Hand-held
frequency counter

up to 40 MHz.

Mini organ

with maxi sound.

Metal detector

sophistication in practice.

selektor

metal detector

The metal detector featured in this article is very professional in appear-
ance and operation and performs better than many high-priced commercial
detectors. It is both highly stable and sensitive. The construction is very
simple since all the obstacles usually associated with this type of project
have been removed.

. high boost

1A.M. Bosschaert)

The average tone control circuit in an electric guitar usually consists of
little more than a capacitor and a potentiometer and can hardly be ex-
! pected to produce very good results. An active tone control is much more
effective and this high boost circuit can either amplify or attenuate the
treble within a range of 35 dB.

new synthesiser

The success of the Elektor Formant synthesiser led us to the opinion that
there is a great interest among our readers in the field of electronic musical
instruments, especially synthesisers. The new Elektor synthesiser is of
modular construction and can be expanded into a polyphonic instrument
with 'programming' facilities. This, the first article in the series, explains
the basics behind the design.

LCD frequency counter

This is the first in a series of projects featuring a frequency counter mod-
ule with a liquid crystal display. The high performance is out of all pro-
portion to the simplicity of the circuit. Two switched ranges are available,
the first up to 4 MHz for use in monitoring frequencies in microcomputers
and the second up to a maximum of 35 MHz to cover CB transceivers and
, general use.

telescope control

I A camera mounted on a telescope can, with a long enough time exposure,
enable the more distant stars to be observed. The circuit in this article
enables the telescope to track the star accurately for the period of the
exposure.

solar powered receiver

The design described in this article is a low cost portable receiver that can
1 be powered with surprisingly few solar cells.

mini organ

Thanks to ITT's special organ 1C the electronics for the mini organ can
easily be built in an afternoon. It uses a full size keyboard and is poly-
phonic. Its performance is so good that it really must be seen to be be-
j lieved.

telephone amplifier

A pound for a minute seems a lot of money to hear Granny's faint voice
ten thousand miles away and then not understand a word she's saying.
This telephone amplifier provides a solution and enables the whole family
to listen in to the conversation.

sine-wave oscillator

' (L. Boullart)

I This particular design is not at all complicated as far as construction is
* concerned and yet it boasts a distortion level of only 0.01%! Its fre-
quency range extends from 10 Hz up to an inaudible 100 kHz.

i Teletext decoder part two

: This article deals with the practical side of the Teletext decoder. Particular
attention is paid to the modification of a TV set if the decoder is to be
! built in; calibration procedures; and, last but not least, the instructions
for use.

Speak to your computer

The removal of the key pad interface between man and computer has long
been the pipedream of every programmer. When using the M 6800 system
| it can become reality.

market

UK EDITORIAL STAFF

T. Day E. Rogans

G.H.K. Dam
E. Krempelsauer
G. Nachbar
A. Nachtmann
K.S.M. Wal raven

advertisers index

UK-22

/AT XS and CS without plug 240. 115. 50.24or
VAT SEE THE NEW ANTEX SOLDERING IRONS

SUPPLIERS I

ANTEX ELECTRONICS LTD.

Mayflower House. Armada Way. Plymouth. Devon PL1 1JX.
Telephone (0752) 667377. Telex 45296

WITH

BUILT-IN

FINGER

PROTECTOR/GUIDE
AND FUSED -

MAINS PLUG //A

ANTEX- THE 'HOT TIP' FOR SOLDERING

a zery special magazine
for a i/ery special price ]

Increasingly in demand, amongst
electronics hobbyists and professionals !

A 1982 subscription delivered direct to your door for £ 2.00 less
than last year - saving £ 1,30 on the cover price.

That’s what we call a real offer !

Complete the Elektor Readers Subscription and order card, and your cheque/P.O. for £ 6.50.

It's probably the best deal you'll make in 1981

Because of steadily increasing postal charges this offer can only apply to the U.K. Mainland. Subscriptions overseas
by post remain the same as last year! Even though the cover price and postage have increased.

We hope you all agree that this is fair to everyone.

P.S. If you live overseas and really can't wait to get your hands on Elektor we will rush you a copy every month by
airmail for £ 17.50. Wherever you live don't forget to order your 1982 Binder.

selektor

The electromagnetic spectrum
crisis

We have only to listen in on the broad-
casting bands at night to notice how
some stations appear to be jumbled up
on top of one another. Though local
ones can be received quite well, there is
considerable interference on the distant
transmissions, for the region of the
electromagnetic spectrum devoted to
radio, necessarily very small, is every
day becoming more congested. It is,
therefore, a finite resource, upon which
ever growing demands are made. This
problem could be called the electro-
magnetic spectrum crisis, and in some
ways it is more acute than the so-called
energy and materials crises which have
received so much attention.

To explain the difficulty that the devel-
oped world is certain to have to face
over the radio spectrum, it is necessary
to explain the scientific background. To
radiate or receive a radio wave it is
necessary to use some kind of aerial
structure, or antenna. This structure is
necessarily related to the length of the
electromagnetic waves: ideally its size
should be at least a substantial fraction
of the wavelength. If it has to be made
smaller, efficiency declines rapidly, and
even a powerful transmitter connected
to it would produce only a feeble radio

The wavelength is given by dividing the
velocity of light by the number of times
that the transmitted wave oscillates in
a second, that is, its frequency. The
velocity of light is approximately
3 x 1 0 8 metres per second, so a wave
with a frequency of 10000 cycles per
second (hertz) has a length of about
30 km. Obviously, an aerial comparable
in size with this would be very unwieldy
and suitable only for a very large land
installation. Because of this, frequencies
below 10 Hz are of little practical use
for radio purposes; it is a limit set by a
physical law which is likely never to be
circumvented.

The upper limit is just as rigid. The
main attraction of radio waves for
communication is their great penetrat-
ing power. They are little absorbed by
the atmosphere, so are able to propagate
over long distances with only a small
loss. Unfortunately this ceases to be
true for frequencies of the order of
20000 million hertz and above. At such
frequencies absorption of radio energy
by the atmosphere is already consider-
able and it rises rapidly at higher
frequencies.

So we are trapped, for most com-
munication purposes, in this radio

'window'. Obviously, the usable radio
spectrum is a natural resource of quite
fixed extent. That is one side of the
spectrum congestion problem; the other
is the width in the electromagnetic spec-
trum taken up by each radio trans-
mission.

Modulation

All radio transmitters use a technique
known as modulation to impress the
intelligence to be transmitted on to the
radio wave. An unmodulated radio wave
simply consists of an oscillation at a
particular radio frequency. The only
information that it conveys is the fact
that it is present. To communicate we
must cause it to vary in some way. The
simplest thing that we could do would
be to keep turning it on and off in the
familiar Morse code, the way the
earliest transmitters were used for tele-
graphy. Usually, however, we wish to
transmit the human voice or other
complex signals, and to do this we must
alter the radio wave in accordance with
the characteristics of the signal wave-
form that we wish to convey.

Most existing radio transmitters modu-
late the carrier wave in one of two ways.
They either vary its amplitude in sym-
pathy with the modulating signal, a
process called amplitude modulation
(AM), or they vary the frequency of the
wave, a technique known as frequency
modulation (FM).

Although the carrier wave was initially
on a fixed, single frequency, as soon as
it is modulated energy begins to be
radiated at other, additional frequencies.
What happens is that two new sources
of radio energy appear, called sidebands,
which are seen above and below the
carrier frequency and each separated
from it by an amount just equal to the

t

I vv\/W w W\/\/W w ^

Figure 1. The amplitude of a modulated
carrier wave (al varies at a rate corresponding
to the frequency of the modulating signal it
carries. This complex wave is the sum of the

frequency f c , and two side components at
frequencies equal to the sum and difference
of the carrier frequency and the modulating
frequency. f a . When the modulating wave
itself is complex, as in speech for example,
the side components spread out over bands of
frequencies on each side of the carrier, each
band corresponding to the band of
modulating frequencies. They are known as
the upper and lower sidebands.

elektor november 1981-11-01

modulating frequency. For AM the
width of the transmission in the spec-
trum (we call it the bandwidth) is just
equal to twice the modulating fre-
quency. Radio frequency allocation is
based on each transmitter radiating in
an assigned channel which in practice
should be a little wider than its band-
width.

Theoretically, an AM transmitter should
radiate nothing else but the carrier and
the two sidebands and therefore have a
spectrum entirely contained within a
bandwidth equal to twice the highest
modulation frequency. But in practice
AM transmitters are not quite perfect
and therefore do radiate some ad-
ditional energy, which we call spurious
emission, generally at other frequencies
further removed from the carrier fre-
quency.

In the case of frequency modulation,
the effect is to crowd up some of the
waves in time and spread others out, but
the intensity of the wave remains con-
stant. But the FM wave has a much
more complicated spectrum than the
AM wave and there are, in theory, an
infinite number of separate spectrum
lines corresponding to individual fre-
quencies at which energy is radiated.

In this case we might be forgiven for
supposing that the bandwidth of an
FM wave is infinite. Strictly speaking
that is true, but the energy contained
in each of the very large number of side-
bands from the FM transmission de-
creases very rapidly with increasing
difference in frequency from the
carrier, so we can regard the bandwidth
as being finite for practical purposes.
The mathematics needed to calculate
the bandwidth of an FM transmission
is complicated and depends in particular
on how much the frequency is shifted
by the modulation; the shift is called
the frequency deviation. Here, too, the
total bandwidth occupied can be shown
to be a factor multiplied by the modu-
lating frequency, but that factor has a
value always greater than two, which it
would have been for AM. So FM always
has a wider bandwidth than AM.

Television

As far as the modulating frequency is
concerned, in the case of human speech
it is not necessary to transmit signal
frequencies above 3 or 4 kHz, de-
pending on the quality of voice that is
required, but for music a good quality
transmission should extend to perhaps

11-02 - elektor november 1981

15 kHz of modulating frequency. With
AM the minimum bandwidth for a
voice transmission might be 6 or 8 kHz
and for a high quality music channel at
least 30 kHz. Using FM these band-
widths are substantially greater; for
example, the present BBC VHF/FM
transmissions have a bandwidth of some
200 kHz. Some kinds of signals even
demand very much greater bandwidth
than this, and the greatest sinner of all
is television.

Transmitting a good quality television
signal can mean using a signal frequency
of up to 4000 or 5000 kHz, depending
on the technical details, and for this
reason a relatively wide channel is
taken up. In open broadcasting FM
is never adopted for television; even
AM, used in the earliest television
broadcasts, gives a transmission of great
width in the spectrum. So television is
nowadays transmitted on a modified
form of AM.

There is an element of redundancy in
both the AM and FM transmissions, for
they are symmetrical about the carrier
frequency and therefore the essential
information conveyed by the modu-
lation is represented twice, once above
the carrier and once below. This led to
the evolution of two closely related
alternative forms of modulation, known
as single sideband (SSB) and vestigial
sideband (VSB). Both are derived from
AM. In SSB the carrier and one of the
sidebands generated by the amplitude
modulation process are totally sup-
pressed and only the other sideband is
transmitted. Provided a carrier wave of
the correct frequency can be generated
at the receiver, all the information
needed to reconstruct the original
modulated signal is still present in the
transmission, because the magnitude of
the signal is represented by the side-
band amplitude and its original fre-
quency by the difference between the
sideband and carrier frequencies. In
theory, single sideband transmissions
occupy only half the bandwidth of full
AM transmissions without the loss of
any information.

Vestigial Sideband

SSB is now extensively used in military
and commercial point-to-point and
mobile radio systems. But for television
the alternative, vestigial sideband system
(VSB) is preferred, in which the carrier
and one complete sideband are left in
the transmission but the width of the
other sideband is restricted to accom-
modate only the lower modulating
frequencies. This is marginally less

economical in terms of bandwidth, but
makes for cheaper design of transmitters
and receivers. Obviously, for mass-
produced television sets this is a con-
siderable advantage and the small band-
width penalty is not significant. Using
techniques of this kind it is possible
to transmit a satisfactory television
signal in 6000 kHz or so, but it is still
worth remarking that we could have a
thousand speech channels for every
television channel.

In addition to broadcasting, the radio
spectrum is used for radiotelephones,
for mobile communications with air-
craft, ships and vehicles, for point-to-
point transmissions within countries
and on the transcontinental scale, too,
and for services such as radar and radio
navigation which play a vital role in
our modern society. Given that the
resource is finite, it is clear that there is
a limit to the number of users who can
simultaneously exploit the economic
advantages or radio systems.

Does this matter? After all, the world
existed for a long time without radio.
Will it be a very serious disadvantage
if the growth of radio is shackled by
spectrum congestion? It is obvious that
although there are some areas where the
use of radio might be described as a
luxury, there are others where its appli-
cation is absolutely unavoidable. For
example, it is very difficult indeed to
imagine airlines running without radio
communication, and the policing of our
cities could hardly continue if the use
of radio were denied. In fact, a feature
of the economic development of the
whole western world in recent years has
been the growing use of radio. All the
parameters of economic growth in our
society correlate very closely with that
growth, and it is difficult to foresee that
our economy could continue to develop

"t

Figure 2. In frequency modulation, the

while the amplitude is kept constant. An
analysis of the complex modulated wave
shows, in theory, an infinite number of
separate spectrum lines in addition to the
carrier, corresponding to individual

However^ the energy they contain falls off
fairly rapidly on either side with the

if the expansion of radio stopped. In
particular, if we face an era in which
energy costs are going to rise, the use of
vehicles will have to be more efficient,
and mobile radio is the most important
resource for making this possible.
London's police force is little stronger
in numbers than it was in the 1920s, in
spite of an enormous growth in popu-
lation with a corresponding increase
in crime and in road users. Senior police
officers have no doubt that the ability
to go on policing the region effectively
depends mainly on the use of radio in
cars and of personal radios by individual
officers.

Where public radio paging and car
telephone services are available, they are
showing steady rates of growth, which
means people must think the cost is
justified, while private mobile radio is
playing a vital part for many companies
in enabling them to run their business
more efficiently. Studies have shown,
for example, that commercial delivery
services can make savings of some
20 per cent in fuel and in use of vehicles
by introducing radio. Taxis are now
largely dependent on it, the growing
number of ships in the world are
entirely dependent on it and new cheap
air fares may be expected to accelerate
the already rapid growth of air travel
and thereby complicate problems of
radio-dependent air traffic control.
Already it has been necessary to halve
the width of aircraft radio channels to
ease congestion.

Then there are areas where the use of
radio is just beginning: the UK Trans-
port and Road Research Laboratory has
shown that radio can reduce the costs of
town and country bus services by 20 per
cent and reduce passenger waiting time
at stops by 16 per cent.

Safety of life

It is difficult to quantify the economic
advantages of any one of these appli-
cations; emergency life-saving radio
services cannot be costed at all. And
there is the social value to consider, for
good or ill, of radio and television
broadcasting. Clearly the public at large
puts a high value on such services.
Inevitably, we must conclude that it is
a matter of the utmost importance, for
both economic and social reasons, to
provide radio services as freely as pos-
sible to those who can make good use
of them. So we are faced with a very

1981 - 11-03

difficult problem of spectrum manage-
ment. What are the rules for making
the best use of our electromagnetic
heritage? I believe that they fall into
two categories, avoidance of pollution
and right organization of use.

We are all familiar with the irritating
interference created by electrical
machinery. Everybody knows that a
good deal can be done to reduce this
nuisance by treating the offending
equipment and nowadays there are
quite stringent laws in many countries
about fitting interference suppressors.

A worse pollution hazard probably
comes from communications equipment
itself, which is capable of interfering
with other users and cannot be sup-
pressed easily, for by its very nature a
radio transmitter must be able to
radiate energy efficiently and is likely
to spread spurious radiation where it
is not wanted. As the spectrum gets
more congested we may expect that
the strict regulations about spurious
emission made by radio licensing
authorities will be tightened further.

Minimum Energy

More fundamentally, all energy trans-
mitted is capable of causing interference,
under some circumstances, with other
radio users. Often energy from two or
three transmitters combine in the
receiver to produce a spurious response
which destroys the intelligibility of the
wanted signal. It is not good enough
simply to radiate a very 'clean' signal
from the transmitter, for even the
energy which forms part of the wanted
transmission is itself capable of creating
a polluting effect for other users.
Another cardinal principle must there-
fore be that in all cases the energy
radiated should be the minimum needed
for the job the system is designed to
do.

