda-
mentally linear up to the point at which
ionisation occurs in one or other of the
airgaps. As soon as that happens you
have a few milliseconds left to get the
drive voltage off, otherwise a flashover
will permanently damage the system.
The protection circuit must therefore
operate almost instantaneously, then
hold long enough for the ions to cool
down. Figure 6 shows how this is
achieved in the ESL 63. The high-
frequency noise radiation that ac-
companies the onset of ionisation is
picked up by the antenna - a length of
wire running around the high voltage
circuitry — and detected by T3. Noise
above a certain level is a reliable indi-
cation that a potentially dangerous
situation is building up. When this
occurs the 555 timer will trigger, firing
triac T1. Then... power amplifiers
beware. This loudspeaker hits back . . .

The breakover diode T2 and triac T3
transfer the firing of T1 to the audio
input in the event of the mains power
being off. The circuit is therefore fool-
proof.

This arrangement will on itsown protect
the loudspeaker against accidental over-
loading. The power amplifier obviously
must have a well-designed short-circuit
protection system, even when it is
incapable of delivering an excessive
output voltage. (We noticed a shut-
down apparently caused by the lady of
the house plugging in an electric kettle!)
We have just seen that the protection
system operates drastically and without
warning. This could be very discon-
certing to a listener who is concen-
trating on a loud musical passage: one
instant the loudspeaker is performing
superbly well, the next instant it is
silent. The ESL 63 therefore also has a
soft input-clipper, designed to start
giving an audible distortion-warning

5

about 3 dB below the shutdown level.
How this works can be seen in figure 7.
The signal is fed to the audio step-up
transformers via small series resistors.
Tr2 acts in a preset-adjustable voltage-
threshold circuit. When the input
voltage peaks exceed 40 volts, in either
direction of swing, Trl will be turned
on, drawing extra current through the
series resistors. This will cause an
audibly non-linear reduction in primary
drive voltage.

The clipper circuit could presumably be
disabled without affecting the safety-
margins in any way. We would in fact
have preferred an optical warning device
— also driven from a monoflop —
instead of a headroom reducing clipper.
How about a switchable option?

FRED

The acronym FRED -full range elec-
trostatic doublet — is another Quad
original. Let us now try to understand
how it works. Peter Walker illustrates
his principle of radiating an expanding
wavefront from a flat diaphragm, by
means of truncated delay-line matching,
with figure 8. Others have interpreted
this as producing a virtual point source
about 30 cm behind the loudspeaker as
viewed by the listener. The real ESL 63
situation seems to be more complex.

An acoustical doublet consists of two
equal sources of opposite sign, each
small compared to the wavelength and
spaced a similarly small distance apart.
The net pressure at a distance from the
doublet that is large compared to the
spacing is accurately described by a
cosine function (the cosine of the angle
between frontaxis and observer
direction). This is shown in figure 9. If
we've got it right, what Peter Walker has
done is to arrange the delay and ampli-
tude shading of the diaphragm drive
(with the entire diaphragm behaving as a
phased array of real sources) to
maintain the doublet-like axial lobes
throughout essentially the entire
working range, even though the array
quickly becomes large compared to the
wavelength. The radiation patterns of
figure 10, showing only a slight sharpen-
ing at 8 kHz, illustrate just how well he
has succeeded . . .

This is not only a fine loudspeaker: it is
a fascinating piece of applied physics.

Listening . . .

Two aspects of the loudspeaker-room
interface that can cause trouble in a
domestic listening-environment are early
reflections and standing waves.

A simple case of an early reflection
problem is shown in figure 11. The
loudspeaker is standing on a highly
reflective floor, so that the listener
receives sound along the direct path and
indirectly off the floor. The reflection
will arrive perhaps one or two milli-
seconds after the direct wave and with
rather less intensity. The most effective
approach is to imagine that the floor is

9-36 — elektor September 1981

QUAD ESL 63

transparent with the reflection being 9
produced by an image of the loud-
speaker underneath it. The extra distance
travelled by the delayed wave will
correspond to one or more half-wave-
lengths of some musical frequency
(usually in the mid-range). The problem
is that destructive interference will
cause a partial cancellation of the
pressure due to the direct wave,
resulting in fairly broad response-dips
centred on frequencies corresponding to
odd numbers of half-wavelengths. The
whole-wavelength situation, with the
two waves more or less in-phase, will
result in broad peaks that are usually
even more unpleasant than the dips.

The image-method can be applied to
more complex situations provided one
uses enough images to account for all
the troublesome reflections. Note, by
way of an example, that a double-
bounce reflection would be 'radiated'
by an image of an image. We believe it
was this effect that we ran into on our
first hearing of an ESL 63, in a room
with a tiled floor. The loudspeaker was
well clear of walls — but it 'honked'
slightly. The effect disappeared when
we either moved the loudspeaker into a
carpeted room or else raised it off the
floor on a makeshift standard. (An
empty milk-bottle crate, actually. Op-
tional extra?)

The carpet presumably attenuated the
reflection sufficiently; the effect of the
milk-crate is more instructive: refer
again to figure 1 1. We started by noting
that the ESL 63 radiates as a doublet.

with axis horizontal and about 50 cm
off the floor. Nowf it in the listener, with
his ears about 1 metre off the floor and
about 3 metres away from the loud-
speaker. The vertical angle subtended by
the direct listener-path at the loud-
speaker is . . . etc. etc . . . This produced
the puzzle that the milk-crate operation
is . . . etc. etc . . . This produced the
puzzle that the milk-crate operation
could only have reduced the reflected
wave pressure by a couple of decibels.
Then it dawned. Moving the image
source about 35 cm further below the
floor had caused its listenerpath to be
intercepted by a coffee-table!

A standing wave as a room-interface
problem should be distinguished from a
standing wave behaving itself as part of
the reverberation process. Any standing
wave mode is characterised by its
natural frequency and its degree of
damping. The problem only appears
when a single lightly-damped mode, or a
closely-spaced group of such modes,
occurs in isolation. Musical tones from
an instrument or a loudspeaker,
particularly sustained tones, that
happen to be near the natural frequency
can build up a forced vibration of
distressingly high amplitude. After the
tone stops the mode will decay more or

QUAD ESL 63

aptember 1981 — 9-37

11

Figure 11. An illustration of the image source method applied to the simple case of one
reflection off the floor.

The next step was to shift the working
range of the main driver downwards in
frequency, then call it a woofer. Later
again came the squawker — an epithet
that never really stuck — and now we
have ultra-widerange systems that add a
supertweeter and, sometimes, a sub-
woofer. The separate subwoofer systems
that are becoming available should more
logically be called rumbters or thumpers

The question now is: granted that the
ESL 63 delivers its bass output the hard
way — into the relatively unfavourable
doublet airload — does it need help?
Peter Walker says quite emphatically
'no'. For one thing the ESL 63 bass
response is more extended and better-
controlled than that of its predecessor.
For another, its doublet radiation
pattern gives a more aperiodic interface
to a typical listening room than available
auxiliary bass systems do - because
they are omnidirectional radiators.

On the other hand, people who insist on
realistic (?) reproduction of organ pedal
- or for that matter heavy trucks and
underground trains - may disagree.
That of course is up to them . . . (They
might of course try listening right in
close to a doublet. Proximity effect will
provide quite an amount of bass boost!)

M

less slowly at its natural frequency.
Beats can occur if two or more modes
are excited together (by the same tone)
and then decay independently.

The advantage of a doublet in this room
interface problem is that its output is in
the form of a particle-velocity along the
axis. It will therefore only couple to
modes that have a significant com-
ponent of particle-velocity along the
axis at the doublet position. This is
probably one of the reasons that elec-
trostatic doublets are considered bass-
weak, even when like the ESL 63 they
are not. The room simply doesn't give
the expected (liked?) booming response.

Combating the room-boom problem
with a doublet starts by feeding it with
a low-level sinewave (at the 'right'
frequency, obviously!) and then moving
it — and yourself — around, until the
offending mode is identified. Then try
to find a position or orientation for the
loudspeaker that will sufficiently
weaken the unwanted coupling.

To woof or not to woof . . .

Many years ago, when all available
loudspeakers had a single nominally full-
range cone driver, some innovative soul
introduced the concept of the tweeter.

