selektor

contents

pest pester

■ Is your body itching for summer sun, while twitching at the thought of
those dreaded holiday fiends: mosquitos? Do you find a couple of nights
'in the bush' are enough to turn your skin into a lunar landscape? Not to
mention the sleepless hours spent wrestling with the sheets in a vain and
exhausting effort to shut out that sky-diving drone? This article describes
a circuit to solve your insomnia!

disco lights controller if. o p -t Eyndel

Disco colour light systems have been on the market for some time now.
The modules have certain advantages: they’re not expensive, they're easy
to handle, and above all, they're safe. There is however one disadvantage:
since the module is a single unit it cannot be expanded. This may be
remedied by adding an extra module, as described in this article.

£2 aerial

Readers with an interest in short wave reception, such as DXers, often
have difficulty in finding a suitable aerial.

The Elektor design staff have been working on this problem for some time.
The result is an aerial which can be placed practically anywhere, an active
aerial, which can compete with much bigger and more expensive types in
the 1 .8 MHz to 30 MHz range.

luxury transistor tester ir. stomi

timbug II

The initial design requirements for this 'bug' were that it be able to 'see'
objects in its path and take avoiding action. It should also be as inexpen-
sive and as simple as possible. As the circuit was designed around the ever
popular 555 timer 1C, and the circuit shown here is the Mk II version, it is
not difficult to realise how it came to be christened Timbug 1 1 .

j coming soon .

i musical cube

EDITOR:

EDITORIAL STAFF

TECHNICAL EDITORIAL STAFF

J. Barendrecht
G.H.K. Dam
P. Holmes

E. Krempelsauer
G. Nachbar
A. Nachtmann

K. S.M. Walraven

narrow band I F strip

i more on TV games

| In earlier articles (T played TV games', Elektor October and November
1979) we asked for our readers experiences with the joysticks - and

[ promised to come back on the subject when we had enough data. We
received several interesting reactions to this request, often including other
suggestions and comments. By now, we feel that it is high time to bring all
other interested readers up to date!

variable logic gate im. v. Kerkwijk)

I applikator: smoke detector

noise at high frequencies

Noise in UHF/VHF receivers can be determined by using extensive and
expensive test equipment. However, tests with a noise generator can give
usable results at a much lower cost.

morse trainer

the SC/MP as a mini organ ih.w. wyes) .
the Elekterminal width extender iw. Me

more on junior computer and fuel consumption meter

missing link

toppreamp, elekterminal, active car aerial

market

advertising index

elektc

1980 - 6-01

A look on the bright side

How would you like to knock 10% off
your electricity bill? It can be done, by
using new lamps recently introduced by
Philips!

The basic idea is quite simple. It is well-
known that fluorescent lamps (TL', in
Philips parlance) are more efficient than
normal filament bulbs. So what do you
do? You take a miniature fluorescent
lamp, fold it up, mount it in a bottle
(with choke, starter and all), and add a
standard bayonet (or screw) cap that
fits into a normal light socket. The
result (figure 1 ) is a direct replacement

1

4 1

power rating ('wattage')
light output
efficiency

incandescent
75 W

900 lumen
1 2 lumen/watt
1000 hours

Philips

■SL'

18 W

900 lumen
50 lumen/watt
5000 hours

reflector
40 W

900 lumen
< 25 lumen/watt*
2000 hours-

£ 2.50 .

. £ 5.00

cost of lamps for

5000 hours £ 3.25 1= 5 x)

electricity cost

(at 3.43 p per kWh) £ 1 2.86
total cost. 5000 hours £ 16.1 1

£ 6.25 ... £ 1 2.50 (= 2)4)

for conventional lamps, that uses only a
quarter of the electricity to give the
same amount of light.

There is rather more to it, of course, but
before going into technical details it is
interesting to see what the new lamps
can do. An 18 watt ‘SL’ (for 'Super
Lamp'?) gives the same amount of light
as a conventional 75 watt bulb, and it
has a useful life of more than 5000
hours - five times that of the filament
type. It gives the same type of light as a
normal lamp - not that nasty 'cold'
white often associated with fluorescent
lamps - and it starts after only the
briefest of pauses, without all that
irritating on/off flickering. All in all, a
good and efficient replacement for
conventional bulbs. But what about the
price? Hold on to your seat: about
C 7.50 each!

Which brings us to the economics. You

can look at this in all kinds of way
(energy saving, pounds-out-of-pocket,
practical examples), but in each case the
new lamp turns out a winner!

The energy aspect is illustrated in
figure 2. The total energy used in an
average household is shown at the left
(assuming that heating, hot water
supply and cooking run on gas or oil),
and the electricity consumption is given
in greater detail at the right. The shaded
portion at the top of the 'electricity'
column is what can be saved by using
the new SL lamps instead of the conven-
tional kind: more than 15%, in this
case!

All right, so you can save energy. What
about saving pounds — with a lamp that
costs ten times as much as the normal
type? This requires a little calculation,
as shown in Table 1, to compare the
total costs when using a normal filament

Total energy Electricity only

(electricity, gas.etc.)

64)2-1

1980

lamp, the new SL, or another new
alternative proposed by Durotest and
General Electric in America (amongst
others) : a fairly conventional filament
lamp with a heat-reflective coating
inside the glass. In each case, the total
cost over 5000 hours is calculated —
based on an electricity price of 3.43 p
per unit (kWh).