We must recognize that radiating a
substantial carrier wave in AM or FM
transmission does pollute the spectrum.
All the whistling sounds produced by a
radio receiver in a congested band are
caused by carrier interactions, which are
more serious in their effect than inter-
actions between energy from sidebands.
Because the carrier wave is of constant
amplitude and frequency, and the
person receiving a signal knows its fre-
quency in advance, in principle at least
it ought never to be necessary to radiate
the carrier at all. In practice, it may be
costly to generate it accurately enough
in the receiver, so it is quite common-
place to simplify the receiving equipment
by radiating a substantial but reduced

carrier along with the sidebands. In a
typical AM transmitter modulated by
speech, on average 90 per cent or more
of the radiated power is in the carrier.
This means that less than 10 per cent of
the radiated power is performing an in-
escapable communication role and the
rest is needlessly generating spectrum
pollution, and is retained only to make
the receiving equipment cheaper. This
may have been justified in an era when
the spectrum was relatively uncluttered,
but today it should be looked into very
closely.

Single-sideband modulation, with a sup-
pressed carrier, is, as we have seen, a
perfectly feasible means of transmission,
already in widespread use in the short-
wave (HF) bands. Not only does it cut
the width of the transmitted spectrum
by about half, but it avoids emitting
radio energy other than the minimum
necessary to establish the required
communication link. It is therefore our
best ally in the fight against spectrum
congestion. Recently in the United
States the Federal Communication
Commission gave wide circulation to
a document — the UHF Task Force
Report — which strongly supported the
extension of SSB into the VHF and
UHF bands for land mobile radio.
Surely this must be a straw in the
wind.

Costs

The disadvantage of SSB is the tighter
specification which has to be placed on
the equipment, which is thought likely
to be considerably more expensive.
Against this, present trends in elec-
tronics are bringing equipment costs
down so rapidly that perhaps the han-
dicap will prove much less than had
been imagined.

Probably it is in two-way radio systems
rather than broadcasting that SSB will
first capture a new empire for itself, and
my colleagues and I at the University of
Bath have been able, thanks to generous
support both from the UK Home Office
Directorate of Telecommunications and
the Wolfson Foundation, to conduct an
investigation into the use of SSB modu-
lation in the land mobile radio service
at VHF and UHF. As a result we are
convinced that operation in channels
only 5 kHz wide is perfectly feasible,
as against the 12.5 kHz or 25 kHz
channels now used for AM and FM.
Because carriers are not transmitted,
the total energy radiated is reduced by
a factor of about eight when speaking,
and to a very low level in between
words. To our surprise it appears that
the cost of the equipment may be very
little more: the saving from the reduced
transmitter power rating tends to offset
increased costs elsewhere.

If we exploit all possible technical re-
sources, the problem of exhaustion of
the electromagnetic resource can be
postponed, at least for a generation.

elektor november

Should we fail to take sensible steps,
things could only worsen. The economic
toll already is certainly not trivial, and
deaths have occurred through the
congestion.

It is, of course, true that spectrum
exhaustion is not like exhaustion, say,
of hydrocarbon fuels. The spectrum is
not used up or worn out by use, and
we can always free it again by simply
turning off the transmitters or inter-
ference sources. Though this is true, it is
not a helpful observation. Established
patterns of radio use have led to large
capital investments which cannot lightly
be put to one side. To take an example,
radio equipment in the hands of UK
police forces is now perhaps worth some
£20 million sterling. In theory they
could change to an entirely new radio
system within days of having new equip-
ment provided for them, but in practice,
and quite rightly, only the most com-
pelling reasons would cause this large
capital investment to be abandoned.
Much the same is true of every other

This means, first, that changes will have
to be introduced in an evolutionary
way, as equipment is inevitably replaced
through age and obsolescence, and,
second, that because there is this large
inertia we must be particularly careful
that new systems coming into use do
not offend against the needs of spec-
trum conversation. If any such systems
were inadvertently allowed to be
installed, the chance of phasing them
out again quickly would be small indeed.
We should now be directing some of
our best engineering talent to spectrum
conservation. It would be foolhardy if
we were not to insist on the highest
attainable technical standards in the
design of all our future communication
systems that use radio. It is high time
that we stopped trying to reassure
ourselves that the spectrum congestion
problem does not exist or will go away.
It is with us now, and it threatens the
future of us ail. We must learn habits of
frugality, and who can not say that if
we do so in this field the gift for it may
yet spread to others?

By Professor W. Gosling,

University of Bath.

Spectrum 164.

(707 S)

metal detector

high performance and simple construction

One of the problems that govern the fast growing hobby of treasure
hunting is that economical metal detectors are notoriously unstable
and suffer from a lack of sensitivity while the good ones are very
expensive. The design featured in this article is both highly stable and
sensitive and presents no problems in construction. It is professional in
appearance and operation and performs better than many high-priced
commercial detectors.

One of the projects near the top of our
'most requested projects' list has, for
some time, been a metal detector. The
reason for the delay in publishing an
article on the subject is due to the fact
that construction of a good detector
(and not one that just detects some-
thing) is a very difficult proposition.
The very first major problem that rears
its head is that of stability since this is
dependent on a number of aspects. To
achieve a stable circuit design is not easy
when cost is taken into consideration.
The design and construction of the
search head is also a formidable obstruc-
tion to obtaining a high performance
while still remaining within the capa-
bilities of the average enthusiast.
Furthermore, the search head must be
robust and remain stable after the very
critical setting-up procedure it requires.
In other words, it must not be prone to
'headaches' when knocked. Even the
difficulties of waterproofing and resist-
ance to temperature changes are
problems enough in themselves if the
others are not enough.

What of the circuit itself? There are of
course, a number of methods by which
metals can be detected with the use of
electronics but stability is the governing
factor in performance. It is possible to
design a circuit using microprocessors
and the like that will work wonders but
will require a find of gold sovereigns
every day for a week to cover its cost.
Overall, the difficulties will appear by
now to be unsurmountable and you
may well be asking how we managed to
do it at all. It took a little time but, rest
assured, it has been done and the results
have been well worth waiting for. The
Elektor metal detector performs very
well and is easy to construct due to the
fact that the ultimate horror - the
search head — does not have to be con-
structed by the reader. A complete
hardware assembly, including the head,
is available to readers, ready-made.

How has it all been done? Before
reading on to find out, a note of
warning. Firstly, reading the article
carefully and paying attention to detail
during construction are essential to
produce a high performance metal
detector. Secondly, and not quite so
obvious, treasure hunting with any
metal detector is much like horse riding,
it takes time and practice to become
proficient at it. Don't expect it to
happen in one dayl

Surveying the field

Various methods are used in the design
of metal detectors and they all have
their various advantages and disadvan-
tages.

BFO - beat frequency oscillator

These units are cheap, easy to build, but
suffer due to the high frequencies used
(for economic reasons). They are not
very sensitive to metals, suffering badly
from instability and ground effect and
they do not discriminate between differ-

metal detector

Figure 1 . A simplified diagram of a
conventional oscillator circuit. No form
of compensation is included to allow
for changes in the velue of capacitor C
due to temperature venations.

ent materials. They work on the hetero-
dyne principle of beating two fre-
quencies together and obtaining an
audio tone relative to the difference
between the two signals. The basic trick
is that the oscillator has the search head
as its frequency determining component
and inducing a metal object into its
vicinity will increase or decrease the
search frequency so producing a change
in audio tone.

TR/IB - transmit -receive/induction-balance

These metal detectors are much better
than 8FO types as regards general
sensitivity but still suffer from stability
problems and do not discriminate be-
tween metals. This type requires very
accurate relative positioning of the coils
and can be extremely difficult to set up.
Another disadvantage is the fact that a
slight knock 'on the head’ can cause
false signals.

VLF - very low frequency

The most expensive metal detectors are
usually of this type. Various circuits
will: (a) make the unit sensitive to
metals, (b) discriminate between metals
and (c) provide ground effect control
thus overcoming a lot of shortcomings
inherent in cheaper detectors. However,
due to the low frequencies used, these
types still suffer from instability. The
disadvantages are cost (which can be
high) and difficulty in setting up the
head assembly.

PI - pulse induction

Pulse induction is used in professional
and industrial metal detecting systems.
A good design can be extremely costly
and construction and calibration is
beyond the realms of the average elec-
tronics enthusiast.

The Elektor metal detector to be de-
scribed is a high performance VLF
model utilising phase locked loop
techniques for stability, good discrimi-
nation between metals and elimination
of ground effect. It features a better

elektor november 1981 - 11-05

2

Figure 2. The Elektor metel detector employs a phase locked loop design to provide
the basis of a highly stable oscillator with immunity to drift.

detection range than BFO and TR/IB
types and does not suffer from insta-
bility problems. To achieve all this
'icing on the cake' it will be obvious
that the circuit must therefore be more
complex. However, a glance at the cir-
cuit diagram in figure 4 will show that
it is not as complicated as some readers
may have expected, and furthermore
calibration is extremely simple.

Instability and PLL

As previously mentioned the major
disadvantage of cheaper metal detectors
is that of instability. In cases of severe
instability, the process of detecting
metal can be hampered to the degree
that it becomes impossible to differen-
tiate between the detector itself being
'off tune' and a 'find'.

Why is the Elektor VLF circuit more
stable than others? The simplified cir-
cuit diagram in figure 1 is that of the
more conventional VLF design. The
problem is that the low frequencies used
with a VLF design require a large num-
ber of turns in the search head coil. This
also necessitates that the value of C in
the circuit is relatively large. This is the
point in the circuit where instability will
occur since any capacitor will change
value when exposed to heat. If it
happens that the capacitor is included
in the tuned circuit of an oscillator, it
follows that the oscillator frequency
will therefore be prone to drift.

In this type of oscillator the drift
cannot be compensated for and the
problem is further aggravated by the
susceptibility of the circuit to voltage
fluctuations in the power supply
(creating more instability). As if this
were not bad enough, the oscillator has
a fairly high current-consumption and
therefore batteries have a very short
life.

To overcome these shortcomings the
Elektor phase locked loop design was
developed. This effectively compen-
sates for any drift in the oscillator
frequency due to changes in the value

of the capacitor. This design is slightly
more complex but with the help of
figure 2 is fairly easily understood.

The centre frequency of the VCO
(Voltage Controlled Oscillator) is
initially set by Ra, Rb, Rc and Ca, Cb.
This produces a square wave, which is
converted into a sine wave by the
amplifier A1 and is fed to the search
head LI. A proportion of this signal is
capacitively coupled back to the input
of the phase comparator PC, the output
of which is passed to the control input
of the VCO via the low pass filter con-
sisting of Rc and Ca. The output of the
VCO is also fed back to the other input
of the phase comparator. These two
feedback loops provide an immunity to
drift and form the basis of this highly
stable oscillator.

The Elektor metal detector

It has already been stated that the cir-
cuit of the metal detector is fairly
complex. However, the diagram in
figure 3 shows the details of the circuit
in block form and it will be understood
more easily if each block is treated
separately, bearing in mind that each
one refers to a section of the main cir-
cuit in figure 4.

Since the first block of the diagram, the
PLL oscillator, has already been de-
scribed, we can move on to the phase
locked loop discriminator. A discrimi-
nator enables the user to reject
unwanted materials, for example metal
paper foil. This part of the circuit has
again been designed with high stability
in mind and for this reason a second
PLL is used. The adjustment of the
phase relationship between the transmit
oscillator and the received signal,
provides a method of sampling a pro-
portion of the received signal. This
provides the phase sensitive detector
with its gate input.

The phase sensitive detector 'chops up'
the input signal from the receive head
and, after filtering, the DC voltage out-
put is used by the meter and audio

1 1-08 - etoktor nowntur 1981

circuits.

The auto tune is another special feature
of the Elektor metal detector. Briefly
the principle is to achieve an offset
voltage, by storing it in a capacitor
across an FET opamp which is used as
slow/fast integrator. By feeding this
voltage level back to the phase sensitive
detector we can reset the output voltage
of the PSD, in order to zero the meter
at the flick of a switch. This means that
manual retuning is not necessary when
changing modes or altering the sensi-
tivity of the detector.

The meter circuit is straightforward and
incorporates a battery status check, as
well as indicating the accepted and
rejected finds.

The final section of the block diagram is
the audio stage, but this does not con-
sist solely of an ordinary amplifier.
A gated chopper circuit is also included
which takes the search oscillator fre-
quency and divides it down to approxi-
mately 270 Hz. This is used to chop the
DC voltage output of the PSD before
feeding it to the audio output amplifier
thus providing an audio tone when
metal is received.

This covers the basic building blocks of
the metal detector and we can now
move on to discuss the finer points of
the circuit in detail.

The circuit diagram

The PLL oscillator is formed around
IC1 in the circuit diagram shown in
figure 4. The frequency is set by the
components C3, R12, LI (the search
head), C2 and RIO. The value
of RIO is chosen to set the
VCO initially at mid frequency.

The square wave output of the
VCO (pin 4) is coupled via R7 to the
search amplifier which is formed by
transistors T1 . . . T4. A portion of the
sine wave is fed back via Cl and R1 1 to
the signal input of the phase comparator
(pin 14) where it is compared to the
signal at pins 3 and 4. The resultant out-
put at pin 2 is used to adjust the VCO
to its resonant frequency thus forming
a stable oscillator.

The VCO output at pin 4 of IC1 is also
fed to the input of the phase locked
loop discriminator IC2. The configur-
ation of this 1C provides a convenient
method of adjusting the phase relation-
ship by simply changing the value of a
resistor at pin 11. The actual 'resistor'
value is selected by SI, P4 and P5 and
the resistor networks connected to
pin 11 pf ES5 and pin 8 of ES6. The
level at pin 1 1 of IC2 selects either
ground effect or reject mode with the
aid of the CMOS switches ES5 and ES6.
The phase shifted square wave output of
the discriminator IC2 is the gate signal
for the phase sensitive detector formed
by IC4 and A2. The receive signal from
the search head is passed through an
impedance converter T12 and then
amplified (by a factor of 50) by the
opamp A1 before being passed to the
phase sensitive detector. It is this signal

al detector. Each block cc

Photo 1 . The hi

that is then chopped and sampled by
IC4 and converted to a DC voltage level
by opamp A2. The gain of this stage is
variable by means of potentiometer P6.

The output of A2 is a voltage level that
varies in relation to the size and content
of any metal object in the proximity of
the search head. This signal is now
provided with a variable offset by the
tune control P7 in order to allow the
meter to be zeroed.

Manual tuning can become tedious if it
is required after every change of mode
and therefore an auto tune is a desirable
feature of the circuit. This is carried out
by storing a voltage level in Cl 8. This
voltage is the sum or difference of the
output of A2 and the position of the
tune control. A4 together with IC6 form
the basis of the auto tune circuit which
functions as follows. When the two
CMOS switches S3 and S4 close, the
output of A4, an opamp with a gain
of 4, will be fed directly to the input
of IC6 and cause this opamp to act as a
fast integrator. Since its output is fed
back to the non-inverting input of A2
the original voltage level on the output
of this opamp will be restored. Auto
tune is brought into operation by
flicking the handle switch S3 momen-
tarily. This action will turn on the
CMOS switches ES3 and ES4 thus
increasing the voltage level on pin 2 of
IC6 by a factor of 4. I

S3 also serves as the mode change *
switch when operated in the other .

11-08 - elektor noven

1981

direction, changing from ground effect
to reject mode or vice versa. It will
be apparent then that S3 is effectively a
three-way switch. Mode change is
achieved by the gates N 1 . . . N4 which
are configured as a flipflop (although
they may not appear to be so in the cir-
cuit diagraml). The complementary
outputs of the flipflop control the two
CMOS switches ES5 and ES6.

Many readers may find the going a
little tedious by now, but take heart:
we are almost at the end, in fact only
the meter and audio sections of the
circuit are left. The two functions of
the meter are selected by the CMOS
switch ESI. This is 'normally closed'
and the meter will be connected to the
detector circuit. Pressing the battery
check switch S4 will open ESI and at
the same time, cause T5 and T6 to con-
duct. The meter will now be connected
across the batteries via R19/D1 and
R26/D2.