9-38 - elektor September 1981 analogue LED display

The UAA170IC differs from its IC1 can be adjusted between 0 and
UAA 180 counterpart in that it presents 5.2 V with the aid of PI. The con-
a dot display rather than a row (or bar), nection for the upper reference voltage
In other words, up to two LEDs light (pin 13) is linked to pin 14 and is there-
simultaneously , which cuts down the fore fixed at 5.2 V.
circuit's current consumption consider- By slightly modifying the bath ther-
ably. The three inputs at pins 11 ... 13 mometer circuit the circuit can be
(see figure 1) are of vital importance, converted into a universal LED display,
Whereas pin 11 acts as an input for the which not only (roughly) measures
voltage being measured, the reference voltages, but is also suitable for
voltage at pins 12 and 13 determine the displaying non-electrical values. Together
upper and lower thresholds of the with the barometer described elsewhere
measurement range and are adjustable in this issue and the Elektor humidity
from 0 to 6 volts. sensor, the two bath thermometer

Using, for instance, the bath ther- printed circuit boards and the piece of
mometer circuit published in December Veroboard shown in figure 3 can be
1980, the lower reference voltage at built into a practical weather station. As

analogue
LED display

using the UAA 170 1C

This very useful instrument has a
variety of applications. It can be
incorporated in voltmeters, rev
counters, humidity meters,
thermometers and in fact in any
type of meter that provides some
kind of analogue indication. The
display consists of a circle of
LEDs which are controlled by a
UAA 170 1C.

analogue LED display

elektor September 1981 - 9-39

Figure 2. How to connect the circuit in figure 1 to the barometer board published elsewhere in this issue. The latter provides a +/-15 V and
+18 V power supply.

Figure 3. How to wire the bath thermometer board to the Veroboard. on which the circuit
containing opamps A1 and A2 (figure 1 ) is built. The two inputs are marked C and C' and the
two outputs E and E'. The latter are connected to one display board each. The required break
in the track is shown in the copper track pattern of the bath thermometer board.

it can be seen from figure 1 , the ther-
mometer circuit has been slightly
modified: R4 and the NTC resistor have
been removed. The voltage proportional
to temperature reaches pin 11 of IC1
from the output of the opamp A2 by
way of P2. The two opamps are mounted
on a separate piece of Veroboard.

The temperature output which was
already present on the barometer board
is calibrated to allow the voltage to
change by 10 mV when the temperature
changes by one degree centigrade, so
that this is 0 V at 0°C. This enables
temperatures to be read on a DVM. In
the case of a temperature indicator
consisting of 16 LEDs (16°C display
range), the difference between the
lowest and the highest temperature level
(= 160 mV) will have to be amplified at
the input voltage range of the display
(5.2 V). Thus,

5.2 t 0.16 = 32.5.

A1 will therefore have to amplify the
temperature output voltage by that
factor at least. Using the indicated
values for R4 and R5 this amplification
will be about a factor of 39. As a result,
P2 can easily set the precise scale
division factor. A2 cancels out the
inversion of A1 and allows the required
display range to be shifted, which
without P3 would cover 0...16
degrees. This is a little low for a room
thermometer. If, on the other hand,
the voltage at the wiper of P3 is 120 mV,
the row of LEDs will not start to rise
until the temperature is 13 degrees. If
the temperature is lower than this, only
the first diode will be lit.

From the four opamps inside the
LM324IC, two amplifiers are left to
allow the same circuit to be used again
for the humidity sensor. P3 and R8 can
be omitted here as humidity in the
atmosphere is indicated in 10% steps

between 0 and 100%, so that only the
first 10 LEDs are ever required. R5
should only be 47 K in this case, as the
output voltage of the humidity sensor
is between 0 and 1 V.

Figure 2 shows how the barometer
board and a 'case' containing the entire
circuit in figure 1 are assembled. The
positive and the negative supply voltage
for the LM 324 can be derived from the
barometer board. For this purpose, a
short length of insulated cable is
soldered to both the positive pole of C8
and the negative pole of C9, as these are
the outputs of the two 78L15 and
79L15 stabilisers.

How to calibrate the temperature
display

1. Ground the wiper of PI.

2. Connect 1.3 V to input C,

3. Calibrate P3, so that LED 2 starts to
light.

4. Connect 0.27 V to input C.

5. Calibrate P2 until the 16 th LED
lights.

How to calibrate the humidity
display

1 . Ground the wiper of PI .

2. Connect 1 V to input C.

3. Calibrate P2 until the 10*h LED
lights.

Figure 3 shows how the display is wired
to the board and how the latter board is
constructed: Output E' must be connec-
ted to another, similar display. Inputs
A . . . D must be connected to the
barometer board in the manner shown
in figure 2. Connection C' leads to the
humidity sensor's output.

The (bath thermometer) display should
be modified as follows:

1. Break the track connection between
the wiper and one end of P2 (see
indication).

2. Solder a wire (E) to the disconnected
end of P2.

3. Remove the NTC resistor and R4 (in
other words, do not mount them, if
you haven't already done so). M

9-40 - elektor septe

sr 1981

P. Gabler

Without electricity there is very little
you can do in electronics, therefore, a
good quality power supply occupies a
very important place in any reader's set
of priorities. It should have a wide
adjustable voltage range providing a
current of at least one or two amps. In
addition, it would be a considerable
advantage if the power supply were to
include a built-in meter to display the
output voltage and current levels. The
ideal meter is a DVM which combines
great accuracy with an easy-to-read
display.

The idea certainly makes sense. After
all, plenty of stable power supplies have
appeared in Elektor's pages and so have
several digital voltmeters. So the

wlt/ammeter
fi»* power
supplies

plus . . .
an automatic
range switch

This handy circuit enables a DVM
to be connected to a stabilised
power supply allowing both
voltage and current levels to be
measured. Furthermore, an
automatic range switch is provided
and a compensation network to
measure the actual output current
without increasing the power
supply's internal resistance.

question of combining the two should
be a straightforward proposition, surely?
But ... it isn't, in fact, quite of few
problems are involved.

The circuit we're about to discuss
manages to overcome the various
difficulties and provides a successful
DVM-and-power supply combination.
What's more the meter does not only
indicate the voltage and the actual
output current, but also switches from
one range to another automatically. All
that the operator has to do is flick the
switch to select either voltage or current
measurement.

The basic principle

Figure 1 contains the block diagram for
the well-known 'series stabilisation' type
of power supply. A differential amplifier
compares a reference voltage U re f to a
voltage derived from the output by R a
and Rb- The differential amplifier
output controls a power transistor T
which is connected in series with the
power supply line. Usually a current
limiter is included which reacts to the
voltage drop across a series resistor R s .
More often than not, a bleed resistor R n
is connected across the output to ensure
stability at even very small loads.

In order to calculate the output current,
measuring the voltage across R s ought
to be enough, but, unfortunately, this is
not very accurate. The reason for this
will be explained later. Measuring the
output voltage is really an easy job, as it
merely involves connecting the DVM to
the output terminals. A voltage divider
should not be left out here, as the

volt/ammeter for power supplies

smallest range a DVM can have is some-
where around 1 or 2 V. Right, now let's
get back to the measuring current.

Figure 2 gives a view of the output
section in the power supply including
the power transistor T and the resistors
connected to the output. R1 and R2
constitute the voltage divider for voltage
measurement. The regulator maintains
the output voltage at a constant level
(to the right of R s in other words).

By measuring the voltage at R s - as-
suming we know the value of R s — the
current passing through the resistor can
be calculated quite easily. The point is,
this will not be the 'real' output current
of the power supply, because additional
current passes through resistors R a , Rb,

R ! , R 2 and R n . One way to avoid this
would be to connect another 'R s '
resistor after those mentioned in the
positive line and measure the current
there. This is inadvisable, however, as
this increases the internal resistance of
the circuit.

The author has arrived at a very practi-
cal solution. As figure 3 shows, the
addition of two more resistors, R3 and
R4, turns the supply into a bridge
circuit. Here R x replaces R a , Rb, Ri,
R 2 and R n - If the values of R3 and R 4
are chosen so that their relationship to
each other is equal to that of Rs to Rx
( R 3 / R 4 = R s /R x ) the differential volt-
age Ud between points A and B of the
bridge will be nil, irrespective of the
power supply voltage. Once this require-
ment has been met, the bridge voltage
can be used to measure the 'real' output
current. This is because the bridge
voltage equals the current passing
through R s minus that passing through
R x . Readers with a flair for maths can
check it if they like; we're more
interested in the result.

The meter circuit

Now that we know where and how to
measure, we can start putting our ideas
into practice. The block diagram for the
circuit we have in mind is shown in
figure 4. The bridge voltage that is
required to measure the current is
obtained by connecting the junction
R3/R4 to amplifier IC1 and by Unking
the zero tine of the meter circuit to
junction R s /R1. The amplification of

elektor September 1981 — 9-41

volt/ammeter for power supplies

IC1 is set so that its output voltage is
1 V when the power supply produces
1 A. Preset R4 is adjusted to balance the
bridge with no load on the power
supply. In other words, the bridge
voltage is 0 V. A diode and a capacitor
are connected after the amplifier to
produce a peak value measurement.