"A

Even though the calculations are
weighted against the SL (we have yet to
find the filament lamp that lasts 1000
hours in a normal domestic light fitting,
and the life expectancy and light output
of the lamp-with-heat-reflector represent
the theoretical maximum that might be
achieved!), the new lamp still wins
hands down. Furthermore, as the price
for electricity goes up and the cost of
the new lamp comes down, the differ-
ence will become even more striking.
Philips give two practical examples that
also shed an interesting light on the new
lamp. First, as far as saving energy is
concerned:

Around 1500 million incandescent
lamps are sold in Europe annually.
Approximately half of these are for
domestic use, the rest going to
'professional' sectors (industrial, hotels
and restaurants, schools, etc.). Let us
now assume that, after a certain period,
10% of the consumer market and 25%
of the professional is occupied by the
new SL lamp. This means that, in all,
250 million incandescent lamps have
been replaced by SLs - saving approxi-
mately 14.5 thousand million kilowatt
hours each year, worth roughly £ 500
million! To put it another way, every
family in London could light their home
free of charge for 7 or 8 years, with this

3

2. Discharge tube

3. Fluorescent powder

4. Choke/Ballast

5. Electrode

6. Bi Metallic strips

7. Starter

8. Mounting plate

9. Housing

10. Thermal cut-out

11. Capacitor

12. Lamp cap

A further example illustrates the pounds-
in-pocket principle. In an average hotel,
with 4000 incandescent lamps installed,
about 75% of these could be replaced
by 18 W SL lamps. This would give a
total saving of some £ 30,000 each year!

A look inside

A cutaway view of one of the new
lamps is shown in figure 3. As can be

seen, the fluorescent lamp itself is
folded into a double U-shape, and the
choke is mounted between its legs.
Obviously, folding up a fluorescent
lamp is no easy feat — especially since
the fluorescent powder must be applied
to the inside of the glass tube before it
is heated and bent. Admittedly, Philips
have a lot of experience in circular and
W-shaped lamps, but even so a com-
pletely new coating had to be developed
for this particular application. The
fluorescent layer had to be 'strong'
enough to withstand the severe bending
required; it had to produce the same
'colour' as a normal incandescent lamp;
and it should have the highest possible
efficiency. Apparently, the designers
have succeeded: the lamps are now in
production!

A further problem is associated with the
design target: a direct replacement for
normal lamps. This limits the permissible
size and weight rather drastically! As far
as size is concerned, they are already
quite close - as shown in figure 4. The
diameter of the SL (18 W version) is
72 mm, as opposed to 60 mm for the
equivalent 75 W filament lamp; the SL
is 160 mm long, compared to 108 mm
for its conventional counterpart. Not
bad, certainly when you consider that
the SL can often be used without a
'decorative' lamp shade.

The weight is another matter: 520
grams for the SL, and only 35 grams for
a normal filament lamp. Over a pound!
However, Philips assure us that this is
still within the weight limit set by inter-
national standards for lighting fittings.
One other difference, when compared
to normal lamps, should be noted:
SL lamps can not be operated off lamp
dimmers! As with any other fluorescent
lamp, they don't take kindly to a mains
supply that has been chopped up by a
thyristor or triac.

elektof june 1980

Some other lamps

At the same time. Philips have intro-
duced some other new types of fluor-
escent lamp. A thing that looks for all
the world like two lamps glued together
at top and bottom (figure 5) goes by the
name of 'PL' ('Paired Lamp'?). It is even
more efficient than the SL — a 13 watt
version gives the same amount of light
as a conventional 75 watt bulb — and it
gives the same 'warm' light.

Then there is the 'TLD' (the D stands
for 'Dun', the Dutch word for thin!),
once again available with the same
'incandescent lamp coulour' — referred
to as colour 82 (figure 6).

However, enough is enough: there is
little to be gained by discussing these
other types in depth.

557S

Design and the silicon chip
The world's first major exhibition
recalling the origins of the silicon chip
and tracing its development and influ-
ence in design is now open at The
Science Museum, London. Intended to
entertain and inform all age groups, the
exhibition will continue until the end of
1980. Co-sponsored by the Science
Museum, the Design Council and the
Department of Industry, this is probably
the first exhibition of its type to show
the significance of the silicon chip in
improving the quality of life for every-

Visitors to the exhibition will enter the
first of the two halls devoted to The
Challenge of the Chip' through a display
featuring a microprocessor which is
60 times life size. Exhibits in this hall
have been assembled by The Science
Museum and deal with the history of
the silicon chip and show its devel-
opment to meet increasingly sophisti-
cated electronic requirements.

The Design Council has selected
products and systems which are impress-
ive examples of silicon chip applications
to show in the second of two halls.
Products from about 60 companies are
displayed in ten sections covering
shopping, offices, transport, communi-
cations, production and control, edu-
cation, music, medicine, home and toys.

In the future one microprocessor may
control the heating and ventilation
systems in the average home as well as
lighting, cooking, washing, radio and
television equipment. In addition, it
could answer the telephone, maintain a
record of telephone charges, control an
alarm system, operate the curtains and
do all the household accounts. For the
present time designers have concentrated
on utilising the inherent reliability and
low cost of the microprocessor to
improve existing designs of domestic
appliances. The world's first computer-
ised washing machine is on display and
it uses a microprocessor to provide a
more comprehensive range of washing
and rinsing programmes to suit the
variety of natural and man-made fibres
in current use.

There is a section of the exhibition
devoted to microchip controlled games
and toys which will appeal to children
of all ages. A popular exhibit will be the
multi-loop railway with multi-control
system. Computer games including chess
and bridge are also on show.