The remaining part of the circuit is the
audio stage. It is slightly unusual but
fairly straightforward. The output
from the PLL oscillator (IC1) is fed to
IC3 where it is divided by 32 to pro-
duce a 270 Hz square wave. This is then
'sharpened' by C7 and R27 to provide
a control pulse for the CMOS switch
ES2. In the normal modes an audio
signal will only be heard when the meter
moves to the right, indicating the
presence of a wanted 'find'. This is
when a rising positive voltage appears at
the output of A2 and is fed via R25 and
D4, through the CMOS switch to the
output stage via T8.

It is also possible to use an audible
signal for rejected finds. This may not
be obvious at first but after some
experience with the metal detector this
feature will be found to be a useful
extra. When S2 is switched off, the out-
put of A3 is allowed to pass, via R55/D5
and again ES2, to the output stage. The
audio amplifier consists of transistors
T8, T10 and Til with T10 acting as a
level converter.

Construction

Readers will be in no doubt by now that
the Elektor metal detector has a very
professional appearance and in no way
suffers from the usual nuts, volts and
tobacco tin look that usually grace
home construction projects of this type.
The good news is that, yes, your metal
detector can look exactly the same os
that shown in the photograph here.
Even more important, the major prob-
lem that prevents construction of a
really good metal detector, that of
winding and setting the coils in the
search head, is completely taken care of.
The entire case and hardware assembly,
together with the potted, waterproofed
and tested search head, has been made
available to readers via Crestway Elec-
tronics Limited and it can be purchased
with or without electronics. The overall
design is to fully professional standards
and the finished metal detector will out-

perform many high quality commercial

Should any reader wish to produce his
own search head it is just possible
although not very practical, since the
setting of the coils is extremely critical
and their position makes the difference
between a very good detector and no
detector. However, for those interested
the head consists of two 1 0" coils with
80 turns each of 29 swg enamelled
copper wire slightly flattened (to form
a D) and overlapped by approximately
1". The critical setting is then found by

trial and error.

The electronics of the metal detector is
surprisingly easy to construct. In fact, if
the printed circuit board is used, it will
present no problems at all. Assembly of
the board is carried out in the usual
manner after which the point to point
wiring is completed, using figure 4 as a
guide.

Particular attention should be paid to
the wiring of the potentiometers or the
direction of the rotation may be incor-
rect. If one of them does appear to be
the wrong way round, exchanging the

metal detactc

slektor

aber 1981 - 11-09

wiring at each end of the track (the out-
side tags) will provide a solution.

The handle switch S3 (ground effect/
reject mode and auto tune) is a momen-
tary two-way switch with a biased
centre off position. This must be wired
correctly to avoid confusion in use. The
switch should be fitted so that auto
tune is operated when the switch is
flicked to the right. The contact that is
'made' in this position (usually the left-
hand one) should be the one connected
to K on the printed circuit board.

A note on the printed circuit board,
there are two points marked -9, these
have to be wired together externally. If
stereo headphones are used they should
be converted to mono by shorting the
left and right leads together (not the
common lead).

It will be wise to ensure that all connec-
tions are correct before connecting the
batteries.

Calibration

Any electronic device will only be as
good as its calibration and for this
reason the setting of the three preset
potentiometers PI, P2and P3 should be
carried out with care. Some readers tend
to get a little alarmed at this point but
we can dispel any worries right away.
The metal detector has been specifically
designed to allow the calibration to be
carried out — with nothing more sophis-
ticated than an ordinary multimeter
with reasonably accurate 1 V and 10 V

Initially all controls must be turned
fully clockwise. The unit can now be
switched on by turning SI to the R2
position. The first check is the positive
supply rail and this is carried out by
connecting the meter (switched to
1 0 V DC) across C9 — taking note of the
polarity. The reading should be between
4.7 V and 5.3 V. The negative supply
can be checked across CIO, again with
attention to polarity, and in this case a
reading between -4.7 V and —5.3 V
will be fine. If either reading is outside
these limits the components in the sup-
ply regulators must be checked before

If the supply voltages are correct the
meter can now be connected between
the test point (TP, pin 10 of IC2) and
0 V (negative lead). With the meter
switched to the 1 V range adjust PI for
a reading of 0.55 V. Switch SI to the
R3 position and adjust P2 for a reading
of 0.15 V. After flicking the handle
switch to select ground effect mode, P3
can be adjusted for a reading of 0.82 V.
And that is all - the calibration has
been completed I

Operating instructions

1. Switch SI to the R3 position.

2. Press handle switch S3 to right and
hold.

3. Rotate tune control until meter

4. Rotate gain control to position 8.

| 5. Release handle switch.

Photo 2. Possibly the most important part of
the metal detector, the search head, is
supplied as a complete and finished assembly.

6. Adjust volume control to 1/2 way
position.

7. Using a gentle sweeping motion
sweep the ground in front of you

with a left-right-left movement while
walking slowly forward. The maximum
sweep should be about 1 8".

8. Practice for a while using this pro-
cedure and try to receive signals

from all types of metal.

Once you are confident in the use of the
basic procedures of the machine then
move on to the following: using steps
1-6 set up the machine in basic oper-
ation mode and carry on with the fol-
lowing steps.

9. Flick handle switch to the left and
release.

10. Rotate the reject mode control and
check that the mode is correctly

selected by a swing on the meter, if not
return to step 9 and repeat.

1 1 . Switch SI to the R2 position and
set the reject mode control to

1/2 way.

12. Flick the handle switch to the right
(auto tune) momentarily and re-

13. At this point you should be re-
jecting tinfoil silver paper etc. but

accepting coins (cupronickel) bronze,
brass, silver and hopefully gold.

14. Rotation of the reject control to-
wards zero will decrease the amount

of rejection, and towards 10 will in-
crease rejection.

15. Practice using the above until you
are familiar with the machine, then

try the following.

16. The coarse settings of positions R1,
R2-, R3 on SI works as follows:

R1 will give the maximum rejection of

R2 will give average rejection and is the
most used position.

R3 will give a minimum metal rejection,
in other words, receives most metals.

17. Ground effect control

On some soil the so-called ground
effect may be experienced, due to min-
eralised soil. The same effect will appear
when using the detector on beaches.
This phenomenon produces the effect
of an almost continual and apparently
haphazard series of finds. The indication
of ground effect can be verified by a
drop in audio output when lifting the
search head away from the ground. It
must be noted that ground effect is a
phenomenon that all metal detectors
have to overcome. The Elektor metal
detector is able to eliminate this prob-
lem when the procedure described here
is followed:

18. Depress handle switch to the left
momentarily and release.

19. Rotate the ground effect control
whereupon the meter should move.

20. Adjusting this control from zero to
10 will increase or decrease the

compensation for ground effect and the
best setting is when you can lift the
head away from the ground, with no
change of signal.

21. Gain control, this control sets the
sensitivity of the machine and is

normally turned up towards position 10
for most of the time. However, if a
beach is being searched for coins, which
will usually be just on the surface, then
turning the gain down will make the
detector sensitive to coins on or just
below the surface and anything buried
deeper will be ignored.

22. Auto tune switch

At any time in use, if changing
mode or resetting of the gain, reject and
ground controls, then momentarily
operation the handle switch to the right
will reset the tune position back to zero

Photo 3. The distance between the head and the handle is fully adjustable.

1 /

elektor novomber 1981 - 11-11

on the meter, but note: do not hold this
switch over while actually searching for
metal as the fast integrator circuit will
be constantly in use.

23. The slide switch S2, at the left hand

rear of the case, is normally in the

nearest position to the front panel. This
corresponds to an audible indication
when the meter moves towards the
right. If the switch is in the other pos-
ition then an audio signal will be present
at all times whether the meter is moving
towards the right or the left, but a
change in tone will be heard as metal is
detected. This mode is useful for general
searching, since you do not need to look
at the meter all of the time.

24. Battery test button

Pushing this will give a meter
reading indicating the state of the bat-
teries. The batteries should be replaced
when the meter reading falls below 10
on the right hand half of the scale. M

Notes for treasure hunters

Successful treasure hunting with any
metal detector requires plenty of luck
and experience. We are of the opinion
that the former is somewhat unreliable
but the latter can be gained with prac-
tice. After a period of use and time in
which to become familiar with the
metal detector, you will get to know
little tricks like telling the difference
between ferrous and non-ferrous ma-
terials from indications given by the
detector alone. Steel is indicated by a
more rapid movement of the meter. It
is not really understood why, but
bronze tends to give a warbling note in
the headphones Small coins and rings
lying on edge in the ground will produce
a very sharp on/off signal. It is also
useful to know that the larger the size
of a buried object, the more area the
signal will be on for. We are also reliably
informed that gold actually gives a
double pulse in the form of an echo.

For consistent results, do not move too
fast I neither sweeping nor walking) and
keep the head within 1/2 to 1 inch
above the ground. This becomes a lot
easier with practise. After use always
dean the search head with a damp
doth.

Above all, respect other people's prop-
erty, and request permission before
searching an area of land, it is almost
always granted. The Elektor metal
detector head is perfectly water-proof
and can therefore be used in rivers
which can prove quite fruitful.

We are given to understand that growing
a beard and rising at unheard of hours
does bring a measure of success but we
have no documentary evidence to prove
this. The seasoned treasure hunter can
be recognised by this downcast, glazed
expression and constant mumbling. His
hand never leaves the shovel that travels
everywhere at his side and he wears
headphones in bed I

Treasure hunting can be rewarding,
frustrating and damaging to your social
life but it will never be boring.

1-12 — elektor

boost

A.M. Bosschaert

All electric guitars and bass guitars
have some sort of tone control,
although this is generally very
unsophisticated. This high boost
circuit can provide a great
improvement, for, as its name
suggests, the treble can either be
boosted or cut by about 35 dB.

In addition, the turnover point of
the tone control can be set to
three different frequencies by
means of a single switch.

It is a well-known fact that for some
inexplicable reason, electronics has
never succeeded in catching up with the
quality of electric guitars. The average
tone control circuit on an electric guitar
usually consists of little more than a
capacitor and a potentiometer and can
hardly be expected to produce very
good results. An active tone control is
much more effective and the high boost
circuit, for instance, can either amplify
or attenuate the treble over a ± 35 dB
range. The circuit is compact in size,
allowing it to be fitted inside the guitar
body, if desired. Its current consump-
tion is low enough for a battery to be
used as a power source.

In addition, the tone control is equipped
with a turnover point switch. Under-
standably, not every guitar owner is
going to be prepared to cut holes into
his/her precious instrument. Taking this
into account, the circuit enables both
the treble and the turnover point to be
selected with a switch that is possibly
already fitted to the instrument (as in
Stratocaster and Les Paul copies, for
instance). This means that the guitar can
be provided with a number of practical
facilities without having to be dis-
figured.

Operation

Figure 1 contains the tone control cir-
cuit diagram. At the heart of the circuit,
IC2 constitutes an active tone control
together with R5 . . . R9, PI, C3 and
C4. The tone control is preceded by an
emitter follower that is built up around
T1 . This serves as a buffer for the high-
impedance guitar pick-up at the circuit's
input.

The DC offset of IC2 is determined by
the resistors RIO and R 11. As a result,
half the supply voltage is fed to the IC's
non-inverting input. The output of the
opamp also determines the bias of T1
via resistors R3 and R4 in the feedback
loop. The opamp used is not a common
type, but has been selected for this
purpose because of its low current
consumption.

As mentioned previously, the tone con-
trol uses a three-position switch to
select one of the three turnover points:
250 Hz, 800 Hz and 2500 Hz. The
setting is altered by means of electronic
switches which connect R8 and R9 in
parallel to PI. The electronic switches,
ESI and ES2, are controlled by the
D-type flipflops FF1 and FF2. These
are wired so that the count cycle is as
follows: 00-01 -10-00-01 -etc. When the
count is 00, no resistor will be connec-
ted in parallel to the potentiometer and
the turnover point will be at its lowest
level (250 Hz). When the count is 01,
however, ESI will switch on, so that R8
will be connected in parallel with PI
and the turnover point will be at 2500
Hz. Then there is the 10 count, where
R9 is in parallel to PI and the turnover
point frequency becomes 800 Hz.

Switch SI controls the FF1 and FF2
counter. This switch is operated with

the aid of PI. A type of monoflop is
made using the two other electronic
switches in IC3 for the purpose of
switch debouncing. Operation is as fol-
lows. PI is usually turned to adjust the
treble. If a different turnover point fre-
quency is required, the potentiometer is
turned fully anti-clockwise, thereby
opening the switch in the pot. The
potentiometer is then turned in the
opposite direction. Now the treble
control can be adjusted as required.
After the switch has been operated
three times, its initial turnover fre-
quency will be restored.

SI can also be operated independently
from PI. For this SI must be replaced
by a pushbutton type. Depressing the
pushbutton will then step through the
turnover frequencies.

The current consumption of the circuit
is exceedingly low, slightly over 0.5 mA,
which means that the circuit may
comfortably be powered with a small
9 V battery.

Construction and setting-up

Figure 2 shows the printed circuit board
on which all the components are
mounted. The board is small enough to
be fitted inside a guitar, but it can also
be housed in a separate case, which may
be preferable for aesthetic reasons.

Most electric guitars have at least two
potentiometers: one acts as a volume
control and the other as a tone control.
All that has to be done therefore is to
replace them by two different types.
PI is a potentiometer with a built-in
switch. The switch is indicated as SI in
the circuit diagram. The volume control,
P2, is a 'normal', logarithmic type.

The potentiometers and switch SI are
now connected up to the printed circuit
board and so are the guitar pick-up and
the battery. There are two possibilities
for switching the power on and off. The
first involves using the supply switch S2,
which means linking the dotted J con-
nection on the board.

The second alternative is a slightly more
elegant solution. The jack socket on the
guitar is replaced by a stereo version.
When the plug is inserted, a 'short' is
caused between the connection for the
spare amplifier channel and ground,
since the plug is a mono type. By con-
necting the negative terminal of the
battery to the second channel connec-
tion of the socket and the circuit earth
to that of the socket, the power supply
will be switched on automatically when
the lead is plugged into the guitar.

If housed in a separate case, the circuit
may be provided with a small mains
power supply. After all, the circuit
barely consumes 1 mA.

How to use the circuit

Readers should know by now how to
use the circuit, but just to make things
clear: P2 serves to regulate the volume
and PI controls the treble. The turnover
point is altered by turning PI fully anti-
clockwise until it 'clicks' and back
again. For each 'click' a lower turnover

. 9 9 99

nnnnnnn

point is selected. Three operations are
required to return to the original turn-
over point. The order of selection is
high-middle-low. The first change will
only be a very subtle alteration to the
sound, whereas the lowest turnover fre-
quency will give the greatest change. M

11-1

liber 1981

For some time our design team have
been exploring the possibilities of a
successor to the Formant, the Elektor
synthesiser. Following the trend in
technology, it was felt that a new
concept was needed rather than just a
rehash of a basic idea.

As regular readers know, the Curtis ICs
have just recently made their appearance
on this side of the water. These have
been specifically designed for syn-
thesisers and are probably the farthest
that any company has dared to venture
in terms of musical 'chips' so far. No

die new
synthesizer

programmable and portable

The success of the Elektor Formant synthesiser led us to the opinion
that there is a great interest among our readers in the field of electronic
musical instruments, especially synthesisers. The availability of the new
Curtis 1C described in last month's issue prompted us to embark on an
entirely new design.

Since the size of the Formant did not lend itself to portability it was
decided that the new synthesiser should be truly portable without any
lack in terms of performance. The new synthesiser is of modular
construction and can be expanded into a polyphonic instrument with
'programming' facilities. This, the first article in the series, explains the
basics behind the design.

further prompting was required to look
into the possibilities of utilising these
new ICs to form the basis of an entirely
new synthesiser.

It was felt that a modern synthesiser
should take on a new look together with
simpler operation. This is the first
article in a series describing a truly
portable and fully operational instru-
ment that can be constructed in modu-
lar form allowing expansion up to a
polyphonic keyboard. Further it was
decided that the ability to 'program'
different sounds was a facility that
could be extremely useful.

In this article we begin by discussing the
fundamentals behind the new design.
Basically, the concept of the new
synthesiser is that of a set of modules
which could be combined in various
ways leaving the reader free to build
four different types of synthesisers
using the same basic printed circuit
boards. The possibilities are as follows:

1 . a simple synthesiser

2. a simple synthesiser including a
preset facility

3. a polyphonic synthesiser

4. a polyphonic synthesiser including a
preset facility.

the new synthesiser

Throughout the series, readers will be
expected to be familiar with the struc-
ture and operation of synthesisers in
general. However, anyone who is new
to this rather complex field can find all
the basics in FORMANT Book One.