The output voltage of the power supply
is derived from the voltage divider
junction R1/R2 and is fed to the
inverting input of amplifier IC2 (since
the polai ity of the voltage at the upper
end of R 1 is negative with respect to the
ground of the meter circuit). IC2 is
amplified so that the output voltage is
1 V when the power supply's output
voltage is 10 V. This IV 'standard-
isation' (IV per A and 1 V per 10 V)
allows current and voltage levels to be
measured without any difficulty when
the DVM is adjusted to the 1 V range.

To save having to alter the measuring
points any further, an automatic range
switch has been added after the
amplifiers IC1 and IC2. This has the
advantage that the DVM does not have
to be readjusted during use.

Figure 4 shows two voltage dividers
followed by the comparators K1 and K2.
Each comparator activates the voltage
divider whenever its input voltage
exceeds 1 V.

Finally, there is a single switch allowing
the operator to select between voltage
and current measurements.

The practical construction

The circuit diagram for the volt/ammeter
circuit is shown in figure 5. The dotted
area contains the existing stabilised
power supply to which the meter circuit
is to be connected. R1 and PI constitute
one side of the bridge, PI being used
to regulate the bridge voltage to 0 V
when the power supply is under no load.
The combination of IC1.D1 and Cl
together with associated resistors acts as
an amplifier/peak rectifier. P2 adjusts
the amplification, so that the voltage at
Cl will be 1 V when the power supply
provides a 'real' output current of 1 A.
Next follows the voltage divider made
up of R6, P3 and R7. The output of
comparator A1 will switch when the
voltage level at its input exceeds 1 V.
This threshold level may be adjusted
with P4. When A1 switches, the elec-
tronic switch ES4 (which was 'off' up
until then), will be activated allowing
only one tenth of the measurement
voltage to reach the DVM. A1 in turn
switches A2, which then operates a
relay to display the correct decimal
point. Thus, two measurement ranges
are available: 1 A and 10 A. The voltage
divider can be calibrated for the 10 A
range by means of preset P3.

The voltage measurement section is
basically the same as that for current
measurement, except that the diode and
the capacitor used for peak measure-
ments are excluded, as they are not
needed. After all, a stabilised power
supply ought to be able to produce a

2

constant voltage! The output voltage
measurement is derived via resistor R2.
This voltage level is fed to the inverting
input of IC2 (since the voltage at R2 is
negative with respect to the meter
circuit's ground, as was seen earlier). P5
sets the amplification of IC2 so that the
output voltage of this 1C is 1 V when
the power supply output is 10 V. The
output of the comparator A3 is set by
P7 to switch at 1 V. This will activate
ESI which is placed in series with part

Figure 3. By adding resistors R3 and R4, a
bridge circuit is formed.

of the voltage divider R 1 3, P6 and R 1 5.
P6 serves to calibrate the voltage divider.
By means of the relay Rb. A4 selects
the correct decimal point in the display.
The voltage ranges thus obtained (and
indicated) are 10 V and 100 V.

The switches ES2 and ES3 connect
either the voltage resulting from the
current measurement or that of the
voltage measurement through to the
input of the DVM. The switch SI |
determines which of the two (current or I

Figure 4. The block diagram for the current measurement circuit including an automatic range

voltage measurement) switches is
selected.

In principle any DVM may be used in
this circuit, but the conversion section
for the decimal points, as shown at the
right-hand side of the diagram, was
designed specifically for the universal
digital meter published in Elektor in
January 1979. This uses common anode
displays.

The power supply for the circuit needs
to be a symmetrical 5 V unit which is
entirely separate from the stabilised
supply being measured. This is vital,
because the meter circuit's ground
connection is connected to the positive
pole of the stabilised supply. It therefore
follows that the zero line of the meter
circuit must never be linked to the zero
of the power supply. The same 5 V
power supply will feed the DVM.

Last but not least . . .

Understandably, this type of circuit
needs to be calibrated with due care, if
it is to operate with any accuracy. But
before we proceed to the actual cali-
bration process, it should be pointed
out that the current meter section will
only work for 100%, provided the out-

put resistance of the supply (at a zero
load) remains constant at every output
voltage level. In practice this means that
the negative feedback must be derived
from the wiper of the calibration
potentiometer. As a result, the total
resistance of R a and Rb in figure 1 will
stay constant at any output level.

Since a DVM is required for the meter
circuit, it must be assumed that this has
been calibrated beforehand. First the
power supply is adjusted to an (un-
loaded) output voltage of 10 V and then
PI sets the voltage at Cl to 0 V.

Using the formula U = I • R. the drop in
voltage across the current limiting
resistor in the supply is calculated with
respect to an output current of 1 A (the
resistor drawn in the diagram has a value
of 0.33 SI but this value may vary from
one supply to another). The supply is
loaded so that the output current is
about 1 A. The voltage at the non-
inverting input of 1C is measured on the
DVM and the supply's output voltage is
adjusted so that the measured voltage is
equal to the calculated value. After that
the voltage at Cl is measured again and
the value is adjusted to 1 V with P2. A
better method is to measure the output
current with the aid of an accurate
ammeter, if available, then adjust the

voltage at Cl to 1 V for a current value
of 1 A.

Next P4 is set to make the output of
comparator A1 switch over at 1 V. The
voltage divider R6, P3 and R7 is cali-
brated by putting a maximum load on
the supply and then measuring the
voltage at Cl. The DVM is then con-
nected to the meter's output with SI on
I and P3 is adjusted until the voltmeter
indicates one tenth of the previous
measurement.

Now the voltmeter section has to be
calibrated. The output voltage of the
power supply is adjusted to 10 V
precisely (measure with the DVM). The
voltage at IC2's output is set at 1 V with
P5. P7 is then adjusted until the output
of comparator A3 just switches over.
The voltage divider is calibrated by
adjusting the supply to a maximum
voltage level and then measuring the
output voltage with the DVM. The
latter is then connected to the meter
circuit's output, SI is set on U and P6 is
turned until the meter indicates one
tenth of the value that has just been
measured. Finally, the decimal point
connections are linked to their corre-
sponding connections on the voltmeter

M

uber 1981 - 9-43

ICs that talk

There is a simple way to produce
synthetic speech. Feed the desired
words through an analog-to-digital con-
verter and store the output in a memory.
As required, this data can be recovered
and passed through a digital-to-analog
converter to produce a spoken message.
Easy, yes, but no-one does it this way.
With good reason: it would require at
least 64 Kbits of memory for one
second of speech!

For a viable system, some kind of
drastic data reduction is required. Up
to this point, all manufacturers are in
full agreement; from here on, they all
differ. Broadly speaking, there are two
main approaches.

The first is to make full use of the ex-
perience gained in telecommunication
systems. Post Office engineers dis-

In real life, the techniques used are
rather more complicated and less easy
to explain. 'Signal coding', 'Waveshape
coding', 'Adaptive delta modulation’ -
you name it, they use it. Quite sen-
sational results can be obtained, too:
the total data requirement can be
reduced- from 64 Kbit/s to about
2 Kbit/s. Bearing in mind that this sort
of thing can be done to normal tele-
phone conversations, without the
listener noticing the difference, it is
understandable that several manufac-
turers have adopted this approach for
their 'synthetic speech' chips. The
National Semiconductor 'Digitalker' is a
prime example.

The second approach to the problem is
to analyse the way in which humans
make speech sounds, and then try to

chattering chips

'Great minds think alike', they
say, and it certainly seems true of
major semiconductor manufac-
turers. Within a short period of
time, 'talking chips' have been
announced by Texas Instruments,
General Instruments and National
Semiconductor — to name just a
few. Some of these systems are
in a price range that even makes
them viable for use in toys.

With talking clocks, computers,
washing machines and telephone
exchanges just over the horizon, it
is worth taking a look at the
principles involved.

covered long ago that there is an awful
lot of redundant 'information' in nor-
mal speech. This doesn't refer to the
'Mmmm's and 'welllU's or colourful
legal or political phraseology; in
straightforward everyday speech, there
is relatively little real information per
word. As an example, think of vowels:
when they occur, they tend to last for a
while. Instead of giving a series of 12-bit
digital samples at a rate of 8- to 10-
thousand samples per second, it would
be sufficient to give a code that uni-
quely defines that particular vowel
sound, followed by a further code to
define the duration.