Small computers, using microelectronic
circuits, are now being used to help
doctors diagnose complaints and store
the medical records of their patients.
Some examples of this equipment are
shown together with new instruments,
using microchip devices, which can help
identify handicaps such as deafness in
new born babies.

Another special exhibit shows the
enthusiasm with which designers at
British Rail are pursuing potential
applications for the microprocessor to
bring about improvements in passenger
travel through the more efficient issue
of tickets and the provision of better
passenger information facilities. The
proposition that robots can be used to
do all heavy work on the railways, keep
trains and stations clean and move mail
and parcels automatically from road to
rail is being investigated. The movement
of coal from the mines to the point of
bulk use is already well established as an
industrial conveyor belt but British Rail
explains that the microprocessor could
make the 'merry-go-round' completely
automatic by driving the train on its
closed loop journey to load and unload
the coal, weigh it and produce all the
necessary documents.

Fuel injection systems were one of the
first of an increasing number of silicon
chip applications for the motor car. A
special Lucas Electrical display shows
how the microprocessor can be used to
conserve fuel, control exhaust emissions,
monitor driving instruments and control
the power delivered to the road wheels.

Teaching machines

The ubiquitous pocket calculator is
probably the best known and the most
widely used microprocessor inthe world.
No attempt is made in the exhibition to
show the tremendous variety of designs
and functions of these machines.
Instead there is a derivation shown
which helps children learn simple arith-
metic and another, based on a simple
calculator, designed to encourage an
interest in spelling. An ingenious
information centre into which different
memory capsules can be plugged to
retrieve several subjects as well as
different languages is shown to demon-
strate the versatility, compactness and
general utility of the microprocessor as
a teaching aid.

The musical chip

Microelectronics is helping to produce
and reproduce music in better ways. A
typical electronic organ which can
simulate the sounds of many instruments
and do so with a variety of accompani-
ments and in different rhythms illus-
trates the tremendous advances made in
organ design since the development of
the microchip. The music synthesiser is
another device shown to demonstrate
the remarkable possibilities which the
silicon chip has brought to music.

6-04 - el 6

pest pester

Is your body itching for summer sun, while twitching at the thought of
those dreaded holiday fiends: mosquitos? Do you find a couple of
nights 'in the bush' are enough to turn your skin into a lunar landscape?
Not to mention the sleepless hours spent wrestling with the sheets in a
vain and exhausting effort to shut out that sky-diving drone? Read on
for the circuit to solve your insomnia!

In many respects, summer can be a
mixed blessing. It's wonderful to step
out of the dark, dreary days of winter
into the bright sunshine with holidays,
parties and picnics to look forward to.
At the end of the day, after a soothing
shower, you slip between the cool
sheets and then . . . you are rudely
interrupted by that most infiltrating
insect: the mosquito.

Your worries are now over! Elektor's
designers have suitably sized up the
situation and have come up with a
sizeable solution: the Pest Pester. It
couldn't be simpler or smaller. Any
mosquito on the rampage will buzz off
immediately upon hearing the circuit's
squeak.

It is a welcome change to have such a
simple, yet effective circuit fill the pages
of the leading article, instead of the
highly complex computer systems
which usually get the honour. The Pest
Pester consists of exactly nine com-
ponents all told. Before dealing with the
circuit's construction however, it might
be a good idea to see what we're up
against. How do mosquitoes 'tick'?
Mosquitoes:

Their Habits and Idiosyncrasies

It is common science that certain high
frequency noises keep annoying insects
at bay. So there's nothing new on that
score. Every now and then the would-be
'inventor' of an electronic mosquito
chaser allows his name to be splashed
across the headlines. Invariably, though,
it all boils down to the same principle.
Unlike in the past, when mosquitoes
were swatted or sprayed regardless of
their gender, occupation or creed, these
undersized public enemies are now
going to be dealt with in a biological
manner. That is to say their private lives
have assumed a new significance, for
apparently, although they all buzz, only
the females sting. Thus, these are the
ones against which to take strategic
action. Nature gives us another helping
hand by narrowing the foe down to
several million mothers-to-be. These
have been discovered to avoid their men
like the plague (you might have thought
that the damage had already been done,
but then who are we to judge the wiles
of nature?). The obvious solution is
therefore to reproduce the male's buzz,
thereby making the bedroom a safe
place to sleep in.

The next thing to consider is the fre-
quency. All frequencies between 1 and
30 kHz were tried and the best results
were obtained around 5 kHz.

Does it work?

Can mosquitoes really be 'buzzed off' so
easily? The next best thing to asking a
mosquito about it, was to talk to a
parasitologist. Our man was highly
sceptical about it and even went so far
as to say that certain tones would attract
mosquitoes, rather than keep them
away. We put it down to parasitical
pessimi sm.

elektor june 1980 — 6-05

P”« P e5ter

Another highly effecitive method is to
use a blue light to attract mosquitoes
towards a chicken wire screen where
they meet an excruciating, high voltage
death. Electrocution is a cruel solution.
Previous high frequency devices in-
cluded a 'bat simulator' (bats are
renowned mosquito eaters) which,
unfortunately, fooled no-one. Why use
your hard earned money to buy a 'Pest
Pester' when you could make it so easily
yourself? Admittedly, its effectiveness
remains to be proved. On the other
hand, its ineffectiveness remains equally
unproved. In other words, you have
nothing to lose and will probably have a
lot of fun in the process.

It is hoped that upon reading this article,
hobbyists will eagerly produce their
soldering irons and send in their
empirical experiences to our editorial
staff. Who knows? You might be the
one to come up with the ultimate
frequency!