Why the preset facility?

You only have to analyse the synthetic
sounds used in Pop, Rock-and-Roll and
Jazz to realize that the number of
different presets required is surprisingly
few. The audience recognizes the
characteristic sound immediately, which
is why a lot of rock bands use one
particular sound regularly as a kind of
'label'. Furthermore, setting individual
modules is extremely time-consuming
and, on stage especially, this can be a
real nuisance. Things can be simplified
by providing a 'manual/preset' switch
for all the elements required to produce
a complex sound effect: filter fre-
quencies, attack and decay times, VCF
resonance factors, the interval between
two VCOs or the envelope amplitude,
and so on. This is illustrated in figure 2,
where inputs 1 ... 4 are for the preset
control voltages.

Since only very few variations are
regularly used in practice, they can
easily be stored as 'programs' and
'called' when needed with the aid of
a single switch or a decimal keyboard.

If only four situations are required per
parameter, say, it is not even necessary
to store the exact control voltage
values in memory. CMOS analogue
switches can be used to select the
desired voltages as shown in figure 3a;
in effect, this works like the rotary
switch shown in figure 3b. The only
data that must be stored is the 'setting
of the switches'.

Obviously, the possibility of full manual
control by means of knobs on the front
panel must be maintained as an option
— if only for special effects.

The individual boards belonging to the
compact model do not need to be
modified if the circuit is provided with
a preset facility at a later date. The
voltages which control the filter fre-
quency, attack times, etc. are fed to
the corresponding modules in the
compact model by way of the potentio-
meters on the front panel.

For the preset option, provision must be
made to break the connection from the
front panel controls, as required, and
drive the modules from some fixed
(preset) voltage instead. As mentioned
above, CMOS switches are the obvious
solution. For the VCOs, four different
preset voltages can be selected as shown
in figure 3. The envelope waveform
from an ADSR module can be 'voltage-
controlled' by passing it through a VCA
(figure 4), and selecting any desired
waveshape from the VCO output is
only slightly more complicated, as
illustrated in figure 5.

As can be seen, eight voltage controlled
switches are needed to select the

i. With the addition of an LFO. whic

riangle signal, and .

diffei

>ise gen

c

D

4

3

.J.

Figure 2. The control inputs 1 ... 4 on the various boards can either be linked to the wipt
the potentiometers on the front panel or to external control voltage inputs.

waveforms (but if a 4066 is used, only
2 ICs are necessary). The principle is
fairly straightforward. When the preset/
manual switch S2 in figure 5 is in
position A, SI' in IC2 will be closed.
The waveform can now be selected with
SI. Since the analogue switches in IC3
are switched 'off' when the manual/
preset switch S2 is in position A, the
data at the BCD inputs will have no
effect on switches S2' . . . S4' in IC2.
Resistors R6...R10 make sure the
switches are held fully 'off' when no
voltage is applied to the control inputs.
With S2 in position B, however, external
data at the BCD inputs of IC3 will
select the output waveshape.

As can be seen in figure 5, the octave
'range' switch and the fine tuning

potentiometer are connected to the
VCO control input via IC4, in the
'manual' mode. When S2 is switched to
'preset', the pre-programmed control
voltage is selected instead.

Driving the preset inputs

Since the synthesiser is made up ot
individual modules it can be controlled
with the aid of external voltages.
That's all very well, but how can this be
put into practice? As an example,
let's assume that sixteen preset voltages
are required for any given setting.
Corresponding digital information can

1981 - 11-17

Figure 7. The block diagram of a polyphonic synthesiser. A polyphonic keyboard provides various control voltages (KOV) and gate pulses which

be stored in EPROM, as shown in
figure 6. The locations for any given
setting can be scanned in rapid suc-
cession and passed through a D/A
converter and 16-channel (analogue!)
multiplexer to sixteen sample-and-hold
units. These, in turn, drive the control
inputs of the various modules.

The polyphonic version
In the polyphonic version (figure 7)
the number of basic units required
depends on the number of keys that are
to be played at the same time. Each key
has to be provided with a VCO, a VCF,
a VCA and the corresponding envelope
generators. All the parameters are
controlled centrally either by the knobs
and switches on the front panel or by
the stored preset information. This
means that when the synthesiser is
expanded into a polyphonic instrument
there is no need to modify the front
panel. This has the advantage that the
user does not have to buy everything at
once; instead, the monophonic unit can
be extended by adding other boards. It
is important however to have a poly-
phonic keyboard with separate control
voltage outputs. M

8

gmber 1981

LCD fr

LCD

frequency

counter

The nice thing about our hobby is that
once in a while an electronic device
comes along that just begs to be used in
one way or another, and not just for
the original purpose that it was designed
for. Just such a device in the neat little
package from Thurlby Electronics, the
FM77T. This minor miracle is actually
a complete frequency counter module
that includes a 4% digit LCD display, all
measuring just 60x 38x 10 mm. It
incorporates a CMOS LSI chip and a
6.5536 MHz crystal oscillator and the
total will measure and display up to
4 MHz with no external components.

As to be expected, we haven't stopped
at just a 4 MHz frequency counter in
our use of the module. In fact this, the
first of the articles featuring this device,
incorporates a prescaler enabling a
count frequency of up to 35 MHz to
be achieved, and it is all mounted in a
hand held case. We plan a further
counter with a capability of 150 MHz
followed by a digital capacitance meter.
There are even more projects in the
pipeline but they must remain undis-
closed at the present time.

professional appearance and performance

This is the first in a series of projects featuring a frequency counter
module with a liquid crystal display. The high performance is out of all
proportion to the simplicity of the circuit. Two switched ranges are
available, the first up to 4 MHz for use in monitoring frequencies in
microcomputers and the second up to a maximum of 35 MHz to cover
CB transceivers and general use.

The complete counter and printed circuit board are housed in a hand
held case that has been specifically designed for a module of this type.

The counter module

Since the heart of the counter lies in
the FM77T module it is interesting to
see just what this package is capable of.
Besides being just a 4 MHz counter with
a reading rate of 10 per second it can
also serve as digital frequency readout
for a radio or a tuner. It is possible to
select any one of 26 preprogrammed
standard IF offset frequencies. The
module can then be used to display
the received frequency by measuring
the frequency of the local oscillator of
the receiver.

The 3 decimal points are also selectable
together with the kHz, MHz and LW
legends at the right hand end of the
display. Two other inputs are of im-
portance, both requiring a high level
(supply level) to activate. One is a
'hold' input which freezes the display
and the other is a 'reset' input which
will return the display to zero.

The maximum display is in fact 39999
but the counter will overflow, in which
case the correct reading will be the
display minus 40000. Thus 5,9 MHz
will be displayed as 1.9 MHz. The
module will operate on supply voltages
between 4.75 V and 7 V with a typical
power consumption of a little over
1 mA. A word of warning here, particu-
lar care must be taken to prevent in-
correct connection of the power supply
because this will convert the module
to an expensive inmate of the rubbish

The basic frequency counter

The first in the series of projects is a
hand held frequency counter having
two ranges, 4 MHz and 35 MHz (40 MHz,
if you really push it!). The specifi-
cations are given in table 1 and they are
very good, especially when the total

Frequency range 1 : 2 kHz . . . 3.999 MHz
input sensitivity: 30 mV RMS
Frequency range 2: 100 kHz . . . 39.999 MHz
input sensitivity: 80 mV RMS. 1 00 kHz ... 20 MHz
1 50 mV RMS, 20 MHz ... 30 MHz
450 mV RMS, 30 MHz . . . 35 MHz
900 mV RMS, 35 MHz ... 40 MHz
Maximum input voltage: 50 V RMS
Input impedance: 1 Mfi/IOpF for Uj n < 700mV
Calibration: none required

Power supply: 9 V battery or NICAO or external 8 ... 1 :

power /charging source
Current consumption: 40 mA max.

Automatic decimal point shift
Automatic kHz or MHz display

cost of the counter is taken into con-
sideration. Probably the most notable
point is that the input level can be
anywhere between 30 mV and 50 V
without fear of damage occuring.

Block diagram

The simplicity of the counter will be
apparent from the block diagram shown
in figure 1. The input amplifier is
followed by the divide-by-ten prescaler
to produce a maximum count of better
than 35 MHz. The divider can be by-
passed with the aid of a switch to obtain
a maximum count of 3999.9 kHz. The
automatic decimal point and legend
display is controlled by the lower block
in the diagram.

The circuit

Only a very few components are re-
quired as can be seen from the circuit
diagram in figure 2. The two diodes D1
and D2 are to protect the circuit from
excessively high input levels (50 V is an
absolute maximum). T1 and T2 to-
gether form a 'super source follower' or
impedance converter and will convert
the high (1 MS2) input impedance to
about 220 £2 for the amplifier N1. The
amplifier is in fact a TTL inverter but
its output will still be analogue for

1981

LCD freqi

3

Q

r&i D

a ■ KH;

'Hill'll

dp2

:v 3

'eOLW

. 1 2 3 4 5 6 7 8 9 1011121314 16 ±

r ’j -T- 1

Photo illustration. The i
Vero 65-2996H case.

4

Parts list

Resist ors:

R1 = 1 k
R2 = 1 M

R3.R6.R9 = 470 SI
R5 = 2k2
R7 = 4k7
R8= ion

Capacitors:

Cl =■ 100 n MKH
C2 ■ 1 00 p/6 V
C3 = 220 (i/16 V
C4 . . . C6 ■ 1 (i/6 V tantalum

Semiconductors:

T1 - BF 256A (not B or C)

T2= BF 494
IC1 - 74LS04
IC2 - 4030
IC3 - 74LS196
IC4 - 78L05
D1,D2- 1N4148
D3 . . . D7- 1N4004

Miscellaneous:

51 « DPDT miniature switch

52 ■ SPOT miniature switch
battery » PP9or NICAD equivalent
Hand held case = 65-2996H from

Vero Electronics Limited
Module = FM77T from Thurlby Electronics
Limited, Coach Mews, St. Ives. Huntingdon,
Cambridgeshire.

small input voltages. At pin 2 it will be
between 1.5 V and 1.8 V peak to peak
for an input voltage at Cl of 30 mV
RMS. The analogue waveform is con-
verted to digital by the pulse shaping
circuit formed by N2 and N3. Next in

the chain of events is a divide-by-ten
IC3, which can be switched in or out
of the circuit depending on the position
of SI.

The remaining gates N4 . . . N7 are
used purely as a decoding circuit to

determine the position of the decimal
point and the legend (MHz or kHz)
again depending on the position of SI .
The power supply for the counter is
slightly more complex since four supply
options are possible. A standard 78L05

regulator is used as the maximum
current consumption for the entire
counter is not more than 30 mA. In
the first case, power can be provided by
an ordinary PP9 dry battery which will
give approximately six hours of contin-
ual operation. Resistor R9 will not be
needed. If on the other hand the PP9
is to be replaced with a NICAD, R9 will
be required since this provides a path
for the NICAD charging current when
the counter is supplied with an external
8 ... 12 V AC source. The value of R9
is dependant on the NICAD used and
must be calculated to provide a charge
current of 20 ... 25 mA when the
NICAD is discharged. Resistor R8 acts
as a limiting resistor to prevent excessive
dissipation in the regulator. The final
option is an external DC power source
and this will appear in greater detail in a
future article.

Construction

Having said all there is to say about the
relatively simple circuit we can move
on to construction. The case used for
this project is the hand held box from
Vero, part number 65-2996H. The point
to bear in mind is that the interior of
the case will become fairly full therefore
it would make sense to assemble and
connect the component parts of the
counter together as far as possible
before mounting the whole into the
case. Ribbon cable will be ideal but
leave enough length to allow assembly,
particularly between the module and
the printed circuit board. It should be
remembered that the case is plastic and
is therefore defenceless against a sudden
attack by a hot soldering iron.

At this stage it will be as well to double
check that all wiring is correct (in-
cluding the battery connector).

The two switches can be mounted in a
small piece of circuit board material
that is fixed with super glue to the
inside of the lower half of the case.
Take care to space them apart enough
to allow them to fit side by side. The
module can also be fixed in place with
two or three drops of super glue. Only
a very small amount will be sufficient.
The printed circuit board can now be
mounted using three very short screws.
The BNC socket can be fitted last of
all and making this connection is the
only soldering that needs to be carried
out inside the case. The miniature
socket for the external AC power source
can be mounted at one end of the
battery compartment.

Before finally closing the case, ensure
that connecting wires can not become
trapped between the two halves of the
case. If satisfied that all is as it should
be a power source can now be con-
nected to check the operation of the
counter. There is no calibration required
therefore the reading will be correct
right from the start. M

telescope control

seeing distant stars . . .

Stopping the movement of
heavenly bodies (on a
photographic plate at least) is the
purpose of this article. A camera
mounted on a telescope can, with
a long enough time exposure,
enable the more distant stars to be
observed. To do this the telescope
must be able to track the star
accurately for the period of the
exposure. Two 24 V synchronous
motors are used to enable the
telescope to traverse at an
accurate speed in either direction.

An important part of astronomy is
of course observing stars and other
heavenly bodies. Observations can be
carried out with the aid of a telescope
or even, in some instances, with the
naked eye. A vast number of stars still
remain unseen however and in order to
study them, somehow these (or some of
them at least) must be made visible. The
method of doing this is to take a photo-
graph of them through a telescope but
in order to be any good it must be a
time exposure.

Exposure times varying from some
minutes or even hours are necessary for
the very distant stars to become visible
on film and this can cause some prob-
lems. During a fairly long exposure the
'position' of the star will change due to
the rotation of the earth and the result
on the film will not be a dot but a dash.

Since halting the rotation of the earth
is a little beyond the scope of this
particular article we will have to deal
with the problem in a simpler manner.
Following a star with the telescope can
obviously not be done by hand if a clear
image on the film is to be obtained. The
movement will have to be so small and
carried out so slowly that any star-
watcher would probably revert to stamp
collecting before sunrise. The answer is
of course a motor drive.

Motor drive

For a motor drive to operate precisely
a synchronous motor is an obvious
choice. An electric clock motor could
well spring to mind since the movement
of the telescope can be compared with
the hour hand of a clock. However, a
clock motor will invariably be too weak
and therefore, either a motor specifi-
cally designed to drive telescopes or a
motor and gearbox combination must
be used. It must be a synchronous
motor because the speed of this type of
motor is directly dependent on the
frequency of the AC supply that powers

Almost all synchronous motors available
in the U.K. have been designed for use
with a frequency of 50 Hz. This pro-
vides them with a very stable speed,
exactly what we want for our purposes
— nice and simple! Unfortunately not,
there is a snag. Yes, the motor speed
should be constant but it must also be
accurately adjustable within certain
limits. This is necessary because the so
called 'astronomic day' isn't exactly
equal to our common or garden 'earthly
day' of 24 hours, but slightly shorter.
Moreover the length depends on the
seasonal 'wobble' of the earth. We are
currently working on an article to
rectify this using PLL (Planet Locked
Loop). The AC supply used to drive the
synchronous motor must therefore be
finely adjustable. This article provides
the circuit for just such a variable
frequency AC power supply designed
for 24 V 50 Hz motors.

The AC generator

The circuit diagram for the AC gener-
ator is shown in figure 1 . The two
opamps A1 and A2 together form a
triangular waveform generator. This

waveform is 'rounded off' by R5 and
the two diodes D1 and D2 because a
synchronous motor prefers a more
sinusoidal supply.

The frequency of the generator is
adjustable between approximately 43
and 61 Hz via PI. The preset poten-
tiometer P2 is used to adjust the output
amplitude of the bridge amplifier.

A bridge configuration is used because
it is able to produce an output ampli-
tude of twice the supply voltage level.
In this case, the amplified AC voltage
between the points A and B has a peak
value of 32 V. This results in an RMS
value that is fractionally lower than the
desired 24 V that is needed to drive the

Left and right

To be practical, the telescope must be
able to turn in both directions to
enable it to track any star in any pos-
ition. Synchronous motors however will
only drive in one direction, they are not
reversible. The simple answer to this is
to provide the drive with two motors
arranged to drive in opposite directions
on one shaft. If one of the motors is
now supplied with power, the other will
just free-wheel backwards.