1

simulate this artificially. This, in turn,
can be done in two different ways. You
can draw up a list of all possible sounds
that can occur in human speech (so-
called 'phonemes'). It turns out that
there are less than a hundred in all, so
that's not too bad. Having coded these
into a memory, the information re-
quired to string them together into
words and sentences is astoundingly
little: only 70 bits per second! This
system, as used in the Votrax 'Speech
Synthesiser', has one major disad-
vantage: the output sounds extremely
artificial, since all natural inflection is
missing.

: igure 1. Simplified circuit of the National Semiconductor 'Digitalker'.

2

Figure 2. Drastic data reduction techniques
are used in the Digitalker system. A 'slice' of
the original speech is shown at the top.
Phase-angle adjustment converts this to a
symmetrical signal (b); after adaptive delta
modulation and half-period zeroing, it would
look as shown in (c). A digital code, corre-
sponding to the signal shown in (d), is all that
is actually stored in memory.

Alternatively, you can make electronic
'lungs', 'vocal cords', 'mouth and nose
cavities' and 'lips', and apply fairly
complicated control signals to these at a
relatively slow rate (once every 25 milli-
seconds or so). This is how the Texas
Instruments 'LPC Solid State Speech'
system works.

The Digitalker

As shown in figure 1, the basic Digi-
talker system consists of only a few
components: the Speech synthesis chip,
a 128 Kbit memory (sufficient for
aboutl 28 words), a crystal oscillator
and filter/amplifier.

The data in the memory is derived from
spoken words. These are sampled and
converted to digital information, after
which several compression techniques
are used to reduce the data to a manage-
able quantity. The first step is to remove
redundant (repetitive) information.
When several nearly identical periods
appear in the total signal, these are
replaced by the codes for a single period
and a further code indicating how often
this must be repeated. On average, this
can reduce the total data required to

about one quarter.

A further technique used to 'compress'
the volume of data is known as 'adapt-
ive delta modulation'. This is based on
the fact that speech waveforms are
relatively smooth: sudden level j'umps
are uncommon. Therefore, the infor-
mation required to define the difference
between two successive samples is less
than that required to define the absolute
level of any given sample. In plain
language: given a particular voltage level
at one point on the waveform, you only
need to add or subtract a small amount
to obtain the next level.

A further step is called 'phase-angle
adjustment’. Without going into higher
maths and Fourier analysis, it is almost
impossible to explain how this works.
However, we can attempt to give a basic
idea. It is fairly well-known that any
complex signal can be described as a
mixture of sinewaves, each with differ-
ent amplitudes and phase angles.
Furthermore, human hearing is not very
sensitive to phase information. Given
these assumptions, it is not such a total
surprise to find that the same set of
frequencies and amplitudes can produce
an almost infinite variety of different
waveforms, provided the phase angles
can be varied at will. The trick used in
this system is to adjust the phase angles
to produce a waveform that has mirror
symmetry and, furthermore, a low
amplitude for at least half of the period.
It then becomes possible to reduce the
low-amplitude part to zero and mirror
the rest - reducing the total data re-
quired to one quarter!

The total effect of the latter techniques
- phase-angle adjustment, half-period
zeroing, adaptive delta modulation and
mirroring - is shown in figure 2, for a

3

small portion of the speech waveform.
The original waveform is shown in
figure 2a, and the phase-angle adjusted
version in figure 2b. Believe it or not,
these two signals sound the same! The
next step, in figure 2c, shows the effects
of adaptive delta modulation and half-
period zeroing; finally, figure 2d shows
the signal that must be stored in mem-
ory (in digital format).

When it comes to reproducing the
spoken words, a large degree of flexi-
bility is required. Male or female voice?
Loud or soft? And so on . . . However,
the processor chip takes care of all this,
on the basis of data stored in ROM by
National Semiconductor themselves (ac-
cording to the customer's specification).
So for the purposes of this article —
dealing with general principles — we will
skip that part.

The Votrax speech synthesiser

The basic Votrax system is even simpler
than the Digitalker. With only slight
over-simplification it can be shown as
a single 1C! (See figure 3).

A 6-bit input (P0 . . . P5) selects one
of the 64 'phonemes' that the SC-01 can
produce; furthermore, a 2-bit pitch
control input helps to add inflection to
the 'speech' output. As shown in table
1, the 'phonemes' are the elementary
sounds that can occur in normal speech.
Only a very low data rate (approxi-
mately 70 bits per second) is required to
tell the chip how to tack these
phonemes together to produce words.
This system has obvious advantages. The
total memory capacity required is
minimal, and programming is relatively
easy. Furthermore, Votrax maintain a
library of phonetically programmed
words and they can even supply a

Figure 3. The Votrax system requires little more than a single Id Control signals are applied
to its inputs, at a rate of only 70 bits-per-second, and speech appears at the audio output.

Table 2.

K8 K9

10100 01101 01111 1010 1010

ioioo oiiio oioii iooo 1100

10011 10001 01010 0110 1001

11010

10010 01101 00111 1000 1100

10001

1010 1010 1001 0111 1000 100 101
1000 1100 1101 1000 0100 100 011
0110 1001 1111 1011 0101 010 000

1110 31 01 0111 001 011

00000 10100 01011 1011 1000
00000 10001 01011 1011 0110
00000

00000 10011 00111 1010 0110
00000 10010 00101 1011 0101

V- VOICED
UV - UNVOICED
E = ENERGY
R = REPEAT
P = PITCH

K1 . . . K10 - FILTER PARAMETERS
Table 2. This sequence of digital code words will make the Texas In

Figure 5. A further block diagram of the Tl syst

micro-computer system for automatic
conversion of English text into phoneme
sequences.

However, there is also a disadvantage:
the total capability of the system is
determined by the actual phonemes that
the chip can produce. This is an ex-
tensive list, admittedly, but it is limited
— and the same phoneme comes out the
same way each time. Even when full use
is made of pitch control, and with
amplitude control added at the output,
the final result always tends to sound
artificial.

Texas Instruments Solid State
Speech

A prime example of the third approach!
As shown in figure 4, the system is
closely based on the natural human
speech mechanisms. A simplified dia-
gram of 'the real thing' (figure 4b) is
very similar to a block diagram of the
speech 1C (figure 4c).

The more extensive block diagram
(figure 5) gives a better idea of the
necessary control signals. Starting at the
left: the noise generator and tone

generator provide the basic signals for
unvoiced and voiced sounds, respect-
ively. In the actual system, four bits
control the pitch of the tone generator
and a further bit 'operates the switch'
for voiced and unvoiced sound. Then
the level is set, using a further four bits.
Finally, a total of 40 bits set the filter
parameters or one bit indicates that the
existing parameters can be maintained
('repeat'). An example is given in table
2: the total code sequence for the word
'Help!' As can be seen, a good 500 bits
are needed for approximately half a
second of speech; this corresponds to
something over 1 Kbit/s, which is
fairly typical for this system.

To work out those bits, for a given word
or phrase, is another matter entirely.
However, Texas really uses modern
technology to the hilt, for prospective
customers! On the European scene, it
works like this. A customer either sends
in a high-quality tape with the desired
texts, or else sends in a word-and-phrase
list. The example given by Tl is rather
unfortunate:

"CLOSE THE... DOOR" (!)

The intended meaning of those dots is
to distinguish the phrase 'close the'
from the word 'door'. Instead of 'door',
you can insert other words like 'win-
dow', 'blinds' and so on.

alektor September 1981 - 9-47

6

Figure 6. A three-chip self-contained system using the TMS 5100.

7

Figure 7. The TMS 5200 is intended for use in an existing microprocessor system.

If a written list is sent in, any desired
accent must be clearly specified. The
final result will sound like the original
human speaker! In fact, this is one of
Tl's handicaps: they already have an
extensive vocabulary on file — 'in

standard TV/radio mid-American ac-
cent'! No good here . . .

Given the actual recording (provided by
the customer or made by Tl), it is
converted into a digital code and trans-
mitted by satellite to Houston. At that

end, it is run through a computer that
extracts the pitch and voicing informa-
tion. At the same time, the program 1
computes an optimal set of coefficien' i
for the filters. This is followed by a
'frame repeat analysis', where similar
sets of filter control bits are replaced by
a single 'repeat' bit.

Then, all this digital information —
corresponding to the 'raw' conversion —
is transmitted back, via satellite, to
Nice. There, the data is auditioned by
an expert speech technician (who, in the
words of a Tl engineer, is 'an extremely
multi-lingual young lady'); audible de-
ficiencies are edited by hand. At this
point, the customer is called in for final
evaluation and, it is hoped, approval.