The circuit

Being so small and simple the 'Pest
Pester' needs very little explanation.
The circuit must be able to run for
extended periods of time on one penlight
cell (AA), it was decided.

Considerable time was spent in dis-
cussing various IC's and supply voltages,
when, convinced that actions speak
louder than words, one designer built an
astable multivibrator (AMV) with two
transistors. It used a speaker out of a
telephone headset and a penlight cell for
power. It worked so well that even at
0.7 volts it continued to oscillate
(a remarkable feat in itself).

Figure 1 shows the schematic. Using the
given values, the oscillating frequency is
approximately 5 kHz. As mentioned
before, this frequency seemed to be the
best, but it may be changed of course
by replacing R2, R3, Cl, and C2 with
appropriate values.

Figure 1. Schematic diagram of the Pest
Pester. The telephone earpiece produces a
5 kHz tone.

A few more particulars — you may have
noticed that C2 is four times larger (in
value) than Cl. This causes the output
to have a duty cycle of around 25%.
This is all quite deliberate because there
will be many more harmonics in the
output than there would have been at a
duty cycle of 50%.

The speaker (it must be a crystal type)
is connected between the two collectors
of the transistors. This may seem a little
strange at first, but it allows the output
swing to be double the supply voltage.
Some may recognise this as a sort of
bridge amplifier — it is. The 'Pest
Pester's' current consumption is
extremely low, using only 300 p A. This
means that with a penlight battery
(which is usually good for 500 mA/hr)
the 'Pest Pester' will torment mosquitoes
for 1500-2500 hours! Specifications
like this should make them cringe.

The printed circuit

This time we’re offering two printed
circuit boards for the price of one. This
gives hobbyists as much freedom as
possible when thinking about a case. It

R1.R4- 10 k
R2.R3- 560 k
Capacitors:

Cl ■ 82 pf
C2 = 330 pf
Semiconductors:

T1.T2-BC 547
Miscellaneous:

Crystal earpiece/speaker from a
telephone headset.

comes in a round as well as a rectangular
version. They are simply cut apart (they
are sent as one board). Figure 2 shows
both printed circuit layouts along with
component positions.

All kinds of cases are suitable. The
prototype was mounted in an old
'glue-stick'. The case should be large
enough for the battery. For the glue-
stick case, a 2 mm bolt was soldered to
the copper base at the heart of the
circuit board. This served as the negative
battery contact.

The positive battery contact was
inserted in the lid of the stick. The
speaker and circuit board were placed in
the bottom of the glue-stick with the

battery above. The positive lead was
attached to the inside and was fixed to
the top edge so that by turning the cap
the 'Pest Pester' could be turned on and
off. In our particular case, the glue-stick
was a little too small for the speaker and
the crystal had to be removed. Fortu-
nately, this caused no problems.
Figure 3 gives an illustration of this. M

Figure 3. The entire unit may be mounted
into an empty glue-stick.

6-06 -i

june 1980

disco lights controller

A disco light system consists of groups How it works
of coloured lamps turned on and off to

the rhythm of the music: red lights for Disco lights operate from an audio
the low notes, yellow for the middle signal. This means that, in order to
notes and blue for the high notes (of control a set of disco light, some sort of
course any colour may be used for any audio signal will have to be generated,
tone). In this way, the music becomes a Every time the red (low) lights are to go
visual as well as an aural experience. on, a low frequency tone will have to be
Even though this effect is fine, it comes generated, for the yellow lights (middle)
nowhere close to that produced in a a medium frequency tone and the blue
Discotheque. The lights must be able to lights (high) will need a high frequency
do more than just flash with the music, tone. For this reason, the disco light
This disco light module gives that little controller generates three tones: 50 Hz
bit extra. (low), 500 Hz (medium) and 2 kHz

It is a simple circuit with which an (high).

disco lights controller

'lights' to face the music

Disco colour light systems have
been on the market for some time
now. All you have to do is
connect an audio signal to three
coloured lamps and you're ready
to throw a party. The modules
have certain advantages: they're
not expensive, they're easy to
handle, and above all, they're safe.
There is however one disadvantage:
since the module is a single unit it
cannot be expanded. This may be
remedied by adding an extra
module, which this article will
now describe.

F. Op 't Eynde

existing disco light system may also The block diagram of figure V shows the
flash in succession (running lights) or three tone generators. All three generate
leave a 'space' in succession (inverse continuous squarewaves at the three
running lights) or on and off (beacon), frequencies desired. The electronic
The running speed or the flashing switches determine which of these
frequency may be varied. This unit is (if any) reaches the output. Each of the
designed for three channel colour generators has its own switch. The tone
systems and will work on home-made or mixer also has one with which the
commercial models. controller can turn all the lights on and

1

Figure 1. Block diagram of the disco light controller. The device generates special tones, so that
it may be used with a conventional disco light system.

off for the beacon function.

The electronic switches are controlled
(opened and closed) by the low fre-
quency squarewave oscillator (LF) and
the run counter connected to it.

The circuit

Figure 2 shows how figure 1 is put into
effect. It is a very simple circuit with
four CMOS IC's along with a few other
'bits and pieces'. The three tone
generators constitute a NAND gate
(from low to high: N1, N2 and N3
respectively) each. To get such a gate to
oscillate, a resistor and a capacitor are
added. The four electronic switches are
all in one 1C, so that is easy enough —
ESI , ES2 and ES3 are the electronic
switches for the running light and ES4 is

for the flashing light. The two mixers in
the block diagram are also very simple,
each consisting of only three diodes. D1 ,
D2 and D3 mix the tones together to
produce the light signal, and D4, D5 and
D6 sum the output of the electronic
switches to provide the running light
output.