The drawing in figure 2 illustrates how
power is supplied to each motor. The
points A and B in this drawing corre-
spond to points A and B in figure 1 . The
two relays Rel and Re2 are controlled
by the three position switch (left, stop,
right). Motor Ml is switched on by Rel
and motor M2 by Re2. The switch and
relays are arranged in this manner to
enable the two motors to be controlled
remotely which may well be desirable
to prevent vibration of the telescope.
Since movement of the telescope will be
very slow, the LED's have been included
to give an indication of direction, if any.
The capacitor C4 must be unpolarised,
that is, an electrolytic will not do. The
control electronics require a stabilised
supply of 18 V while the power stage
T1 . . . T4 can be connected to an
unstabilised 24 V DC source. K

aleklor november 1981 - 11-23

solar \\\l/y

\\ s

The sky's the limit as far as fuel costs are concerned and, like Atlas, we
are all finding it increasingly difficult to shoulder the burden. One
consolation: when Earth's resources are finally depleted, we will still be
able to turn to the Mother of all energy, the sun.

Slowly but surely, solar cells are at last coming down in price, so that
one of these days 'sun-powered' radios, kitchen equipment, heating
systems, etc. are going to be economically viable.

Developments in technology have always fascinated hobbyists and
scientists alike and readers are invited to participate in an experiment
and build a solar powered receiver.

The need to save energy has become a
daily part of our lives, not least because
of the huge publicity given to it. The
media are really going to town and so is
the government, what with the adver-
tisement campaigns on television, radio
and in the newspapers. The effects are
sometimes quite astonishing. The notion
'room temperature', for instance, seems
to have dropped to a sub-zero level.
People are reluctant to open the curtains
in the morning for fear of losing precious
warmth.

Contrary to what might be expected,
the result is a vicious circle. For now
that less energy is being consumed, gas
and electricity companies are losing
money and are therefore forced to put
up their rates. The consumers then react
by 'tightening the belt' even further, so
that the prices go up again . . . etc. . . .
etc.

The fact remains: the world's resources
are being exhausted and at an alarming
rate. There's only one thing for it, new
substitutes will have to be found. One
of the main alternatives currently being
tried out on a large scale is solar energy.
Developments have not yet reached the
stage where the 'big' domestic ap-
pliances, such as washing machines and
central heating systems, can be powered
with solar energy, but the 'small' ones
can and this also includes radios. The
design described here provides a low-
cost portable receiver that can be
powered with surprisingly few solar
cells.

Low power

Solar cells are now readily available in
all sorts of shapes and sizes, although
the smaller kind tend to be triangular. It
is unfortunate that the really effective
ones are very expensive. What's more,
each cell can only generate about
0.5 ... 0.6 volts, which must be taken
into account when designing a solar
powered receiver.

The circuit's current consumption and
power supply voltage will therefore
both have to be very low.

One design that goes a long way towards
meeting these requirements is the
medium wave receiver published in
March (Elektor 71, p.3-32). This re-
ceiver already has a current consump-
tion low enough for it to be powered
with solar energy. Unfortunately, how-
ever, the audio amplifier in the circuit
requires a rather high supply voltage,
which means an awful lot of solar cells
would have to be connected in series for

Let's forget about the audio amplifier
for the moment and look at the receiver
section. The prospects here are much
more favourable. The ZN414IC from
Ferranti which is at the heart of the
receiver fits the bill perfectly. The IC's
supply voltage range is between 1.2 and
1.6 volts and only requires 0.3 mA. In
other words, two or three tiny solar
cells are all that is needed.

1 -24 - elektor november 1981

15V

Figure 1. A solar cell receiver design for beginners. The circuit includes a ZN 414 1C, a single transistor that acts as an audio amplifier and a high
impedance crystal ear-phone.

Figure 1 shows a simplified version of
the MW receiver which includes an
ear-phone. Its current consumption is
very low (less than 0.5 mA| and even on
an overcast day the circuit provides a
reasonable reception using the cheaper
kind of solar cells. Provided three
cells are used which, when combined
together, produce about 1 .5 ... 1 .6 V,
the components may have the values
indicated in figure 1. If the output of
the cells reaches as high as 0.6 volts
each, the third one may be omitted —
only then it is advisable to lower the
resistance of R2 to 470 J2. The ear-
phones must be high impedance crystal
types, as otherwise T1 will be over-
loaded and the receiver will not func-

As can be seen from figure 1, this MW
set needs very few components and yet
it leads to a fully 'self-sufficient' radio.

Medium and short wave superhet

Obviously, a set that is run on only two
or three solar cells is bound to have a
flaw somewhere. You can hardly expect
to drive a first-class receiver on a power
supply of 1.5 V!

Readers who are prepared to dig a little
deeper into their pockets and buy six
cells can build a much better design that
only requires a supply of 3 volts. The
prototype constructed at Elektor did in
fact look (and soundl) very promising.
The circuit diagram is shown in figure 2.
The receiver covers one medium wave
band and two short wave bands. The
first of the latter extends from 1.7 to
5.1 MHz and includes the popular
'fishermen's' frequency range. The
second range is situated between about
5.1 and 15 MHz, and includes the
49 metre band (= at around 6 MHz).
The prototype had a sensitivity of about
2 pV.

What kind of receiver is it? Well, basi-

cally, it is an ordinary superheterodyne
with a few tricks here and there in the
circuit to ensure low power operation
with a minimum current consumption.
Provided the volume is not turned too
high, the whole receiver should not need
more than 5 or 6 mA — which could get
the circuit mentioned in the Guinness
Book of Records!

Now let us take a closer look at figure 2.
First of all, the audio amplifier. Regular
readers will recognise it to be the 'ulp
amp' published in last year's Summer
Circuits' issue (circuit no. 55). The
amplifier is constructed with ordinary
transistors and operates at any supply
voltage level between 3 and 12 volts. It
will produce about 100 mW maximum
power. The design of the output stage,
which has no quiescent current adjust-
ment, ensures that current consumption
is extremely low (1.5mA) which makes
it eminently suitable for the solar cell
power supply.

Now what about the actual receiver
section?

The power supply voltage here is
stabilised in the circuit around T6 . . .T8.
Again, any voltage between 3 and
12 volts may be applied. Transistor T1
amplifies the RF input signal and S2
allows the tuned circuit to be switched
from one frequency range to another.
The local oscillator is built up around
T3 and T4 and can, of course, be
switched between the three frequency
ranges with the aid of the same switch
(S2b). Transistor T2 mixes the oscillator
and the input signals, after which the
455 kHz IF signal is filtered by means
of a ceramic filter (TokoCFM 2-455 A).
The IF signal is then amplified and
demodulated inside the ZN 414.

Time to go into a few more details.

The RF amplifier T1 is an ordinary
amplifier stage which is biassed at an
unusually low level. There's nothing

extraordinary about the oscillator T3/
T4, except that it is not usually in-
cluded in receivers at this particular
spot. The design selected is fairly
reliable at the low supply voltage and
yet capable of producing a sufficiently
powerful oscillator signal. This oscillator
has the added advantage that it only
requires a single pole switch to select
the various frequency ranges. The oscil-
lator coils (L4 . . . L6) do not have taps.
The mixing stage T2 and the filters
following it are quite straightforward.

As for 1C 1 , this was discussed in depth
in the article on the MW receiver pub-
lished in March. Since the 1C used, the
ZN414, consumes very little current,
there was no need to look for a more
economical component. Its automatic
gain control, however, has been adapted
to suit the more serious nature of the
present design. After all, the solar cell
receiver needs to be of a much higher
quality than its MW counterpart (which,
remember, was meant to teach budding
electronics enthusiasts how to build
thier first radio!) and the straightfor-
ward version shown in figure 1.

Normally speaking, the AGC's range
would only be about 20 dB which is
certainly not enough for a good receiver.
This is remedied by deriving an ad-
ditional control voltage from the output
of IC1 by way of D1, R44 and R45, as a
result of which T5 will be 'cut off', that
is to say, it will stop conducting, in the
presence of powerful signals. Thanks to
this modification the AGC now has a
very reasonable range of about 50 dB.

All in all, the circuit is very cost-effective
and although it looks rather compli-
cated at first it should be easy enough
to build. The only really expensive
items are of course the solar cells, but
rumour has it that in the near future
these too will be available at a much

11-26 — elektor november 1981

the mini nr&m

ITT recently introduced the highly
interesting SAA1900IC which, the
company claims, is the 'key' to a com-
plete organ. The data sheets, however,
are a little ambiguous about the chip's
possibilities. On the one hand, they pro-
vide a great many fascinating technical
specifications, but on the other hand,
they modestly christen their brain-
child a 'One Chip Toy Organ 1C'! This
article intends to discover whether the
1C is in fact just another 'toy', or
whether it can be recommended for
serious musical instruments as well.
And the best way to find out is, of
course, to get hold of such an 1C and

As readers can gather from the heading,
the test not only proved successful, our
designers had so much fun that the 1C
was up-graded from 'Toy Organ' to
'Mini Organ'!

The SAA 1900

The 1C includes a keyboard scanning
facility. This scans the 56 single key
contacts, 37 of which belong to the
'solo' keyboard and 19 of which consti-
tute the 'accompaniment' manual. It is
fully polyphonic with separate outputs
for CHORD and BASS. These are con-
trolled by the accompaniment section.
In addition, there are two separate voice

musical fun and games
on a single chip

Building an electronic organ from
scratch can be a tedious and
expensive business costing
anything over £ 200 and several
weeks of hard work. Now for the
good news: this particular organ
incorporates all the electronics
needed on a single compact board
and can be built in a matter of
hours. Thanks to ITT's special
'organ 1C', the mini organ is highly
economical and provides a
remarkably good performance
(that is, as long as the organist
isn't 'all thumbs"!). Fun for
everyone ... at the price of a
single chip.

1

.ASS Q <V\,

outputs, 4' and 8\ controlled by the solo section. This set-up is very practical A few nasty shocks like that and the

'solo' part of the keyboard. and offers the advantages of a church organist will be 'conditioned' to avoid

Whereas the 37 upper keys provide the organ, where a set of stops allows for a such mistakes.

melody manual, the 19 lower ones are variety of volume - and manual - com- The tone signal's asymmetrical square

used for the left-hand accompaniment, binations. wave form, which is full of harmonics.

By 'accompaniment', something more is With only 19 keys, the accompaniment provides a sufficiently resonant sound,

meant than the usual 'one-finger' tech- manual's possibilities are rather limited. In addition, the lower manual controls

nique. Instead, genuine polyphonic keys After playing a few chords 'up the scale’ a separate bass output. Since this is

are available, so that the organist is free the player will run out of notes and monophonic, only the lowest note in

to invent chords. Furthermore, the inadvertently striking a solo key will each chord that is played will reach the

accompaniment manual has its own out- lead to a jarring surprise: the C at the output. The IC's circuitry can be

put, enabling the volume of the left- beginning of the solo manual is an changed to allow the highest note of a

hand backing to be adjusted indepen- octave lower than B (on the accompani- chord to be heard at the bass output

dently from that of the melody on the ment manual) preceding it! instead (this requires a logic 1 level at

iber 1981

pin 11). The Elektor printed circuit
board is however designed to cater for
the first version, as this is the one most
frequently used.

The bass note, which is derived directly
from the IC's output is full of harmon-
ics and, with the aid of a low-pass filter
consisting of a resistor and a capacitor
(R 26/C 13) connected in front of a
summing amplifier (figure 2), its sound
is enhanced with the rich resonance that
is characteristic of an organ.

The upper manual has two voices. The
sound produced by this keyboard is
made up of two square wave signals
which each have a separate volume con-
trol and differ by one octave in fre-
quency. It is a pity ITT did not add
a third or even a fourth square wave
signal at one octave higher, as this
would have given the 1C a really pro-
fessional touch. Actually, it does not
really matter, for when the two square
wave outputs are mixed, the solo
manual produces a very satisfactory
sound. Sub-octaves can be produced by
divide-by-two's, provided only one key
is struck at a time. If several keys are
depressed simultaneously, the corre-
sponding signals will be mixed inside
the ICs and will appear in a mixed form
at pins 21 and 22.

The circuit diagram

e 3. The structure of the keyboi
lx. Only 15 wire connections (to
sponding 56 pins on the ICI are
red for 56 keys to be scanned.

The 1C does not include a clock oscil-
lator so that this will have to be added.
The economical 74123 TTL 1C serves
the purpose admirably.

The two transistors T1 and T2 con-
stitute a phase-shift generator. This pro-
duces a low-frequency sine wave signal
in order to frequency-modulate the
oscillator. The result is a slight vibrato,
which is exactly what an organ should
have.

P6 and P7 adjust the vibrato's amplitude
and frequency, respectively. Provided
the two potentiometers are both trim-
mers (as indicated on the board), the
vibrato effect can be switched on and
off by means of SI. By mounting P6
and P7 on the front panel, the frequency
and the amplitude can be regulated
continuously. In fact, the amplitude
can be turned down to zero, so that SI
may even be omitted here.

P5 serves to alter the entire voice range.
When the wiper of P5 is set in the mid
position, the oscillator frequency should
be 500 kHz. This value must be main-
tained, as the scanner matrix tends to
act rather 'strangely' at higher fre-
quencies.

If any component tolerances should
, prevent P5 from adjusting the frequency
into the desired range, the easiest way
to alter the frequency is to select two
equal capacitor values for Cl and C2.
The clock frequency's precision can best
be checked with the aid of either an

.

a

alektor

- 11-29

6

Figure 6. The keyboard's mechanical structure. Pieces of Veroboard are ideal for wiring the keyboard. Together with a plastic coyer and the
contact blocks, the Veroboard pieces are glued to the keyboard frame.

placed by a potentiometer on the front
panel. Pins 2, 21 ... 23 of the SAA 1 900
represent the signal outputs which are
connected to the output (link F) by
way of a mixing stage (PI . . . P4 and
IC3). The latter consists of an inverting
opamp in a well-known adder circuit.
The bass signal (pin 2) is then sent
through a low-pass filter before reaching
the input of IC3. Outputs A . . . E enable
the organ to be extended, if necessary,
at a later date and are therefore not con-
nected.

The keyboard matrix

The scanner matrix shown in figure 3 is
connected to pins 4. . . 10 and 12 . .

. . 19, as indicated in the IC's pin assign-
ment. Each one of the 56 junctions in
the matrix is made up with a keyboard
contact and a diode (see figure 4). A
special control circuit inside IC1 scans
every junction 'row by row 1 , one after
the other, until the key that was de-
pressed is detected. In other words,
, instead of 7 x 8, only 7 + 8 contacts are
required.

Note: In the matrix shown in figure 3,
^ the lowest note (the first key at the far
left of the keyboard) corresponds to
contact number 56. This is situated in
the lower right-hand corner, not in the
upper left-hand corner, of the figure.
To make things clearer: contact num-
bers 56 and 1 are assigned to the lowest
and the highest notes, respectively. As
long as the contacts are wired in the
manner shown in figure 3 and every
matrix point is linked in the manner
indicated in figure 4, nothing can go
wrong.

It is not advisable to construct a matrix
on the basis of wire and copper track

junctions and then connect each one to
a key. In practice, the simplest way to
construct a matrix is shown in figure 5.
This enables the 56 diodes to be incor-
porated on the keyboard as well. The
contacts may not be interconnected. It
is best to glue each contact block onto
a perspex, plastic or pertinax surface.
As far as mounting the diodes is con-
cerned, several pieces of Veroboard
should be cut into strips and glued
behind the contact blocks, with the
copper tracking pattern side facing up-
wards. The 15-wire connection between
the keyboard and the printed circuit
board is made by using two (7 + 8)
pieces of flat cable (see figure 6).

7

The power supply and the
amplifier

The organ circuit requires a total of five
different supply voltages: +15 V, -5 V,
-15 V, -18 V and finally a non-stabil-
ised voltage of about —23.5 V. Two
integrated voltage controllers (IC4 and
IC5) are already included on the organ
printed circuit board. These produce
—15 V and — 5 V respectively. The re-
maining voltages (— 23.5 V, -18 V and
+15 V) have to be connected to the
organ board. The need for a suitable
power supply is met by the circuit in
figure 7. This is in fact a dual power
supply and produces +15 V and —18 V

Figure 7. The power supply circuit. The use of a 250 mA transformer enables additional
circuits to be powered with up to 150 mA of current as well as the organ board itself. The

1981 - V

with the aid of two voltage regulators.
In addition, the required non-stabilised
23.5 V voltage can be derived at the
negative pole of the electrolytic capaci-
tor C2. This is done quite easily, since
the wire bridge to the right of D4 on the
board in figure 9 provides a suitable
connection point.