All right, so you've got your 'speech
data' stored in ROM. So how do you
use it? According to the application,
you can choose between two different
'chattering chips'. The older version
(the 5100) was used in the Tl 'Speak &
Spell' unit; it is intended for talking
games, clocks, washing machines and
telephone exchanges - in short, any
application where it must 'talk' on the
basis of simple control signals. The
block diagram of a basic three-chip
system is given in figure 6. Alternatively
there is 'big brother', the 5200, intended
for use in conjunction with an existing
microprocessor system. This is shown in
figure 7.

The link with microprocessors opens an
infinite variety of applications. You can
store standard words in EPROM,
together with a complete set of pho-
nemes - the latter can be used to con-
struct other words, as required. It is even
conceivable that you can design a
speech-analysis program, so that you
can talk into a microphone and use the
resulting data in a program. Alterna-
tively, we may be able to supply Texas
Instruments' standard library on tape,
as part of the Elektor Software Service
- pick out the words you want, and
burn them into EPROM! This is getting
interesting! . . .

Maybe, in the near future . . .

In conclusion

Chattering chips are here — there's no
doubt about it. Chattering? We've even
heard them produce an absolutely true-
to-life 'human' voice! Of the systems
proposed at present, Texas Instruments
seems the winner — both in price and
in capabilities. Price-wise, it wins on
points from National — and the pro-
gramming is easier and, therefore,
cheaper and quicker. Quality-wise, it
beats Votrax — even though the latter
is cheaper on memory. And when
you're looking for 'the' system for the
future, you want something that is both
cheap and good.

There are other systems, of course.
AMI, General Instruments, Hitachi,
Intel, ITT, Matsushita, Philips, TSI . . .
However, they all use (variations of) the
systems described above. Enough is
enough! For one article, anyway. M

9-48 - elektor September 1981

transistor

u

I II! H

lie

ignition soldering
alu min ium

H.M. Wolber

A year has passed since Elektor pub-
lished the transistor ignition system in
the 'car issue' (April 1980) and yet the
project continues to interest and in
some cases, judging by the letters we
receive, puzzle our readers. This article
will try to answer some of the more
common queries and provide a few
practical, constructional hints. Many
readers have sent us their ideas on the
subject, for which we are very grateful,
as they have proved to be very useful
indeed.

The capacitor across the contact
breaker

Quite a few questions concern the
capacitor across the contact breaker.
As the April '80 article stated, this ca-
pacitor must be retained even with elec-
tronic ignition. Its capacitance, however,
may not exceed 0.1 /jF. In combination
with R1 a higher value will produce an
excessive time constant, effectively
retarding the ignition timing. One way
to solve the problem would be to lower
the value of R1, but then the current
through the contact breaker will be too
high and increase the wear rate. This
of course leads to a deterioration in
ignition timing — one of the main prob-
lems that electronic ignition systems are
designed to cure. If the standard capaci-
tor has a value greater than 0.1 pF it is
best to try and fit one that is exactly
that value. If necessary, a normal low-
voltage paper type (at least 60 V) may
also be used, since the electronic ig-
nition system does not produce a high
induction voltage across the contact
breaker. Bear in mind that the capacitor
will need to be thoroughly water
proofed. The original capacitor is left
where it is but unconnected. Should
anything go wrong with the electronics,
resort to the original capacitor.

Adapting the ignition system to different
engines (or vice versa?)

The Elektor transistor ignition was
originally designed for 4 cylinder 4
stroke petrol engines with a maximum
engine speed of 6000 rpm and an ig-
nition system with a single spark plug
per cylinder and a single contact
breaker. For this reason the monoflop
and the RC time constants in the cir-
cuit are based on the ignition frequency
of this type of engine.

The ignition frequency depends on the
engine speed, the type of engine (2/4
stroke), the number of cylinders and the
ignition system (one or more contact
breaker sets, spark plugs per cylinder).

Using the circuit for any other type of
engine and/or ignition system than
the ones mentioned means having to
modify the circuit of the transistor ig-
nition. Not only will the monoflop and
RC times be involved here, but also the
switching stage. In 6 and 8 cylinder
engines and motor cycles, very low
impedance spark plugs are sometimes
used, for which our transistor ignition is
totally unsuitable. Unfortunately, we
are unable to modify the circuit to cater
for every individual requirement, as
many people have requested us to. We
can, however, say that the circuit will
operate correctly with engines where
the firing frequency is up to 12,000 per
minute and where the spark plug
(including resistors if applicable) have a
resistance of at least 1.5 S2. In other
words, the so-called 'super spark plug' is
not suitable here.

How lo convert from a transistor ignition to
a conventional type and vice versa

A number of readers have asked us
whether it is possible to use a switch
with several contacts or a relay to
switch from a transistor ignition to a
conventional system and vice versa.
This could certainly be done with a
switch, but the connections between
the switch and the transistor ignition
should not exceed 1 0 ... 20 cm, so
in most cases the switch cannot be
mounted on the dashboard. On the
other hand, a relay would be very
unreliable, as any vibration could cause
the ignition to become intermittent,
which would be highly inconvenient, to
say the least.

Turn the engine off before switching
from one system to another!! Otherwise
the switching process might cause an
'artificial' ignition to occur with detri-
mental results for the engine.

Constructors use aluminium a great deal
because it is relatively inexpensive and
very easy to work with. However,
soldering it causes problems, for using
an ordinary soldering iron and solder
just does not work. The solder does not
seem to feel particularly attracted to
aluminium. Nonetheless, Mr. Wolber has
come up with a solution which allows
'normal' (60/40) solder to be soldered
on aluminium.

The problem with aluminium is that it
oxidises quickly. In other words,
aluminium is always covered in a layer
of oxide. As you know, this is a very
good insulator and prevents solder
from even touching the aluminium.
However hard you try to scrape the
oxide layer off, oxidation occurs so
rapidly that a new layer will appear
as soon as the 'old one' is scraped
off.

To prevent oxidation, apply grease or
oil to the solder point. Then scrape off
the layer of oxide with a sharp object
underneath the oil. The oil prevents
oxygen from reaching the clean alu-
minium. Now drip hot flux onto the
area with a soldering iron, causing
the oil to evaporate and the flux to
cover the area. The area can now be
soldered. To ensure a solid connection
make sure plenty of heat is produced
by using a soldering iron of at least
100W. If necessary, the area can be
held over a gas flame to remove all
traces of oil before soldering.

Elektor infocard 21

There are two typing errors on the Elektor
infocard 21 (Standards 5) in the tone and
frequency table referring to musical instru-
ments. The 'B' tone in octave 0 and octave 1
should be 30.8677 and 61.7354 Hz respect-

Summer Circuits' Issue, circuit no. 11

The second sentence in the third paragraph
should read: 'This can easily happen if the
positive supply is always present before the
negative is applied."

elektor septembe

High performance cassette drive
mechanism

Papst Motors Limited have announced a new.
very high quality cassette drive mechanism
which out-performs many studio reel-to-reel
tape transports. Known as the MOD 303, it
breaks entirely new ground in that the
laminations for the three d.c. motor stators
employed in the unit are an integral part of
the pressure die-cast chassis. The capstan and
two spools are driven directly from the motor
shafts, there being no belts, pulleys or other
mechanical interfaces to reduce reliability.

The cassette drive can be operated either
horizontally or vertically at a cassette tape
speed of 1.2,2.38 or 4.75 centimetres per
second. For extremely high quality repro-
duction, tape speedsof 9.5 and 19centimetres
per second are available as an option. How-
ever, it is interesting to note that running at
4.75 centimetres per second, the MDD303
specifications exceed the DIN45511/1 re-
quirements for studio tape recorders running
at 38 and 76 centimetres per second.

rated speed, a weighted DIN peak wow and
flutter of ± 0.07% and only 35 seconds to fast
wind a C60 cassette. The deck runs from 24 V
d.c. and consumes 0.65 amps. Operational
temperature range is 0 to 50° C with a typical
operating life of 5,000 hours. The MOD 303
weighs 1,550 grams and measures only
140 x 95 x 76 mm, which is approximately
half the size of any competitive machine.
Three sets of contacts are included within the
MDD 303, one to indicate that the cassette is
loaded, and two designed for record lock-out
applications using the tabs on the back of the
cassettes. The latter two can be used for other
purposes defined by customers if required.
The pinch wheel and record/playback heads

accurate speed control. The motors include
Hall effect sensors to sense drive shaft

rotation. This produces three pulses per

revolution and can be employed as a tape

position counter, end of tape monitor and to

actually sense tape motion.