The low frequency squarewave generator
is made using NAND gate N4. Its
construction is similar to that of the
three tone generators. The only differ-
ence here is that the resistor has been
replaced by a fixed resistor and a
potentiometer for varying the oscil-
lation frequency.

The signal produced by this generator
not only controls the flashing light
switch ES4, but also feeds the clock
input on the run counter IC3. The run

counter has four outputs 0, 1 , 2 and 3.
The '3' output is connected to the reset
input on the 1C. This tells the counter
to start counting over again when it
reaches three.

To provide the 'moving hole' effect the
three outputs of IC3 are inverted. This
is achieved by the EXOR gates N5, N6
and N7 controlled by switch S2.
Inversion takes place when the control
inputs of the gates are taken high, to
+5 V, by switch S2. Switch S3 selects
the operating mode; normal (using the
audio input), flashing light (the well
known beacon effect) and 'running'
lights. Potentiometer P2 has been
included to allow adjustment of the
output amplitude for matching the
sensitivity of the disco light system in

Readers with an interest in short
wave reception, such as DXers,
often have difficulty in finding a
suitable aerial. The aerial they are
really looking for is either too
expensive or their town council
refuses permission to place it. And
a small aerial often suits the bill
but not the reception. Moreover,
there are few types of aerials
which enable the entire short wave
range to be received.

The Elektor design staff have been
working on this problem for some
time. The result is an aerial which
can be placed practically
anywhere, an active aerial, which
can compete with much bigger
and more expensive types in the
1 .8 MHz to 30 MHz range.

The genuine DXer has a hard time these
days, what with community aerial
systems and local laws and restrictions.
Shortwave aerials are often regarded as
eye-sores — and unfortunately, they
often are!

However, there are many kinds of
aerials and usually one can be found
which can be used indoors or outside
without getting in everybody's way.
Reception quality of course greatly
depends on local reception conditions.
People who live in flats, for that matter,
are always at a disadvantage, with so
little breathing and thus 'receiving'
space. Since electromagnetic waves do
not penetrate steel reinforced concrete
well, a rod aerial will always have to be
placed outside the building. Such an
aerial will produce only a marginal

the best of their ability.

Of course, there are many other kinds
of SW aerials, but they all manage to
have one or more of the disadvantages
already mentioned. An aerial which
reacts to the magnetic component of
the field, therefore, always has at least
one advantage when compared to the
others. One of these is the magnetic
loop aerial, but this is hardly ever used
due to its lack of efficiency. A type
universally adopted in transistor radio's
for medium wave reception is the ferrite
rod aerial. It is also used as a directional
aerial for navigation or military pur-
poses.

Small loop aerials

The magnetic loop aerial is small with

atrial

active window aerial
(patent has been applied for)

signal. Of course this might be remedied
by adding an aerial amplifier to the
paraphernalia of wire and metal, which
in turn will add to the noise. Further-
more, the aerial is highly sensitive to,
and will pick up all sorts of, man-made
noise (QRM). It cannot be made to pick
up only a weak station.

The aerial may be of varying lengths. As
a 1/4 lambda aerial, the smallest (tuned
to resonance) aerial, it tends to be fairly
long, especially for lower frequencies
and in addition is tuned so that the
bandwidth will be limited. The more
conductive the earth under it is, the
better it works as the earth serves as the
dipole's counterpart.

DXers with gardens can improve the
electrical quality of the earth under the
aerial by digging trenches, fanning out
away from the aerial and embedding
copper wire in charcoal in them. When
this is done the ground under the aerial
will have to be kept continually moist.
All in all, a job not to be taken lightly.
To the flat dweller such an installation
may seem nothing more than a far
fetched fantasy and he will have to
make do with a tiny rod aerial. The
length of the rod aerial will be many
times smaller than the smallest wave
length to be received and is therefore
reactive (needs to be tuned to the
required frequency). Now an amplifier
will definitely have to be brought into
the picture and fitting it between the
aerial and the receiver is bound to cause
quite a few problems. To make matters
worse, numerous electrical appliances
are used in a block of flats, all of which
conspire in interfering with reception to

regard to wave length, so that its energy
pick-up is quite low. Yet, these aerials
are an attractive proposition. To start
with, they have a polar diagram in the
shape of a figure of eight as shown in
figure 1. It can be seen that very sharp
zero points appear, for from certain
directions reception is virtually nil.

As its name suggests, it works on the
magnetic rather than the electrical
component of the electromagnetic field.
This penetrates concrete more success-
fully before reaching the aerial free of
interference. In blocks of flats especially,
it will prove to be an asset.

One disadvantage is that the loops in
use up to now have not been very
successful above 7 MHz. Elektor, how-
ever, has chosen this type to work on.
Considering its advantages, it is strange
that so little interest has been shown in
it. After all, its drawbacks (low ef-
ficiency and narrow bandwidth when
tuned) should not be too difficult to

It is placed in a magnetic field. With
regard to the electrical field this is
turned 90° as shown in figure 2. The
aerial is therefore positioned vertically,
standing perpendicular to the magnetic
field as a loop. A voltage is induced in
the loop which causes current to flow
through the aerial and to the receiver.
This naturally produces another
magnetic field around the loop, so that
it can operate as a receiver as well as a
transmitter aerial. Part of the energy
received is therefore beamed out again.
You could say that part of the energy
received seems to be dissipated in a
resistance. This is called radiation

1980-6-09

Uind.