Obviously, the organ's tones, however
resonant, will be inaudible without the
addition of a miniature amplifier and
a loudspeaker. The LF output of the
organ (point F) can be connected to any
HiFi or PA amplifier. With the aid of
P8 . . . P10 the volume at all the IC's
outputs (except for the bass signal) can
be readjusted once the board is con-
structed. Furthermore, the opamp's
feedback resistor (R20) can be altered
in value to match the signal level to the
amplifier. (A higher value will cause the
volume to be turned up, whereas a
lower one will turn it down,)

The keyboard

A great many keyboards are available at
an even greater variety of prices. Since
an easy-to-play keyboard involves a con-
siderable amount of precision, both
with respect to the keys and to the con-
tact blocks, it is bound to be one of the
most expensive items in the instrument.
How much it costs to build the organ
will therefore depend on the price of
the keyboard.

The sound

The prototype produced a remarkably
convincing sound, which did away
completely with our initial scepticism
(evoked by the IC's size!). After hearing
the first few notes, it was quite obvious
that the organ isn't just another toy, but
an 'adult' instrument. When it is com-
bined with a monophonic synthesiser,
the accompaniment can be played on
the organ with the left hand while the
right hand plays the melody on the
synthesiser keyboard. M

11-32 -ele

sr 1981

uplifiar

telephone

amplifier

makes distant callers loud and clear

'Keeping in touch' is easier said than done, despite the modern
telephone networks that stretch to the four corners of the globe. For
one thing, a pound for a minute seems a lot of money to hear Granny's
faint voice ten thousand miles away and then not understand a word
she's saying. Elektor has come up with a solution in the form of an
amplifier which, when connected to the telephone, enables the whole
family to listen in to the conversation.

\

\\

Some callers, of course, don't need
amplifying, as anyone blessed with an
old aunt who bellows hearty greetings
down one's ear at eight o’clock on a
Sunday morning will agree. Here an
attenuator would be more appropriate!
But then that is an exception. Distant
and sometimes even local lines can be
very poor indeed, so that an amplifier is
really practical. For instance, when
relatives ring up from South Africa, say,
or Australia, it would be much more
economical if the whole family could
listen instead of having to 'queue up'
to say a few costly words. What's more,
the amplifier drowns any interference
caused by crossed lines and thousands
of 'clicking' relays, so that the once
distant voice sounds as loud and clear
as if the person were sitting in the same
room.

Now that we know what the amplifier is
for, we can study the circuit diagram in
figure 1. Looking at the drawing from
left to right, the circuit starts with a
pick-up coil, the centre contains an am-
plifier and at the other end there is the
loudspeaker. The pick-up coil operates
according to magnetic principles: any
alteration in the magnetic field that is
radiated by wires in the telephone set or
in the receiver will be fed to the am-
plifier. This slightly roundabout system
is necessary, since a direct electrical
connection to the interior of a tele-
phone is forbidden.

The rest of the circuit diagram in figure
1 comprises very few components. LI
I represents the telephone pick-up coil

1981

1

Figure 1. The circuit diagram of

the telephone amplifier.

Parts list

Resistors:

R1 = 100 k
R2 = 39 k
R3 = 2k2
R4 = 680 fi
R5 = ion

PI = 4k7 (5 kl preset
P2 = 10 k linear

C5.C10 = 100 p/16 V
C6 = 10 p/16 V
C7 = 100 n
C8 = 47 n
C9 = 220 p/16 V

Semiconductors:
T1 = BC547B
IC1 = LM 386

Cl = 27 n
C2.C4 = 2p2/16 V
C3 * 22 p/16 V

LI = telephone pick-up coil

LS = 8 Sll'/i W miniature loudspeaker

SI = on/off switch

- 11-33

which is specifically designed for this
type of application. A very low AC
voltage is induced across the coil and
this is amplified by transistor T1 and
the amplifier IC1 and then fed to the
loudspeaker.

There are two ways in which the volume
can be- adjusted: either by using PI to
set the threshold value or by means of
the volume control P2.

A printed circuit board has been de-
signed for the telephone amplifier, the
details of which are shown in figure 2.
Using a miniature Japanese loudspeaker
and a 9 V PP11 battery, the whole
circuit will easily fit into a plastic case
of roughly 120 x 65 x 40 mm. A mains
power supply may also be used, pro-
vided the supply voltage is very well
stabilised, as otherwise there could be
some mains hum.

The construction is very straightforward
indeed and so we can proceed with the
setting-up, which primarily involves LI
and PI. First of all, the best position
for the pick-up coil has to be found.
Ideally speaking, this is underneath the
telephone, but this would mean having
to raise the 'phone a little, since
the coil is about 3 centimetres high.
Another solution is to fit LI onto the
side of the telephone so that it is close
to the amplifier. Readers should decide
for themselves what the best practical
solution is.

Now for the preset PI. This adjusts the
maximum volume. Above a certain
level, the sound reaching the amplifier
input will be so loud that acoustic
feedback ('howl round') will occur. This
is a kind of echo that has got out of
hand and produces a high-pitched tone.
After setting P2 to maximum, PI is
adjusted so that this just does not occur.
It would of course be feasible to omit
all the components to the right of P2
and use Hi Fi equipment to reproduce
the caller's voice, but then, that is up
to the reader. M

1981

siiie-wmv oscillator

very low distortion ....

Nowadays, an entire function generator can be constructed from
a few simple ICs. When measuring low-frequency equipment, such as 1
audio amplifiers, it is highly desirable to use a low distortion, reliable
sinewave generator.

This particular design is not at all complicated as far as construction
is concerned and yet it boasts a distortion level of only 0,01%! Its
frequency range extends from 10 Hz up to an inaudible 100 kHz and is
very simple to operate.

Where modern Hi Fi equipment -
especially the home-constructed
kind — is concerned, accurate measure-
ments are almost impossible to carry
out. Although frequency characteristics
and squarewave signals can be checked,
there is little that can be done towards
measuring the actual distortion level,
that is, assuming the unit is properly
built and the amplifier is working
correctly. Fortunately, most up-to-date
designs are so good, that the distortion
level will be negligible. In any case, it is
hardly worth buying an expensive
oscillator with an extremely low distor-
tion factor and a first-class distortion
meter to carry out one or two measure-
ments.

An oscillator can be built in a number
of different ways, each particular design
having certain advantages and disadvan-
tages. For low-frequency measurements,
where the frequency needs to be varied,
it is best to use a 'Wien bridge oscillator'.
This type of circuit provides low distor-
tion and allows the frequency to be
changed fairly easily with the aid of a
stereo potentiometer or a dual-ganged
capacitor. The design described here is
quite compact and straightforward, but
nevertheless it is eminently suitable for
measuring frequency characteristics
and distortion levels. In addition, a
Schmitt-trigger has been included in the

circuit to provide squarewave output
signals.

The oscillator circuit
Although most readers will probably
know how an oscillator containing a
Wien network works and have reference
books in which they can look it up, it
might not be a bad idea to go into the
matter here.

Figure la shows a network of two
resistors and two capacitors. This
constitutes the frequency-determining
section of the Wien oscillator. If the
transfer function is calculated as U| /U 0 ,
the result will show that there is only
one frequency with no phase shift
between Ui and U 0 - This will be at the
frequency: F = 1/(2.ir.R.C.). At this
frequency the ratio of U| to Uo will be
exactly 1:3. If the U, voltage is am-
plified by a factor of three and then fed
back to U 0 , as drawn in figure 1b, a
perfect oscillator is obtained (since the
U| and U 0 signals are in phase at that
particular frequency). Unfortunately,
however, there aren't any opamps
available with a manufactured gain of
three. As figure 1c shows, this does not
matter, as the solution is to connect the
RC network between the non-inverting
input and the output of an ordinary
opamp. A voltage divider (R1, R2) is

Figure 1 . This drawing shows how a Wien
bridge can be used to create an oscillator.

The voltages U i and Uo will only be in phase
for one particular frequency, where U ) is but
1/3 of Uo- By amplifying the voltage U i by a
factor of three and feeding it back to Uo, an
oscillator can be obtained.

tcillator

november 1981 - 11-35

connected to the inverting input of the
opamp. The R1/R2 ratio is calculated
twice, so that the amplification factor
will be:

R1 + R2 _ 2.R2+ R2_

A u = :

R2

R2

-= 3.

There will now be a sinewave signal at
the output of the opamp with a fre-
quency based on the formula provided
above. In practice, the amplification
factor of three is rather critical, since it
is very difficult to maintain both at the
opamp side and at the RC network side.
If the amplification slightly exceeds a
factor of three, the amplifier's output
will produce an everincreasing output
signal until this is limited by the supply
voltage. The opamp will then produce a
squarewave. If, on the other hand, the
amplification factor drops slightly the
oscillator will either stop operating
or simply not start in the first place.
Then no output signal will be produced
at all. Obviously, some sort of control
system is needed to adapt the gain so
that the circuit oscillates, without
affecting the supply voltage. Only then
will the output be able to produce a
symmetrical sinewave signal.

Usually, such a control system can be
set up by choosing a temperature-
dependent resistor for either R1 or R2)
When the output voltage increases, the

current passing through the temperature-
dependent resistor will also rise, causing
its resistance to be altered. As a result,
the opamp's amplification will be
reduced. If, however, the output voltage
drops, less current will pass through the
feed-back resistors and so the resistance
will change causing the gain to be
increased. This method leads to an
equilibrium, where the output voltage
is constant.

The circuit diagram
Figure 2 shows the circuit diagram for
the sinewave oscillator. It looks quite
different from the block diagram in
figure 1 . The opamp here is discrete in
structure and consists of transistors
T1 . . . T4. The input stage is formed
by a cascode circuit containing a bipolar
transistor (T1) and a FET (T2). In order
to obtain a considerable open loop gain,
a Darlington transistor was chosen for
T3. By means of the current source
formed by T4, the collector is linked to
the negative supply voltage.

The bridge section comprising the
resistors and capacitors is connected
between the collector of T3 and the
base of T1. Use has been made of a
stereo potentiometer with a logarithmic
characteristic to regulate the frequency

continuously. Switch SI acts as a
range switch and provides other
capacitor values. The four positions
provide an overall frequency range of
10 Hz ... 100 kHz, which amply serves
most audio purposes.

The amplitude is stabilised with the
NTC resistor R19. The type chosen
has a resistance of 1k5 at 25°C. This
produces an output amplitude of about
1.5VRMS- (t is ver V important to use
the right type of NTC here, for if the
wrong one is used the distortion level
will rise alarmingly. The one used here
is housed in a glass package with a
maximum power dissipation of 20 mW.
The latter is vital, as current passing
through the NTC has to heat it.

The output signal is fed to P3 by way of
Cl 3 and preset P2 (which presets the
maximum output voltage). This
output level control is followed by an
attenuator which has been drawn
separately.

Switching S2 connects a Schmitt -trigger
in series with the output lead, so that
squarewave signals are also available.
The Schmitt-trigger consists of transis-
tors T5 . . . T7 and the associated com-
ponents. As readers can see, the circuit
has a standard structure and could have
been copied straigth out of a text book.
However, the square wave signals it
produces are of a high enough quality

circuit. Some of the resistors and capacitors shown in figure 2 are soldered directly to the
various potentiometers and switches.

a variable output impedance. The latter
version is much more straightforward.

for audio purposes. The only disadvan-
tage involved in the circuit, is that the
mark/space ratio is somewhat depen-
dent on supply voltage, although
this is really of minor importance in
this particular application.

Figure 3 shows two attenuator circuits.
Usually signal generators feature an
output impedance of 600 Q. The same
can be done here by using the attenu-
ator in figure 3a. In order to obtain an
output impedance of exactly 600 Q
for every stage, the resistors will have to
have rather "odd" values. If, on the
other hand, a slightly irregular output
impedance is acceptable, the standard
values indicated in the circuit diagram
will do. The output impedance will be
around 565 fl. It is only at the highest
output level that the impedance will be
altered between 0 ... 5 k (depending on

the position of P2 and P3). Provided
readers do not require on a standardised
output impedance, they can use the
attenuator in figure 3b. The output
impedance will then of course no longer
be constant, but in most cases this does
not really matter.

Finally, we still have to mention the
LED D1 with its series resistor. This
indicates when the oscillator is activated.
At the same time, the LED ensures that
the circuit's current consumption is
equal for both the positive and the
negative supply if it is battery-powered.
(The current consumption of the LED
has been chosen to be the same as that
of the Sch mitt-trigger circuit.)

Construction and calibration
The easy way out is, of course, to buy

the board and the parts from your local
retailer, solder the lot together and
Bob's your oscillator. The device will
indeed oscillate, but don't be surprised
if it turns out to have a considerable
distortion factor. Obviously, a little
bit more is required. To start with,
R5 . . . R7 should be metal foil types
having a 1% tolerance. The stereo
potentiometer PI needs to be a good
tracking type. Capacitors Cl . . . C4
should also be 1 % types, if available.
This is not absolutely essential, but it
does lead to an accurate scale division
for every range. T1 must be a low-noise
transistor. Nowadays, various Japanese
transistors have an ever lower noise
factor than the types mentioned in
the parts list. A good example is the
2SC2546, but unfortunately this type is
not yet readily available. In addition.

Resistors:

R1.R4 = 22k
R2.R3 = 1 k5
R5 = 560 SI
R6 ■ 100 k
R7 = 330 SI
R8 - 2k2
R9 - 18 k
R10-8k2
R11 - 1200
R12.R15 = 680 SI
R13- 1k2
R14 = 1 k
R16.R18 = 8200
R17 ■ 560 0

R19 = NTC 1 k 5 at 25°C Philips
type 2322 31152
PI = 47 k log stereo
P2 = 2k5 preset
P3 = 2k5 potentiometer

Cepacitors:

Cla.Clb = 1 n
C2a,C2b = 10 n
C3a,C3b = lOOn

C4a,C4b = 1 u (not an electrolytic type!
C5.C1 2 = 1000 p/16 V
C6 = 47 p/16 V
C7 - 470 p

C8.C9.C1 3 = 220 p/1 6 V

CIO = 68 p

C11 =330 p/16 V

Semiconductors:

01 = LED

T1 = BC549C, BC550C
T2 = BF 245B, BF 256B
T3 = BC516
T4 = BF245C
T5.T6.T7 = BF 494

Miscellaneous:

51 = double pole. 4 position wafer switch

52 = double pole toggle switch

53 ■ double pole switch

the voltage at T2 will have to be
measured once the circuit has been con-
structed. This transistor should be a
type that has a drain current of 12 /jA
at a gate-source voltage of -3 V (we will
come back to this later). For this reason,
it might be a good idea to place a tran-
sistor socket in this position on the
board first.

To get back to building the circuit, part
of it is not mounted on the board. This
includes the frequency-determining net-
work at the circuit input (everything
required for connections A . . . C) and
the switch S2 together with potentio-
meter P3 and the attenuator.

In the input circuit the capacitors are
soldered directly to switch SI and
likewise the resistors R1 . . . R4 to the
stereo potentiometer. This section is
then connected to the board by means

Photo 1. Shown on a spectrum analyser
are the oscillator's distortion residues at a
frequency of 1 kHz. The first large peak
represents the 1 kHz signal. The first ever

respect to the 1 kHz signal, or rather at
0.006%. The first odd harmonics con-
tribute 0.01% (-80 dB).

of three wires, after the rest of the com-
ponents have been mounted. Points
D . . . F are then linked to switch S2
and potentiometer P3 is wired. Finally,
the resistors belonging to the attenuator
are mounted straight onto the switch.
Then the supply voltage has to be con-
nected up by way of switch S3.