Gordon Hesketh-Jones.

Papst Motors Limited,

Pamell Court.

East Port way,

Andover,

Hants SP103LX6.

Telephone: (0264) 53655

(2063 M)

probe to be slid along a track at a low angle
without damaging its surface, whilst being

sharp enough to probe coated boards if

required. Plug-in leads facilitate quick probe

cable replacement.

As well as testing PCB's, the Hy-Trak 100D is
appropriate for a wide range of similar tasks,
e.g. as a continuity tester or fault finder for
cable assemblies, wire wrapped boards, or
wiring looms. It can also be used by skilled
personnel as a de-bugging tool for isolating
faulty IC's.

Omnitest Limited,

Highcliffe House,

4 1 1/4 13 Lymington Road,

Highcliffe. Christchurch,

Dorset PR 3195.

Telephone: 04252 77731.

Telex: 418449

(2071 M)

Short-circuit locator and
micro-ohmmeter

The Hy-Trak 100D is a brand new instrument
for precisely locating shorts on PCB’s. Aural
and visual indicators enable the user to probe
quickly to the exact fault location, even when
the short-circuit is invisible. The instrument
provides significant advantages compared with
other similar products. It is the first of its
kind to use a digital display, giving high
accuracy and resolution. It also has compre-
hensive input protection and a radically new
probe design. DC injection is used, enabling
probing to be carried out on capacitively
loaded circuits. The Hy-Trak 100D doubles as
a very accurate Micro-Ohmmeter.

The instrument has five resistance ranges,
including the ultrasensitive ’Hy-Trak’ range
having a resolution of 100 Micro-ohms. To
locate short circuits, two probes are placed on
the suspect PCB tracks. Hy-Trak then displays
the resistance of any interconnecting path.
For resistances less than 60% f ad., an on-
board oscillator produces a tone which
increases in pitch as the probes are moved
doser to the fault. When the resistance is
within 2% of fjd. on any range, the pitch
doubles to emphasise fault location. On the
’Hy-Trak’ range, short circuits can be pin-
pointed exactly, even on large voltage planes.
The probes are made of a new material that
does not bend or distort under load. Probe
tips are an all-in-one design that cannot fall
out during use. Long screw threads are used
to prevent damage to the probe body. Probe
tips have a hyperbolic profile which allows a

7-segment decoder driver

Cosmocord Limited is now marketing a
7-segment decoder driver incorporating input
latches and constant current output circuits,
for use with common cathode type LED
displays rated at a nominal 20 mA at 1.7 V
per segment, without the need for current
limiting resistors. The 9368 device accepts a
4-bit binary code and produces output drive
to the appropriate segments of the 7-segment

feature means that data can be routed directly
from high speed counters and frequency
dividers into the display, without slowing
down the system dock or providing inter-
mediate data storage. Another feature of the
9368 is its ability to be driven from an MOS
device in multiplex mode without the need

9-50 — elektor September 1981

Neons for illuminated push
buttons

IMO Electronics have now added a mains
operated (neon) to their series of illuminated
push button switches and indicators.

The illuminated push button series 01.00 now
has the added advantage of a red neon housed
in a glass envelope complete with limiting
resistor mounted on a Tl 3/4 miniature
groove base.

The higher brightness of this neon is achieved
by incorporating a lens at the end of the glass
envelope, therefore focusing the light output
in one direction and a special diffusor will
prevent the two anode hot spots being
prominent and allowing an even overall
distribution of light output.

The button colours designed for this red neon
are red, orange and yellow.

Available in both 110 VAC and 240 VAC
the lamp contacts are fully isolated from
switching contacts which are rated at up to
5 amp @ 240 VAC.

IMO Electronics Ltd.,

349 Edgware Road,

London W2 IBS.

Telephone: 01 7239047/8

(2072 M)

DIY speaker range

An entirely new range of speaker components
is being introduced by Wharfedale. The kits,
named Speekercraft, will enable music and
hi-fi enthusiasts to construct their own

from the company’s well-established ready-
assembled speakers. All the drive units offered
under the Speakercraft label can be mounted
in various combinations to give a custom-
made loudspeaker ideally suited to the
personal taste of the listener.

The Speakercraft component range offers a
comprehensive selection of treble, midrange
and bass drive units as well as a variety of
crossover networks. The units offer a high
degree of efficiency, above average power
handling, excellent appearance, and above all
good sound quality at a reasonable cost.

Each component comes separately packaged

with construction plans and a Speakercraft
booklet which gives all the information
necessary, including drawings, for building the
speakers. The booklet shows designs for
six speakers: the L60. L80. L100, E50, E70
and E90. They roughly approximate to
Wharfedale's new Laser series and the high-
efficiency, high-power E series.

'Speakercraft',

Wharfdale I Rank Hi-Fi),

Highfield Road,

Idle, near Bradford,

West Yorkshire.

Telephone: 0274 611131

(2064 M)

Digital I.R. meter

Tranchant Electronics Ltd. now have available
their new digital (micro-controlled) I.R. meter.
The N2700 is part of a new generation of test
instruments from Norfolk Industrial Elec-
tronics Ltd. and is designed primarily to
measure the insulation resistance of capacitors.

and has other IR and current measurement
applications. The N2700 utilises an auto
ranging Pico-ammeter to facilitate the dynamic
range of current required.

There are 5IR ranges and 4IL ranges. An
integral variable DC power supply is incorpor-
ated to provide the test voltage. A pre-charge
timer function enables the component under
test to be charged for a pre-determined time,
prior to measuring the I R or leakage current.

A pre-settable pass/fail limit may be
programmed into the N2700 for both I R and
I L. The soak time and indication of parameters
is by 3 digit displays: OIR/IL 2) Test V
3) SOAK TIME. The IR/IL display indicates
the pass/fail limit set, initiated by the key
pad. There are 5 Ohmic ranges: 0 to 200 M,
2 GST. 20 GST. 200 GST, 2 TST, and 4 leakage

10.00 pA. Test voltage is continuously
variable in 1 V steps, 2-1000 V DC. Maximum
output voltage is 1 KV @ 5 mA max., with
short circuit protection. Soak Time is 0-99
seconds.

Tranchant Electronics lu.k.) limited,
Tranchant House,

53 Ormside Way,

Redhill,

Surrey RH1 2LS

Telephone: Redhill <0737) 69217

(2061 M)

Signal generator

The LSG-16 is a wideband mains operated
signal generator with a frequency range of
100 KHz to 100 MHz |300MHz on har-
monics) over six positions. It has internal
modulations of 1 KHz or can be modulated
externally between 50 Hz and 20 KHz.
Crystal oscillator facility is also provided for
1 MHz to 15 MHz. The LSG-16 is housed in
an attractive professional case and com-
petitively priced at £55 plus VAT.

Sinclair Electronics Ltd.,

London Road,

Huntingdon,

Cambs PE 17 4HJ.

Telephone: 0480 64646

(2066 M)

September 1981 - 9-51

Wireless intercom

New from electronics company Eagle Inter-
nationel is a 2-channel FM Wireless Intercom.
Model Tl 250WA can be used conventionally
by means of 'call' and 'talk' keys or, by
simply depressing the 'auto' key, for complete
’hands-free' communication. Switching from
‘receive' to ’transmit' is achieved by built-in
voice ectivated circuitry. For optimum
performance the user should be 1 metre from
the unit, although It will work up to a

distance of 3 metres. LED indicators show

when the unit is in ‘transmit’ or 'receive'
mode. Volume and voice-trigger threshold are
adjustable and a channel selector switch
allows up to four units to be used in the
same vicinity, two on channel 1 and two on
channel 2. Transmission is through the mains
system thereby eliminating the need for
connection leads.

Eagle International.

Precision Centre.

Heather Park Drive.

Wembly HAO 1SU. Middx.

Telephone: 01-902-8832.

(2057 M)

Audio analyser

A Mk2 version of the Lindos LAI Audio
Analyser has been introduced — a compact
all-in-one test set incorporating a low distor-
tion oscillator, a digital frequency meter and a
comprehensive measuring section that can
check signal levels, frequency response,
weighted or unweighted noise, rumble, wow
and flutter and distortion.

Improvements over the Mkl include the
adoption of BNC sockets as standard, a larger
meter scale with 0.1 dB markings, an audio
band filter that meets CCIR468-2 require-

and the incorporation of both CCIR and I EC
'A' noise weighting curves as standard.