Figure 2. An electromagnetic wave consists of two main components, an electrical field end a
magnetic field. This figure shows that they are 90° out of phase to each other.

resistance and varies according to the
aerial used.

If the average value of the radiation
resistance is calculated for a loop aerial
of, say, 40 cm in diameter, this turns
out to be less than one tenth of an ohm
at 30 MHz, in other words a negligible
amount.

The aerial has two kinds of resistance:
load and material resistance. The latter
may be considered in series with the
radiation resistance. Since the resistance
of a rounded conductor of 2 it x 40 cm
is hardly worth mentioning, the result
obtained is the substitute shown in
figure 3. The voltage source represents
the induced voltage in the aerial, L
stands for the inductance of the aerial
and Rb for the load resistance. By
means of a fairly complicated math-
ematical calculculation, it may now be
established that the smaller the induct-
ance the greater the current passing
through the aerial. At the same time,
the greater the flux contained the
greater the current. Thus, it is safe to
say that the aerial with the highest
possible 4>/ L ratio is the best one to use.
Once we had got that far, finding the
right shape for the aerial was chicken
feed. Since this had to be a question of
trial and error, however, a few consider-
ations had to be dealt with first. The fre-
quencies which we are concerned with
are fairly high, so the 'skin' effect will
arise to a certain extent. (This means
that the current will mostly flow to the
outside of the conductor) . This being the
case, a solid rod of copper will have no
more effect than a hollow drainpipe. In
addition, the fact that the current will
pass through the outside of the
conductor will really make it immaterial
whether the conductor has a tubular
form or not. In fact, it could be
flattened out, thereby creating a thin,
flat conductor.

A few measurements proved this point.
Hardly any difference in self-induction
between the thin sheet copper and a
hollow tube or massive rod was noted.
Thus, the obvious conclusion was to
stick to the thin sheet copper for further
measurements, since this can be bent
into all shorts of shapes. The results of
tests carried out on various shapes are
given in table 1 . Note how a broad loop
aerial (14) produces better results than a
large, narrow specimen (10). The ratio
of the aerial loop's surface area to self-
induction was used as a criterion.
Another point of interest in this table is
the fact that six loops wired in parallel
(25) also produce very low self-induc-
tion. This may be explained as follows.
If two coils are wired in parallel, the
value of self-induction will be halved.
But this will only occur when the coils
do not affect each other and do not
generate mutual induction voltages.
When broad foil is used, there will also
be a number of coils switched in parallel,
but these do affect each other. This is
only partly prevented by foil, with the
result that the selfinductions present are

more significant than is the case in 25.
The ideal distance between loops has
been found to be about a tenth of the
diameter of the loops. Nevertheless,
copper foil is the best choice, because
the aerial then takes up less space and is
easier to construct.

There are two types of magnetic loop
aerial: a resonant and a non-resonant
type. This is determined by whether the
aerial has been tuned or not. The
resonant type uses the layout in figure 4.
As a capacitor has now been placed in
parallel to the load, the reactance of the
aerial is 'tuned out'. It should be noted
that whether or not one half of the
power is re-radiated depends on the
type of matching, i.e. power matching
noise matching. The advantage of this
type of aerial is that more power is
available as compared with the untuned
version.

A drawback is that the aerial is narrow
banded and must therefore be tuned. If
it is to be placed in the attic or on the
roof this will have to be done by remote
control, which is easier said than done.
Secondly, the current through the aerial
will be 90° out of phase with the flux,
thus with the <J> in figure 2.

An advantage, on the other hand, is that
its transmitter and receiver character-
istics are equal, so that a fairly mobile
transmitter/receiver aerial is obtained
with the directional characteristics of a

The small loop aerial is non-resonant.
This means that the transmit and receive
capabilities are not the same. As the
aerial we are looking for has to suit the
average short-wave listener, this is of
minor importance.

The £2 aerial

For every unit of length any conductor
will have a certain amount of inductance
and capacitance. Usually, the capaci-
tance is disregarded, but since a loop
aerial's width is equal to its length, it
will have to be taken into account here.
Let's take a look at the replacement

layout. Since inductance can be
established per unit of length, it can be
assumed that the layout will look like
figure 5. It now follows that load
resistance Rq should be as small as
possible, because every bit of the aerial
will preferably be circuited with its own
impedance. Ideally speaking, the loop
should be a short circuit. If R(3 is as
small as possible, the figure will be fairly
symmetrical. Kirchhoff's law may be
applied (the sum of all currents to and
from a point is nil). Then, the sum of
the currents in point A will be nil, in
other words: the capacitance will
have no influence whatsoever. The
capacitance concerned will undoubtedly
come up at the Rb connection points. If
sheet copper is used, the ends which are
connected up should be cut to a point,
so that two points come to face each
other rather than two broad areas (see
figure 6).

Optimally, the loop will have to be very
small with regard to the smallest wave
length to be received in order to obtain
a highly homogenous field within the
aerial. A loop having a diameter of 1/10
lambda has a nice, homogenous field,
but a rather weak signal. It is therefore
advisable to use an amplifier as well.
This must be virtually free of noise with
a very low input impedance and be as
well matched as possible to the first
receiver stage. If necessary, a less
homogenous field will suffice and the
diameter may be increased to 1/4 of the
smallest wave length to 2.5 m covering
the range to 30 MHz. Such an aerial will
also partly react to the electrical field,
but in any case generates a large enough
signal for it to be connected to the
receiver directly by means of a 50 - 70 £2
cable.