The power supply may be a straight-
forward mains type, consisting of a
small transformer, a bridge rectifier, one
or two electrolytic capacitors and two
voltage regulators similar to those
published previously in Elektor. Current
consumption will be about 23 mA.
Seeing as the current consumption is so
low, the circuit may also be battery-
powered. Using four 'flat' 4.5 V bat-
teries, the lifespan during intermittent
use will be a reliable 100 . . . 200 hours.
Figure 4 shows the printed circuit board
for the sinewave oscillator. A resistor is
indicated by way of a dotted line next
to transistor T3 . Once the entire circuit
has been built, in the manner described
above, a multimeter is connected
between the two connections of the
dotted resistor and the meter is
switched to current measurement (DC!).
After the supply voltage has been
switched on, the current measured
should be about 15 mA. If it is any
higher than this, a resistor should be
connected in series with the meter and
have a value that makes the meter indi-
cate 15 mA. Depending on the result,
the resistor or, if not required, a wire
link, is soldered onto the board.

The voltage is then measured at T2.
Initially this is measured between the
source and the gate and then the meter
is switched to current measurement and
connected to the drain. Several tran-
sistors of either the B F 245B or B F 256B
type should be experimented with and
the one that comes closest to meeting
the Vqs = — 3V and Id = 12//A require-

elektor november 1981 — 11-37

ment is soldered onto the board.

After this the output voltage can be
measured. This is usually around
1.5 Vrms (measured at the junction
of R7 and C13). If necessary, the out-
put voltage can be altered slightly by
choosing another value for R7. Once
this has been done, P2 is adjusted so
that the output voltage at the wiper of
P3 is exactly 1 Vrms when the latter
is turned fully clockwise. The attenu-
ator can then be used to select a lower
output voltage, 100 mV, 10 mV or
1 mV, which can also be attenuated
with the aid of the potentiometer.

If the squarewave signal observed on an
oscilloscope screen looks slightly asym-
metrical, the solution is to alter the
value of R8. Finally a word of advice:
the output amplitude can be kept at
a constant level by covering the NTC
with a layer of insulating material, such
as foam rubber.

Easier said than done . . .

This circuit is a perfect example of how
an effective device can be built without
the need for a lot of ICs or other fancy
components. All that the enthusiast has
to do is think carefully and decide what
the circuit is to achieve and then select
the best possible components. The
result is a compact design with excellent
features. That is exactly how the
designer created the prototype which
has a distortion level of 0.01% at 1 kHz.
The designer even maintains that dis-
tortion levels as low as 0.0014% at
1 kHz is possible! The frequency range
coveres 10 Hz . . . 100 kHz within
0.1 5 dB. These are very satisfactory
results by any standards. If such a cir-
cuit still is not capable of testing a par-
ticular audio amplifier, the amplifier
must be of such a high quality that
there isn't any point in measuring
anything anyway I M

part tno

teletext

decoder

. . . that does not require modifications to
the TV set

After the detailed introduction to Teletext given last month, this article
deals with the practical side of the project. To keep the whole thing
within manageable limits, the circuit description will be as brief as
possible. This leaves room to pay closer attention to three important
aspects: modifying a TV set if the decoder is to be built in; calibration
procedures; and, last but not least, the instructions for use.

This article concerns the Teletext de-
coder and the circuit-diagram for it is
shown in figure 1. This is a repeat of
figure 13 in part 1, so many readers
may already be familiar with it. The
function of the various LSI chips have
already been discussed in part one,
therefore the operation of this part of
circuit should be fairly clear, at least in
the main outlines.

The ICs 5 through 9 form the external
address counter and page memory (see
SAA5020 of part 1). The remaining
part is the control which is clearly
separated from the decoder in the
circuit-diagram. There are only three
connections between the decoder and
the command/control section, since IC2
can only be given orders in serial form.
This procedure saves 1C pins.

Control (handling)

The layout of the command section is
again divided into two parts (see figure
2) In fact, the real control is carried out
by the keyboard with its encoder con-
sisting of IC17, IC18 and IC19. The re-
maining ICs (IC1 1 . , . IC16) convert the
parallel information of the keyboard,
which is available at points C, D, E,
F, G and H, into a synchronous serial

Figure 2. The >

coder. The shaded

11-40 -

^.-.iruljijiruiriimrinnn^

z — nh|uuiJinjmnnnrLi_

~ |t|t|w|«1w|w|«! i

output. In this context synchronous
means that the clock-signal necessary
for decoding the serial signal is also
transmitted. Figure 3 shows how the
keyboard code is supplied serially to
IC2 (SAA5041). The DLlM-signal is
derived from a 62.5 kHz clock-signal,
which in its turn is produced by dividing
a 1 MHz-signal. This can either be de-
rived from the keyboard itself or from
IC3 (see figure 1). Figure 1 indicates
that the signal from IC3 is derived from
the 6 MHz oscillator (crystal controlled)

R1.R5- Ik6
R2 = 100 k
R3 = 680 12
R4.R15- 1 k
R6.R7.R12- 6k8
R8 = 33 k
R9 . . . R11 = 1k2
R1 3 = 4k7
R14 » 47 k
R16 = 330 n
R17.R18 = 22 k
R19 . . . R21 « 3k3
PI = 1 0-k-preset
P2 = 250-n-preset

Capacitors:

C1,C13,C17= 1 p/16 V Ta
C2.C7 = 1 n

C3.C12 = 10 p/1 6 V Tama
C4 = 330 p
C5.C18.C24

6 = 47 p

100 n

. C22 =

C19 = 68 p/16 V printed C m
C23 ■= 39 p
C28 = 2n2
C29.C30 - 470 n

T1 = BC547B
IC1 = SAA 5030 VIP
IC2 = SAA5041 TAC
IC3 = SAA 5020 TIC
IC4 = 74LS02
IC5 = 74LS83A
IC6.IC7- 74LS161
IC8.IC9 ■ 2114 RAM
IC10 = SAA 5050 TROM
IC11 =4520
IC12 = 74LS123
IC13 = 74LS74
1C 1 4 = 74LS165
IC15 = 74LS132
1C 1 6 = 74LS27

L2 = 33733 (TOKO)

of IC1. So using FI has the advantage
that the transmission speed of the com-
mand switch (DATA and DLIM) does
not have to be trimmed. In order to
make testing of the keyboard possible
without the aid of the Teletext decoder,
the oscillator can be built around N5.
The keyboard is only a local control
for the Teletext decoder so far. Control
over a greater distance using remote
control is possible, but a completed
design is not available yet, although
preparations have been made. The

separation between the keyboard
encoder and the parallel series converter
has already been carried out on the
printed circuit boards. The keyboard
code is fed to the parallel series con-
verter via nine connections. In this way
a selecting circuit can be added, which
makes it possible to choose between
local and remote control.

The converter section is combined with
the Teletext decoder in order to make
it easier to separate the units, as de-
scribed above. This means that the key-

board printed circuit board only con-
tains IC17, IC1 8 and IC19, which will
simplify the installation of this board in
the front of a case. The keyboard may
be used at a distance of up to one metre
from the decoder board. To avoid
problems with interference inherent
with longer distances, buffers must be
added (at the transmitting as well as the
receiving end).

Boards 1 and 2

Figures 1 and 2 give the complete cir-
cuit of a Teletext decoder that is suit-
able for building into a TV set. Figures 4
and 5 show the corresponding printed
circuit boards. Only the shaded part of
figure 2 is shown on board 2 (figure 5).
All the keyboard connections are
brought out together in the form of an
1 1 way ribbon cable between the key-
board and the decoder. With only a few
exceptions, the same is true for the
connections to the video circuit board
(to be described next month).

All boards are 10 cm in width and the
length of the decoder board is 20 cm.
This enables the keyboard and video
circuit boards to the placed beside each
other, near the decoder, allowing the
connections between them to be kept

Modification of the TV set

Although the decoder was not originally
intended to be fitted into the TV set
we will pass on a few hints for readers
who wish to do so.

Only very rarely can the video signal
in a TV set be interrupted. Even sets
with a video recorder connection socket
aren't satisfactory in this case, since
they provide an input or output for the
video signal but the Teletext decoder
needs both possibilities at the same
time.

Most TV sets are still produced without
a power transformer nowadays, so they
may have a live chassis. This problem
can be overcome by the use of an
isolation transformer (figure 6). It is
possible for the transformer to be built
into the TV set without too much diffi-
culty but connecting a video input and
output is slightly more complicated. It
is useless attempting this without a
layout diagram of the TV set. Fortu-
nately a diagram is almost always sup-
plied with the set itself, otherwise it
may be obtainable from the dealer or a
service company.

Figure 7 shows a part of a TV circuit
and the modifications necessary for the
connection of Teletext or any other
video signal. A break must be made
between the video demodulator and
-amplifier. Considering the modular con-
struction of a TV this spot should not
be too hard to find. The signal ampli-
tude should be 2-3 V peak-to-peak. In
the circuit diagram of a TV set this is
usually indicated, as in figure 7, near a
sketch of the waveshape at this point.

1 1-42 -elak tor

fiber 1981

5

Figure 5. The printed circuit board and component overlay of the keyboard section . Since
only the shaded part of figure 2 is placed on this board, it contains just three ICs besides the

Teletext decoder keyboard

Resistors:

R22 . . . R31 = 3k3

Semiconductors:

•Cl 7 = 74LS73
IC18- 74LS148
IC19- 74LS00

21 x digitast switch
(Digitasts serve as wire links)

The interruption of the video con-
nection may influence the AGC (Auto-
matic Gain Control) of the tuners. This
would lead to much poorer reception
quality, or even: no reception at all I
Whether the AGC is influenced (situ-
ation B of figure 7) or not (situation A)
must be checked on the TV circuit
diagram. In the unhoped-for event that
situation B occurs the AGC must be
switched over to a Manual GC (via St>).
The Manual GC control must be ad-
justed separately for each transmitter,
due to the differences in reception.
Fortunately, situation B occurs mainly
in relatively old black and white TV
sets. The video output is easy to con-
nect once the right spot is found.
Figure 7 shows how this is done. The
DC voltage level, on which the video
signal is super imposed, is used to bias
the emitter follower T1 . Via C3 and R5,
the signal can be brought out to a coax
(BNC) connector on the back of the
TV set, or else a connection can be
made directly to the decoder. In the
event that the amplitude of the video
signal is over 3 V, R4 can be replaced
by a trimmer to enable adjustment of
the output amplitude.

The video signal must never be greater
than 6 V. The supply voltage for T1
must also be derived from somewhere
in the TV. Most tuners are connected to
1 2 V, so a supply can probably be found
in that vicinity.

The video input circuit which can be
switched by S a should return the ampli-
tude of the video signal to the desired
value (P3) and super impose the video
signal on the required DC voltage
component. Adjustment of P2 deter-
mines the DC voltage level of the sync
pulses via D1 (see figure 7). With the
component values given, this level can
be set between 0 V and 6 V. If a higher
voltage is required a smaller value may
be chosen for R2.

These modifications are not only
suitable for the Teletext decoder. The
design is such that input and output
impedance are approximately 75 S2.
Therefore a TV set with this input stage
can also serve perfectly well as a video

imber 1981 1V43_

Figure 6. The TV set must be isolated from the mains when it is supplied with a video input
and -output. Generally an isolation transformer can be built in very easily.

Figure 7. Most TV sets can be modified to provide a video input and output. The block diagram
indicates where the video signal must be connected inside the TV. The block diagram of the TV
will either look like situation A or B.

8

Figure 8. The signals from the Teletext decoder mint be mixed in this little video combiner,
before they can be applied to the video input.

monitor, for example with a video
game. In that case a coaxial cable must
be used for the signal connection.

Video combiner

Although the TV set is provided with a
video input stage, the Teletext decoder
signals are not yet fit for this input. The
sync signal from SAA5030 ( 1C 1 ) and
the Y-signal from SAA5050 must first
be combined to achieve this. This is
carried out by T2 and T3 (see figure 8).
The sync signal at pin 12 of IC1 has a
(positive) amplitude of about 0.7 V, so
the base of T3 is short-circuited during
the sync pulses and the output becomes
0 V.

The Y-signal is supplied by one of the
open-drain outputs of IC10. This signal
contains the luminance information of
the Teletext picture. The proportion of
R8 and RIO is chosen in such a way
that the combined video signal is modu-
lated up to approximately 60-70%
white. Since the complete amplitude is
5 V, P3 (at the video input) will have to
reduce the level to 2 to 3 volts.
Switching over from Teletext to normal
reception is simple: Sa (+Sb) is the
selector switch.

With or without frills?

The quickest (and cheapest!) way to
Teletext is to build the simple decoder
described so far into a modified TV
receiver. However, this does severely
limit the flexibility of the system.

In the first place, the Teletext picture
will not be reproduced in colour. This
is not too serious, perhaps, but there is
more: the automatic switching facilities
of the Teletext system are also lost.
Consequently, some useful features like
subtitles for the deaf can not be utilised.
Furthermore, no time indication can be
superimposed on the normal TV picture
and, in general, the Teletext infor-
mation and normal TV picture cannot
be displayed simultaneously. The video
control board (to be described next
month) does add all these facilities
but it is by no means a simple circuit.
The output from this unit can be con-
nected to the video input described
above.

An even more interesting alternative
will also be described. With a little
add-on circuit, it is possible to feed
the Teletext signal into the aerial input
of the TV set. This has the major
advantage that it is no longer necessary
to dig into the inside of the TV set,
provided a seperate receiver section is
also included to provide the video
signal for the Teletext decoder input.
This is no real problem either, as we will
see. In effect, therefore, the basic unit
described so far is suitable for 'standard'
Teletext reception, but the 'frills'
described next month will convert it
into a really interesting project.

This leaves us in a quandary. Readers

11-44 — elektor november 1981

who want all the 'frills' cannot put the
unit into operation until next month;
others, who are prepared to do some
prospecting inside their set want the
calibration procedures and instructions
for use now ... To keep both groups
happy, we will describe the calibration
for the basic unit now, and include the
full 'instructions for use' of the final
version. The latter include the descrip-
tion of some options that are not avail-
able on the basic unit, but this may
serve to whet the appetite!

Basic decoder calibration

Initially, PI, P2 and C9 on the decoder
board should all be set to the mid-
position. In the input/output circuit, PI
and P3 should be set to maximum; P2
sets the DC level, as indicated in the TV
circuit diagram. Hopefully, some kind
of Teletext picture should now appear
- the picture quality is unimportant at
this time. Of the potentiometers in
figure 7, P2 sets the sync level, P3
adjusts the video level and PI sets the
gain of the front-end according to the
strength of the transmitter.

Proceeding now to the decoder board:
this calibration is both simple and
critical - strange but true! The 6 MHz
oscillator is crystal controlled, which
means that the control range of C9 is
limited (±4 kHz). This means that the
line frequency (15625 Hz) can never be
far off. If a frequency counter is avail-
able, the oscillator can be set to exactly
6 MHz when no video signal is applied
to the input. A suitable test point is
line F6 between pin 6 of IC1 and pin 2
of IC3 (see figure 1).

If the vertical synchronisation is in-
correct in Teletext mode (leading to a
'jumpy' picture), this can be corrected
by readjusting PI.

The most important adjustment in the
whole decoder is L2. This coil must be
trimmed up until the decoder synchron-
ises properly on the 'clock-run-in'
bytes of the Teletext signal. It is very
important for the receiver to be tuned
in optimally, in other words that it
gives a sharp, clear (colour) picture.
After pressing the keys 'reset' and
'TXT-nor' the page header or at least
the letters 'PI 00' will appear on the
screen. The core of L2 should be turned
until the time indication becomes
visible. It will be indicated very clearly
within a limited adjustment range and
the correct setting for L2 is in the
centre of this range. It is best to press
the 'reset' key repeatedly during this
calibration in order to wipe out the
nonsense that the decoder displays on
the screen. The decoder should be ready
for use now.

Directions for use

TXT-off Fortunately, it is possible
to switch off the Teletext decoder.
Normally, when the TXT-off-key is
operated the TXT picture immediately
'clears the way' for the normal program.

However, when this key is pressed
during 'mix-mode' the TXT page
remains visible for a few seconds before
it disappears. After that the TV program
can be watched again.

The keys 'full page, half page and
reveal' have no effect on the picture in
this mode. All the other keys im-
mediately bring about a change in the
picture information.

TXT-nor This key activates the de-
coder which means that the normal
mode is selected. The program picture
disappears to 'clear the way' for the
selected page (see description of the
numeric keys).