While the basic LAI will meet most require-
ments two other versions are now being
offered with the needs of the professional
user in mind. The LA1-P includes two extra
meter characteristics, true rmsand CCIR 468-2
Quasi Peak (for noise measurement). The

;7L j

• • O o 4 •

LAI -PI adds Peak Programme Meter charac-

teristics to the list, making it particularly well
suited to the needs of local radio stations and
others who have to meet the I BA and BBC
codes of practice.

All units operate from a single PP9 battery,

with a choice of three add-on backs to suit

different needs. The MAI is a simple mains

adaptor. The MA2 provides mains or re-
chargeable operation, audible monitoring, and
DIN connectors with channel switching for

quick connection to Hi-Fi equipment. The
ST1 Studio Interface provides mains oper-
ation. audible monitoring, and balanced
inputs and outputs with channel switching via
four PO Jacks (XLR's optional). It also drives
up to +26 dBm into 600 ohms.

Lindos Electronics,

Sandy Lane,

Bromeswell.

Wood bridge.

Suffolk IP 12 2PR.

England.

Telephone: Eyke (03947! 432

(2059 M)

Mains filter

H & T Components announce the introduc-
tion of a compact mains filter unit specifically
designed to meet international requirements
concerning the suppression of r.f. from, for
example, electronic ignition systems. The
Series 1179 mains filter unit may be ordered
from a variety of standard styles which offer a
choice of termination layouts. In this way.

the specific layout may be ordered which
meets the individuel wiring distribution of the
host equipment. The basic filter can be
supplied with or without a mains leakage
resistor, and has a current rating of 1 A
continuous or 3 A peak, at 240 V a.c., 50 Hz.
It has been appro^d by an independent test
house, and offers r.f. suppression over the
range 1 50 KHz to 300 MHz.

H & T Components.

Crowdy's Hill Estate,

Kembrey Street.

Swindon,

Wiltshire SN2 6BN.

Telephone: 0793 693681-7

Single dot temperature indicators
Thermindex announces e new and improved
range of single dot temperature indicators.
These selfadhesive dots are 10 mm in diameter
and cover the range 38°C to 260°C. Tempera-
tures are shown in both centigrade and
fahrenheit on each dot. They are packed in
booklets of 50 dots per booklet priced at
£3.50 per booklet, ex- VAT.

The dots change from silver grey to black

upon reaching temperature, the colour change

being permanent, thus affording a permanent

record of the temperature achieved.

Thermindex Chemicals and Coatings Ltd..

P.O. Box 112, Immex House,

Imperial Way,

Watford WD24JD.

Telephone: 0923 33477

(2065 M)

High voltage meter probe

Lightweight, with a built in meter, the
LHM - 80A from Leader measures up to
40,000 V DC simply and safely. With 20 KO
per volt input impedance, and an accuracy of
± 3%, the LHM -80A is an essential tool for
TV servicing, and many other applications.
Made of high impact polystyrene, the
LHM - 80A is 385 mm long, weighs 300 g.
and costs £16+ VAT.

%

Sinclair Electronics Ltd.,
London Road,

St. Ives,

Huntingdon,

Cambs PEI 7 4HJ.
Telephone: 0480 64646

(2067 M)

9-52 - elektor September 1981

Hybrid active filters

A range of precision low pass and high pass

available from Menvier Hybrids Ltd. The
modules are trimmed during manufacture and
require no external components. Customers
can specify the cut-off frequency in the range
50 Hz to 5 kHz and select a Bessel.
Butterworth or Chebishev response approxi-
mation. Amplitude and phase response
accuracy are better than 0.2 dB and 1 .0
degrees respectively.

Samples can be supplied within two weeks,
there are no tooling charges and quantity
prices are from £ 5.90. The modules can be
supplied as 16 pin DIL or 8 pin SIL packages.
Higher order filters can be supplied.

Menvier Hybrids Ltd.,

Southern Roed,

Banbury,

Ox on 0X16 7RX.

Telephone: 0295-56363,

Telex: 837829 MEN Vi

11871 Ml

LCR bridge

AIM Cambridge Ltd. have launched a new low
cost, automatic, digital LCR bridge priced
at £495, The LCR Databridge 401 measures

over eight decade ranges, claiming a basic
accuracy of 0.25% of reading. Measurements
can be made at either 100 Hz or 1 kHz and
either Series of Parallel equivalent circuit
values can be displayed on the instrument's
four digit display. Designed around a 280
microprocessor, the LCR Databridge 401 has
full auto-ranging facilities, with range lock,
and automatically distinguishes between
inductors and capacitors, without operater
intervention. An internal 2 volt bias supply
is provided for use in the measurement of
electrolytic capacitors. The LCR Databridge
401 is supplied with an integral 4 terminal
test fixture and digital outputs for the option
Limit Comparator.

AIM Cembridge Ltd.,

Burrel Road, Industrial Estate,

St. Ives, Huntingdon,

Cambs.

Telephone: (04801 65141

(1995 M)

Exponential voltage to frequency
converter

Specially designed for use in applications
where the requirement of a wide stable
exponential frequency range is of prime im-
portance the Aragom VF01 encapsulated
module also operates in a linear mode with a
minimum change of external components.

The operational frequency range lies between
0.1 Hz and 200 kHz with a present maximum
of 500 kHz. Accuracy in the linear mode is
better than 0.1% central to 6 decades and in
the exponential mode better than 0.1% over
5 decades. Power supply requirements are
t 15 volts. The module measures L.45x
W.30x H.21 mm with 0.040" diameter gold
plated pins on a 0.1 " grid pattern.

Turner Electronics L td..

Unit B2. Old Barn Lane.

Kenley.

Surrey, CR25AT.

Telephone: 01.668.0821

(1996 M)

Miniature LCD frequency meter
module

The Thurlby FM77T is a complete frequency

and costing under £ 20.

With no additional components the FM77T
will directly measure and display frequencies
up to 3999.9 kHz. With external pre-scaling,
this can be extended to 39.999 MHz or
399.99 MHz.

Stability is better than * 1 digit over a 10°C
to 30°C range and is defined by a built-in
crystal timebase. The display is a high con-
trast reflective LCD with 9 mm characters
with user selectable decimal points and
kHz/MHz legends.

For radio receiver applications, the user can

select any one of 23 pre-programmed standard

IF offset frequencies, enabling a reception
frequency to be displayed by measurement
of the Local Oscillator.

The FM77T operates from a single power rail
of between 4% and 7 volts and consumes
only 1 mA. Overall size is 2%"x 1%" x
and the display is mounted behind a textured

Price in the U.K. is £ 19.95
Thurlby Electronics Ltd.,

Office Suite 1,

Coach Mews,

The Broadway,

St. Ives,

Hutingdon,

Cambs. PE17 4BN.

England.

Telephone: (04801 63570

(2056 M)

Miniature mains switching relays

New AZ692/AZ693 series miniature mains
switching p.c.b. mounting relays with 4 kV
isolation and 8 mm creep and air gap, have
been launched by Zettler UK Division.

The skilful design of these relays combined
with carefully chosen materials give a relay
tMiich, whilst being compact and of sturdy
construction, is also inexpensive.

They are available with one changeover, one
normally open or one normally closed AgCdO

dard matrix affording a wide range of com-
patibility.

nation of output voltage and current, n-
the 601 2A convenient, cost-effective
capable of satisfying many different
quirements. For example, an engineer w
need a 20 V 50 A supply, a 40 V 30 A su|
and a 60 V 17.5 A supply to cover a r
similar to that of the 6012 A.

In addition to autoranging, the 6012A

Miniature connector

The Series 1 300 miniature snaplock connector
has been designed for applications in electrical
and electronic equipment in which vibration
or internal movement is commonplace. Both

reports and callsigns

required. The
when the key-

hooded for personal protection, and the free
plug incorporates twin barbed locks, one on
each side of its moulded plastic case. These
locks mate with the fixed, flanged socket thus
ensuring firm and secure mating. Disconnec-
tion is effected by light pressure being placed
on the two locking barbs.

Electrical characteristics of the Series 1300
connector includes current rating, per circuit,
of 5 A, operating voltage of 400 V a.c.
(maximum), proof voltage of 1.2 KV and
contact resistance of 5 milliohms (maximum).
The connector shell is moulded in grey Acetal
Resin, and contains a moulded phenolic insert
with six gold-on-nickel or silver plated brass
contacts. The positioning of the contacts is
such that mis-mating of the connector is
effectively prevented.

H St T Components,

Crowdy's Hill Estate,

Kembrey Street,

Swindon,

Wiltshire SN2 6BN.