The active £2 aerial
Gradually we are getting to the crux of
this printed matter! After all, what it is
all about is how to construct an aerial
which is suitable for short-wave listeners

(SWL's) and easy to set up. We have
opted for the non-resonant magnetic
loop aerial with amplifier. It will be
small in size, easy to build and as good
as any of its larger counterparts. As
table 1 has shown, it should be round in
shape. As far as its material is concerned,
a corrugated aluminium strip three cm
wide is suggested. An advantage of
corrugated aluminium is that its surface
area is greater than you would imagine
from its width. This of course has
nothing to do with the fact that a
broader loop gives better results.

The aluminium strip is bent into a loop.
The diameter must be less than 1/10 of
the smallest wave length to be received.
Figure 7 shows a broadband noiseless
aerial amplifier. Use has been made of a
very quiet transistor: a BFT 66. To keep
the noise factor down to a minimum a
grounded emitter configuration has
been selected.

There are a number of conditions which
the amplifier needs to comply with. A
well known problem with broadband
amplifiers is that they are prone to
overloading, for instance by local
transmitters. If such a transmitter is in
the neighbourhood, distortion in the
amplifier stage may cause the signal to
mix with the other two signals and to
produce a mixture product within the
tuning range of the receiver. As a result,
'stations' are heard where they do not
exist, and existing weak stations are
inaudible. This can be avoided by using
an amplifier with a wide dynamic range.
Furthermore, the amplifier's bandwidth
will have to cover the entire short wave
range and, of course, the noise it
produces itself will have to be negligible.
At a collector current of 9 mA the
BFT 66 has its maximum dynamic range
(approximately 60 dB). Resistors R1,
R2, R3 and diode D1 take care of the
bias, resulting in 9 mA of collector
current. The (unby-passed) emitter
resistor R1 creates a small amount of
feedback, improving the amplifier's IMD
properties at the expense of the noise

n aerial

elektor june 1980 —6-1'

Figure 7. The broedband amplifier for the n -antenna.

figure. If a larger aerial loop than 50 cm
is chosen, the effect obtained will be
partly nullified. Since the collector and
base impedances are highly reactive
oscillations are likely. Thus is it import-
ant to work out the component layout
in such a way, that the connections
between the various components are as
short as possible. In addition, the input
and output will have to be as far away
from each other as possible.

In the design of the amplifier a compro-
mise was reached, this being that the
aerial is not low-impedance terminated.
The result is that the signal production
at lower frequencies drops by 6 dB per
octave. This is no disaster, because as
the noise at lower frequencies increases
up to 20 dB per octave, the net result or
signal to noise ratio will at any rate not
deteriorate. This means that the
receiver's dynamic range meets more

8a

Figure 8a. The power supply for the antenna. When SI is switched in the lower position, the

input is attenuated by 20 dB. The open arrow shows the input from the amplifier and the

shaded arrow represents the output to the receiver.

8b ^

Figure 8b. To avoid any possibility of oscillations, the power supply is placed between the
amplifier and the receiver. The two coils in the above drawing consist of 10 to 20 windings of
0.2 mm enameled copper wire.

flexible demands than when an active
rod aerial is applied, which was seen to
amplify the total signal + noise factor at
lower frequencies as well.

The amplifier printed circuit board has
been designed to form a unit with the
aerial. The supply for the amplifier,
which has been included on a second
printed circuit board, can be connected
to the amplifier by means of a coax
cable (see figure 8a). It is advisable to
place the aerial several metres away
from the receiver to reduce any risk of
oscillation to a minimum and care must
be taken that the receiver and aerial are
not placed on the same metal base.
(Oscillations may be recognized as an
inordinate amount of noise emitted by
the receiver). In figure 8b a method is
shown to eliminate oscillations by a sort
of balun between the output of figure
8a and the receiver.

Last but not least . . .

After reading the information provided
in this article, it should be possible to
build a good aerial which can be put up

in any house or flat. In cases of great
difficulty, copper or aluminium foil
may be chosen and the aerial may be
fixed flat against the inside of a
cupboard door. The aerial is then direc-
tionally mobile and will not be in
anyone's way. One thing which must be
taken into account is that any metal
surfaces must be removed from around

the aerial. Thus, if your windows have
metal frames, it is not a good idea to fix
the aerial to the pane. Then it is indeed
better to use a cupboard door. Of
course, by applying several loops, the
aerial's directional effect may be in-
creased. One way is to place two loops
next to each other (keep in mind that
the individual distance must be at least

1/10 of the diameter of the aerials to
keep coupling between them low). If
the SW receiver has a battery supply, so
that DXing may be continued out and
about, it is possible to derive the supply
from the receiver or use a battery
exclusively for the £2 aerial. The supply
voltage of the amplifier may be between
4 V and 12 V. H

6-14 -elektor june 1980

luxury transistortester

This particular design first appeared on
Elektor's pages in last year's Summer
Circuits issue. Readers voted it as one of
the most interesting circuits and this
article is the elected Elektorised result.
The circuit has been slightly modified
and a printed circuit board now
accompanies it.

In technical literature, the current
amplification is usually indicated as hpE-
For everyday purposes it is not absol-
utely necessary to know the precise hpE
value, but rather to have a rough idea of
its upper and lower limits. The manufac-

l c (collector current) and lt> (base
current).

The Luxury Transistor Tester indicates
the letter corresponding to the transis-
tor's hpE category. Thus an A, B, or C
will appear on the seven-segment d isplay.
An 'F' will appear, if the transistor is
faulty. The circuit has separate connec-
tions for NPN and PNP transistors. A
switch selects the transistor type.