Mix The program picture doesn't dis-
appear but the Teletext page is dis-
played on the normal picture. This
function can only be reset by the TXT-
nor-key or if necessary cancelled via the
TXT-off-key. In the latter case the de-
coder will always return to the mix
function when a following key is
pressed. Note that this function is not
available on the basic decoder.

Numeric keys Page selection is done
by pressing a three digit number, this
doesn't necessarily require the 'mix-'
or 'TXT-nor'-key to be pressed. The
page header is displayed in a rectangle
on the (normal) screen as soon as a
number key is pressed during the 'mix
mode' or when the decoder is switched
off. The page header remains visible for
a few seconds, disappears and returns
again when the selected page is received.
The decoder indicates the reception by
displaying the page number at the
upper left of the screen. Time indication
will also be visible for a short time. The
page is made visible either by pressing
the 'TXT-nor' or 'mix'-key.

RESET Generally the first page (PI 00)
consists of search information, for
example several references to a more
detailed list of contents, therefore this
page will often be consulted in order to
find a desired page. The 'reset' key
returns the page number register to
100 and wipes out the displayed page at
the same time, only the page header
remains. The counter visible in this
header indicates the number of the
page being received at that moment. As
soon as page 100 appears on the counter
the whole screen is written again.

Timed page A page can be selected at
a specified time. In order to accomplish
this a page number has to be chosen
first and then the 'timed-page'-key has
to be pressed. At the upper left of the
screen T 00.00 will be displayed. The
decoder expects the time setting in
hours and minutes. At the desired time,
the reproduction procedure will operate
as described under 'numeric keys'. The
'timed page' function can be cancelled
by pressing the same key once more.

Full page/half page The Teletext page
consists of rather small letters. In order
to improve the readability, it is there-
fore supplied with the possibility of
dividing the page into two pictures. The
'half page’-key selects one half of the

page. Repeated operation of this key
displays the upper and lower half of
the enlarged page on the screen, in turn.
The 'full page' key is operated in order
to reset the page to its normal pro-
portions again.

Reveal Some Teletext pages contain a
number of hidden information (for
example for video games) which can
be made visible by the 'reveal' key.
However, this information can only be
erased by (re-)selecting a page.

Hold Very often the 24 lines of a TXT
page are not enough to display all the
information about a certain subject,
therefore a page can be extended to a
number of pages. Turning from one
page to another is done automatically
with the aid of a control bit that is
transmitted together with the page.
Each page is displayed for approxi-
mately 25 seconds. However, if this is
not long enough it can be held by the
HOLD-key so that the decoder will not
read any more information. Pressing the
HOLD-key activates the decoder again.
The decoder may also return to the
'mix-mode' in this case (see numeric
keys).

Time/B7 Time display is one of the
tricks that only expensive TV sets are
able to perform. The Teletext decoder
enables time to be displayed during a
normal program. By pressing the keys
'B7' and 'time' simultaneously, time is
displayed in a rectangle at the upper left
of the screen for 5 seconds. If only the
key 'time' is pressed either the complete
page header will be displayed in a black
rectangle for 5 seconds or the decoder
will return to 'mix-mode' for the same
period of time. M

Sources

- The history of CEE FAX, S.W. Amos
(BBC Engineering)

- Broadcast Teletext Specification
(BBC, IBA, BREMA)

- Muiiard Technical Information
No. 72: Multitext

- Valvo Technical Information

No. 800407. Teletext and Viewdata

- Valvo development information
No. 67: The integrated Video-IF-
Amplifier circuit TDA 2541 for TV
receiver

- Grundig Technical Information
4/5-'80

- Siemens Short Form Catalog 1980:
Surface Acoustic Wave Filters

LI OB®

elektor

ar 1981 - 11-45

by David Ridyard B. Sc.

Human Speech

Before describing the techniques
employed in this system it is worth
briefly considering the production and
form of human speech .

The vocal tract can be seen to consist of
a d.c. power supply, square wave
oscillator and a resonator. The lungs
provide a constant air pressure on the
back of the taut vocal folds, or vocal
chords, which forces these two flaps of
skin to open. The Bernouilli forces set
up by the air flow then force the folds
shut, and the result is a stream of glottal
pulses, which may be approximated to a
150 Hz square wave. This means a
fundamental at 150 Hz with exponen-
tially decaying odd harmonics at 450 Hz,
750 Hz etc.

S|H‘ak to your
computer

A simple and flexible speech
recognition system using an
M6800 microprocessor and
involving a minimum of external
hardware, recognising ten words
in less than 1.5 seconds, with
better than 90% accuracy.

The removal of the key pad
interface between man and
computer has many advantages
and has long been the pipedream
of every programmer. By using
this M6800 system it can become
a reality for every home
computer.

The vocal tract itself then acts as a
resonator. If it is compared to a perfect
closed pipe, 17.5 cm long, resonances
would be expected at 500 Hz, 1500 Hz,
2500 Hz and so on. These resonances
are known as formants and the form of
speech is determined by the position of
these formants, which are modulated by
the jaw, the tongue, the larynx and the
pharynx.

Basic design criteria

Most of the 'word' information is
contained in the first three formants,
while 'speaker' information is generally
contained in the higher formants.
Consequently the input signal in this
system uses a second order Butterworth
filter to band limit the signal to less
than 3.8 kHz. This has two advantages.
Firstly, it reduces the speed of pro-
cessing required for analysis and
secondly, it means that the response is
far less dependant on the speaker's

Work on speech has also shown it is
quite intelligible when infinitely clipped,
that is to say that zero crossing infor-
mation is sufficient to characterise a
word. This allows us to use a compara-
tor to reduce speech to a single serial bit
stream, thus eliminating the need for
analogue to digital conversion.

The conventional approach to speech
analysis from this point is to perform a
Fourier transform, and to look for the
positions of formants at various points
in the word. However this is clumsy
and slow, and so in this system the
autocorrelation function has been used:

R x (r) = LIM /+ T x(t)x(t+r)dt

T-~ 2 V-T

This lends itself very easily to micro-
processor implementation, when the
input x(t) is two valued, and 'V re-
presents +1 and '0' represents -1,
since multiplication is performed by the
EOR instruction.

The value of r is determined by the
Nyquist frequency for a 4 kHz input, as
0.125 ms. The limits of integration
were chosen to cover one full cycle
of the lowest frequency of interest,
in this case 250 Hz, since the 150 Hz
fundamental is uniformly present.
This is also convenient since the
data required for each autocorrelation
uses exactly 4 bytes, plus 4 bytes on
either side giving a total of twelve bytes.
In order to reduce the memory volume
per word, only every fourth set of
12 bytes is used. It may be possible to
reduce this further by adjusting the
constant value in line 124, $24(36),
without reducing accuracy. By taking
every sixth set, $3C(60), a 50% re-
duction in recognition time would
accrue, provided a compatible diction-
ary is used.

The autocorrelation of the unknown
word is compared to the autocorrelation
of the words stored in the dictionary,
and selection of the answer word is
made by the minimum Euclidean
distance:

Euclidean distance of answer word = D,

-2 |Xk-Y k |
|_K=0 Jm

j MIN

= number of bytes stored per s
correlated word in store
xk = k th byte of autocorrelated
stored word
Vk = k th byte of autocorrelated
spoken word

Dh = distance of h th choice
In order to reduce incorrect decisions a
test is done, to see if the second choice
word is close to the first choice or not.
(Dj -D,)>A

Where A is a constant, specified in line
252.

In its present form, the system outputs
5 ASCII characters for the first choice,
and 5 for the second choice. If D1 dif-
fers significantly from D2, the words
'I THINK . .' appear as a preface to the
first choice.

’981

The input routine uses a standard PIA,
with the input connected to Bit 7, and
the READY line connected to Bit 0.

Filter and comparator design

This circuit consists of three stages,
preamp., filter and comparator. The
preset PI could be replaced by a fixed
100 k resistance, but best results are
achieved by adjusting PI to give a
maximum peak to peak amplitude of
15 V for normal speech. Conversely R7
and P2 could be omitted, and the
comparator reference input tied to
earth, but best results are achieved by
adjusting P2 to make the output bit
stream spend roughly equal amounts of
time at 'T and 'O'. This reduces false
triggering by noise.

Using the existing system
The first thing to do is to create a
dictionary, by reading in the ten words
required. This is done by taking the
Speech Recognition Program, and
replacing line 135 with a software
interrupt:

020F3F 135: SWI

Words are stored at 2000, 2400, 2800,
2A00 . . 4400 and the required
location is determined by the operand
of line 29.

After the READY switch has been
switched to 5 V, the system will wait
for the first '1' to appear, and will then
sample at 8 kFIz, for 0.84 seconds:
therefore, the speaker should begin to
speak the instant that the READY line
is put high, and the READY line should
be returned to zero as soon as possible
after completion of speech.

The five ASCII characters required for
output should then be placed at
23D3, 23D4 . . . 23D7 for the word
stored at 2000: 27D3 . . . 27D7 for the
word at 2400 and so on.

When a complete dictionary of ten
words has been read in, the original
Speech Recognition Program should
be reloaded and run, and the word
under test spoken, using the READY
switch as before.

The output will consist of the first
choice answer on the left and the
second choice on the right. As a safety
precaution, to ensure incorrect answers
do not get through, the threshold in line
252 should be set up. In this case, if the
answer word does not satisfy the
criterion (3), the words 'I THINK . . .'
will appear on the extreme left. Selec-
tion of the precise value of the
threshold should be made using Graph 1 ,
bearing in mind, that in order to reduce
the number of errors, an increase in the
number of no decisions will result:
02 or 03 are usually adequate.

Better results can be achieved using a
dictionary consisting of words which
have been averaged over several utter-
ances of each word (8 in this case). In
order to do this, a word should be read
in eight times and stored at 2000, 2800,
3000 . . . 5800. The dictionary average
program should then be run and the

output address determined by the
operand of line 118. Do not forget to
add in your five ASCII characters, as
before. Doing this ten times is a rela-
tively slow and laborious process,
involving a lot of work loading and
altering programs from disc or tape
but a significant improvement in results
is noted, and once a dictionary is
completed it can be used indefinitely.
Without using the threshold, but with
an averaged dictionary, 90% accuracy
should be achieved for all speakers of
the same sex. Using a threshold of 02,
the errors are reduced to only 4% or
less, while the decision rate falls to 81%.
Roughly double these errors will occur
if an averaged dictionary is not used.
Initially a new speaker, not used to
timing the switch properly, and over
emphasising his speech may achieve less
than 70%, but after a few runs through
the dictionary he will improve to almost
the same standard as the original author
of the dictionary.

Hook-up in an RCA space
chamber

Technicians at RCA Astro-Electronics, Prince-
ton, N.J., prepare the RCA Satcom Itl-R
communications satellite for a series of tests
in a giant thermal/vacuum chamber. The tests
assure that the spacecraft can operate reliably

earth orbit. Scheduled for launching in late
1981, the satellite will provide communica-
tions for all 50 states. Owned and operated
by RCA American Communications Inc.,
Princeton, RCA Satcom satellites serve the
cable TV industry as well as provide commer-

Unfortunately, almost no success has
been achieved in using male speakers
and a female author or vice versa,
although occasionally a freak set of
results can occur, when a female speaks
at exactly one octave above the male.

Adapting the existing system

The first variable is the decision
threshold, in line 252, which has already
been dealt with in some detail.

The next thing the user may wish to do,
is to extend the dictionary. The number
of words is determined by the operand
of line 157. In order to accomodate
each extra word, two lines must be
inserted between lines 154 and 155, in
such a manner as to suitably extend the
sequence, which should be obvious from
lines 135 to 154; i.e. for the eleventh
word:

154.1: LDX £$4800

154.2: STX $ 1A14

The most important alterations which a
user may wish to perform, concern the
output. The existing system uses the
SSB subroutine, which outputs the
ASCII character stored in the A accumu-
lator. If this does not exist, a similar
routine must be written, and placed
at $ E1D1.

The next alteration may be to ask the
speaker to 'REPEAT', if the criterion 3
is not satisfied. This is simply done by
altering the ASCII characters in lines
254, 256, 258 etc. from 'I THINK' to
'REPEAT'. At the end of this output
section a branch back to line 3, should
be entered.

The output of the first and second
choice relies on the fact that the address
of the first ASCII character of the first
choice is stored at $ 1 B08, and the first
ASCII character of the second choice ‘
stored at $ 1 BOA. These addresses could
be used as the target addresses for
branch instructions, to cause the system
to perform whatever function is
required.

Using the characters 0 to 7, and the
words 'WRITE' and 'STOP', octal
programming should only require
relatively simple subroutines, as should
controlling train sets or remote control
of T.V. sets. The latter two applications
are particularly good, since an
casional error would not be fatal.

Secure cordless phone

Known as The Handshake', the digital
accessing facility from Pace Electronics (UK)
eliminates the security risks which have faced
manufacturers of cordless phones. 'The
Handshake' is a means of providing digital
access between the handset and the base
station thus offering total security to the
phone user. Up to now the major problem
phones has been the ability -

Acknowledgements
The author would like to offer thanks,
for the time and equipment lent to
him by the University of Durham,
Department of Applied Physics and
Electronics, especially Dave Isaacs and
Dr. Tim Spracklen. M

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 £ 5.00 postage and packing inclusive.

(£restV\£iy)

^ 1 ELECTRniSlIRR

WOODHILL LANE,
SHAMLEY GREEN,
near GUILDFORD,
SURREY

THE ELEKTOR
METAL DETECTOR

Kit no. 82021. An advanced metal detector available
for the first time to the home constructor. Pro-
fessional quality ABS vacuum formed case, complete
with 10" ready set up and foamed search head. This
kit can be constructed easily in a few hours. High
performance specification includes phase locked loop
oscillator and discriminator circuitry giving four
reject modes woth ground effect elimination. For full
details and specification see the article in this issue of
Elektor.

complete kit now
available

£89.95

p&p £2.00

Send cheque or postal order to Crestway Electronics
Ltd., write or ring with your Access/Barclaycard acc-
ount no. (do not send your card). Please add 40p to
all U.K. orders for postage and packing. All our
prices include VAT.

24 hour answering service on 0483 893236.

Crestway Electronics Ltd.,

Woodhill Lane, Shamley Green, IujgjjmoI

near Guildford, Surrey. | ws a j

CRESTLEK KITS

Ioniser (9823) Negative ion generator £10.50

Talk Funny (80052) Ring Modulator £10.00

Sound Effects Unit (81 1 12) Guns, trains, etc. £ 8.30
Elektornado (9874) 100 W power Amp . . . £19.50

Top-preamp (80023) Hi-fi preamp £34.40

Guitar Preamp (77020) £ 6.50

2VJ Digit DVM (81105) £23.35

AM Receiver (81 1 1 1) Easy to build £10.25

Touch Switch (81008) 12 positions £10.00

200 W Disco Power Amp (81082) £20.85

STAMP (80543) Mini amplifier £ 3.75

Pest Pester (80130) Insect repellent £ 2.35

Aerial Booster (80022) £ 5.65

LATEST KITS.

TV Games extension (81 143)

Disco light controller (81155)

DFM + DVM (81156)

RF test generator (81150)

EPROM programmer (81594)
Frequency counter (82026)

Teletext decoder (82025)

The ELEKTOR MINI ORGAN (82020)
write or phone for details of these kits.

in our 0
• CATALOGUE •

320 big pages packed with #

V data and pictures of ^
fk 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 $ price £1.

In case of difficulty check the coupon below.

make it easy...
with

*•

w a ^ w Easy to build,

9 w w superb specification.

Comparable with organs selling for
Up to £1,000. Send for our new
book giving full construction
details, order as XH55K
:e £ 2,50 inclusive.

* Same day service on in-stock lines

* Very large percentage of our stock lines in sck

* All prices include VAT

$ Large range of all the most useful componets

* First class reply paid envelope with every oier

* Quality components-no rejects— no re man

* Competitive prices

* Your money is safe with a reputable compay

On price, service, stock, quality and secuntyjtiakes
sense now more than ever to make I
first choice for components every time!

Post this coupon now. E 11 8 1

Please send me a copy ol your 320 page catalogue 1 end® £1 25
(md, 25p pip). II I am not completely satisfied I may relui the
catalogue to you and have my money refunded. II you live itside the
U K. send £1.68 or 12 International Reply Coupons

I ELECTRONIC SUPPLIES UD. ^

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