Telephone: 0793693681-7

(2073 M)

Morse keyboard sender

The new Datong morse keyboard is intended
as a direct replacement for conventional or
electronic morse keys. Because hand move-

reduced compared with conventional keys,
sending fatigue is greatly reduced. Also a
built-in 16 character buffer memory means
that error-free morse of excellent quality can
be produced even by a beginner without the
need to develop a specialist skill which is of

The Datong keyboard is the first to use
micropower circuitry. This means that no
external power cable is needed. Instead, four
internal pen cells power the unit for typically
300 hours. Sending effort can be further
reduced by storing standard message routines
in the four 64-character memories. Pro-

batteries are changed.

The exclusive keyboard design combines the
benefits of conventional microswitches (1.3
mm travel, light click action) with a dust and
water resistant wipe-clean surface. The
surface is in fact a tough polycarbonate
membrane under-surface-printed in four
colours. Other features include: built-in
sidetone and "keypressed" pip tone; variable
weighting; speed ranges 5 to 33 and 20 to
132 words per minute; variable extra delay
between letters to ease message editing or to
allow use as a morse trainer; auto-repeat from
memory for automatic CQ calls etc.; an ex-
tensive character set includes punctuation,
accented letters and procedure signals;
"MERGE" key for making non-standard
characters.

Model MK is finished in matt black with panel
printing in white, brown, and red. Dimensions
are 10%"x6"x 1 %" (273 x 150 x 29 mm).
The price is £ 140 + VAT (£161 total I .
Detong Electronics Limited,

Spence Mills,

Mill Lane,

Bramley,

Leeds LSI33HE.

Telephone: 105321 552461.

(2058 M)

1000 W power supply

A new autoranging power supply with the
high performance characteristics and special
design features useful in automatic test
system applications has been announced by
Hewlett-Packard. The HP6012A gives system
designers who need a variety of fixed and
programmable power supplies, operating
freedom and flexibility in a compact, light-
weight and efficient unit. In addition to
filling the need for up to 1000 watts of power
in the lab. the power supply has special
interface and status feedback features for
automatic test systems. Typical applications
include semiconductor burn-in systems, PC
board test systems, and automatic production
processes. Option 002 provides a convenient

iiiiiiiiiiDiiiiiiiin

supply including mode and status indicators,
adjustable overvoltage protection, two 10-
turn potentiometers for high resolution
control, amplified current monitor terminals,
and voltage and current meters. A barrier strip
at the back of the supply provides the necess-
ary terminals for current monitoring, remote
programming and remote sensing.

Interface option 002 provides various features
to the system designer including remote
programming, status readback, remote shut-
down and output bias supplies.

Enquiries Section,

Hewlett-Packard L td„

King Street-Lane.

Winnersh,

Wokingham.

Berks RG1I5AR.

Telephone: (07341 784774

(2060 M)

Hexadecimal push index switches
Designed to satisfy the demand created by
microprocessor based equipment, the
Cosmocord 8000 series of 24 mm push index
switches has been extended to include a fully
hexadecimal version engraved 0 - 9 and A to F
as standard, and with 0 - 15 if preferred. The
Cosmocoder 8000 series offers a wide choico
of output codes including decimal and binary
coded decimal - with or without complement.
Push mounted from the front, these snap-
together switches offer direct visual indication
of the last input, eliminating errors associated

with keyboards. The standard case colours
are black or grey with large, easy to read
engraved white letters on a black wheel.
Manufactured in impact resistant Acetal and
polycarbonate, the switches have contacts of
silver/gold alloy on beryllium copper, offering
a switching capability of % amp and
200 V rms within the limits of 24 watts per
switch.

Cosmocord Limited,

Eleanor Cross Hoad.

Waltham Cross,

Herts.

Telephone: Lea Valley 716666.

jrammed pauses can be inserted anywhere in

(2075 M)

Barkwav Electron!

Intercom Unit to complement their micro-
computer-controlled POLYDEX System.

The new, ergonomically designed unit, with
special key labout conforming to inter-
national telecommunications standards, is
suitable for desk use or wall-mounting and is
available with handset.

The Barkway unit offers considerably
advantage over other manufacturers' dual-
purpose units which are, at best, a com-
promise. because the need to make them
compact hampers audio output and sound
quality capable of a larger unit specifically
designed for direct speech.

Barkway's experience in supplying many
systems throughout the world indicates that
no more than 2-3 percent of direct speech
customers use handsets or dual-purpose units
in the handset mode.

The POLYDEX-System gives a wide range of
facilities to direct speech intercom, including:
automatically call-back; call transfer; sec-
retarial transfer; automatic priority — which
enables a caller to cut into an engaged
extension; interfacing with a paging system or
mobile radio hook-up and a choice of up to
200 stations on 14 channels.

Barkway Electronics Limited.

Barkway. Royston,

Hertfordshire SG8 8EE.

Telephone: 1076384) 666

arranged in a forklike configuration, form an
air gap through which a soft iron vane can be
pass as a switching medium. The open col-
lector of the Hall amplifier conducts (to a

air gap. When, however, the vane leaves the air
gap, the collector output remains blocked
until the vane reappears in the air gap.

Because of this static method of operation,
there is - unlike with, for example, inductive
pick-ups - no lower limit to the working
frequency. The output signal shape is in-
dependent of working frequency. The new
Hall effect magnetic vane switch also features
integrated overvoltage protection, which
provides the circuit with a reliable safeguard
against voltage peaks which occur in the
electrics of motor vehicles.

The HKZ 101 is also ideally suited for many
industrial applications, e.g. as a limit switch,
revolution sensor or position sensor for speed
measurement and for sampling coding disks.
Since the output stage is in the form of an
open collector, LEDs or relays can be directly
driven; levels can also be matched to sup-
plementary circuits.

Siemens Limited,

Siemens House.

Windmill Road,

Sunbury-on-Thames.

Middlesex TWI6 7HS.

Telephone: 109327)85691 Ext. 250

Versatile LED bar indicators

The Sharp GL Series of LED-based bar-graph
indicators from Impectron Ltd. consist of
five, seven or twelve LED segments in a
choice of colours. There are seventeen
different types, all designed for PCB or panel
mounting offering a range of colours, sizes

There are four basic colours: red, yellow,
yellow-green and green, and as well as single
colour arrays the user can specify many
colour combinations.

Overall dimensions of the five segment
versions are 40 x 8 x 7 mm (excluding pins)

LEDs. Impectron can also supply integrated
circuit drivers for all indicators in the Series,
allowing high impedance interfacing to most

Intelligent display module

The DS240 is an intelligent display module
comprising a 2-llne by 40-character dot-matrix
vacuum fluorescent display tube together
with all the necessary drive circuitry. The unit
is designed for easy incorporation into desk-
top computers, point of sale terminals, etc.
An on-board microprocessor makes system
interface a simple matter, two types of
parallel interface and serial (V24) interface at
8 different band rates being available as
standard.

The unit also features a 'self-test' mode to aid
system fault tracing. The display is fluorescent
blue-green, which may be filtered to blue,
green, yellow, pink, etc. (at reduced light
output). Characters are generated under
software control and therefore custom and
special characters are possible. The standard
unit produces 96 standard ASCII characters,
plus 32 non-standard characters. Characters
presented to the unit greater than 9 F (Hex)
produce the standard ASCII characters but

The DS240 measures 10" x 2.3" x
costs £346.1 5 (10 off).

Delpak Electronics Ltd.,

13 Hazeibury Crescent,

Luton LU1 IDF.

Telephone: 0582 415832

,•••••

w m our •

CATALOGUE •

320 1 big pages packed with •

V data and pictures of ^
m over 5.500 items •

••••••

Over 100,000 copies sold already!
Don't miss out on your copy,

On sale now in all branches
WH Smith priced.

In case of difficulty check the coupon below.

* Same day service on in stock lines

* Very large percentage of our stock lines in stock

* All prices include VAT

* Large range of all the most useful components

* First class reply paid envelope with every order

* Quality components-no rejects— no re-marks

* Competitive prices

* Your money is safe with a reputable company

On price, service, stock, quality and security it makes
sense now more than ever to make fWIHKl! your
first choice for components every time!

Post this coupon now.

Please send me a copy of your 320 page catalogue I enclose £1.25
(mcl 25p pSp). If I am not completely satisfied I may return the
catalogue to you and have my money lefunded. If you live outside the
U K. send £1.68 or 12 International Reply Coupons.

All mail to: P.0. Box 3, Rayleigh, Essex SS6 8LR. lei: Southend (0702) 554155 Sales: (0702) 552911