The block diagram

Figure 1 shows the block diagram of the

luxury

tester

transistor

A highly important aspect of
transistors is their current
amplification ratio. This is often
indicated by an A, B, or C behind
the type number in the case of
transistors. Inevitably however
(by decree of Murphy himself),
this is no longer legible. Using the
transistor tester described here,
the correct letter can be read from
a display. At the same time, it can
also be determined whether the
transistor is up to scratch or not.

R. Storn

turer used to have no way of precisely
determining the current amplification
ratio in advance. The best he could do
was make a rough estimate, then after
the transistors are manufactured, they
were selected to meet the required hpE
limits. The type number was then
printed on the case. Although nowadays
this can be determined in advance, the
same type numbering is still used. Two
transistors with the same type number
do not necessarily have the same hFE-
That is why industry uses a letter as a
suffix to indicate the general hFE value.
The letters define the hpE according to
the following values:

'A' for an hFE between 1 40 and 270
’B' for an hpE between 270 and 500
'C' for an hpE of more than 500
The terms hpE and current amplifi-
cation ratio describe the ratio between

transistor tester. Its operation is quite
simple. The voltage across a number of
resistors is compared to a reference
voltage. Here it is important to know
beforehand whether the transistor is
NPN or PNP. The switch that selects the
transistor group also operates an LED to
indicate the position of the switch. This
voltage comparison determines the hFE
group of the transistor and displays an
'A', 'B', or 'C' whichever is appropriate.
If the 'F' on the display does not
disappear when the pushbutton is
depressed, then the transistor is defec-

The layout

The complete layout is given in figure 2.
Also shown is the parts list. The schmitt
triggers in the block diagram consist of
three op-amps wired as comparators.

Figure 1. The block diagram of the Lux

Tester.

The upper half of the schematic,
IC1 - IC3, serves to measure NPN
transistors. The inverting inputs of the
op-amps are connected to a reference
voltage. The non-inverting inputs are
connected to the collector of the
transistor under test (TUT). Resistors
are used as voltage dividers here. The
base drive current is determined by R1
and RIO. At a certain amplification
factor the collector current will also be
fixed. Then the three collector resistors
will be under a voltage determined by
the current amplification ratio and the
value of the collector resistor. If the
amplification factor is 400 and the base
drive is IOjiA, then the collector current
will be 4 mA. With this amount of
current flow, the voltage dropped across
the collector resistor R4 (390 J2) will be
1.56 V. Three collector resistors have
been included and they all have a
certain voltage dropped across them. In
the given example, R2 (220 fi) has
| 0.88 V and R3 (180 ft) has 0.72 V. As

said earlier, R4 has 1.56 V dropped
across it. This makes calculating the
voltages at the 1C inputs easy. The
inverting inputs are all at the same
potential (or voltage). The voltage
at the TUT's collector will be
9 V - 3.16 V = 5.84 V (the 3.16 V is the
sum of the voltages across the resistors
that feed the non-inverting inputs and
the 9 V is the supply voltage). The
reference voltage at the inverting input
is 8.02 V which is determined by R5,
R6 and R11, R12. In the earlier stated
example, therefore, IC3's output will be
low along with IC2's. Only ICI's output
will be high. This is shown by this
simple calculation:

9 V (supply) - 0.88 V (voltage at pin 3)
= 8.12 V. 8.12 V is higher than the
8.02 V reference. If S3 is in the NPN
position a B will appear on the display.
If the output of IC1 were also to go low,
the display would then be C. This would
be correct as the voltage drop across the
resistors would have risen as well as the

current through them. The base current
in this circuit being always the same, the
higher collector current could only be
due to a higher current gain.

If, on the other hand, the outputs of
IC1 and IC2 were high, only segment d
would not light and so an 'A' would
appear on the display. Segments a, e and
f are always on because they're used in
all of the various letters that are
displayed. All of the above is on the
presumption that the transistor is not
faulty, for if it is, then an 'F' will appear.
This occurs only when all the 1C outputs
are high - when the reference voltage is
higher than the collector voltage of the
TUT.

The display control (the circuit
consisting of T1, T2 and T3 along with
resistors R15 . . . R19, R24 . . . R26 and
diodes D3 . . . D5) works quite simply.
If the outputs of IC2 and IC3 are low,
segments d, b, and c of the display are
lit. The common anode of the display is,
of course, always connected to the +9 V

124.R25.

supply.

IC1 controls the three transistors. If
ICI's output is high, only T2 will
conduct so that segment g is connected
to ground. Conversely, if ICI's output is
low then only T1 and T3 will conduct
with the result that segments b, c, and g
are connected to +9 V and these
segments go out.

A similar situation occurs when S3 is
changed to the PNP position. The
outputs of IC4, IC5 and IC6 are then
connected to the display instead of IC1,
IC2 and IC3.

Construction

In figure 3, both sides of the printed
circuit are shown. To make construction
as easy as possible, the display and
switches have been included on the
board. Even the transformer can fit on

it if a board mounting type can be
found, otherwise a little tinkering may
be necessary. The connections between
the supply and the circuit itself have
been deliberately omitted. This makes it
possible to cut off the supply portion of
the printed circuit board and mount it
above (or anywhere else for that matter)
the main board. The entire unit can
be mounted in a Verocase type 502
(75 - 3960 E) or similar. Figure 4 shows
how this is done.

Switch S3 is a 4 pole 2 way and, if
desired, can be attached to the printed
circuit board. For this, a hole may be
drilled into the board and the tumbler
can be inserted without any difficulty.
If this is done it should be possible to
make a slot in the case's lid so that S3
may be operated. The switch's connec-
tions must be wired to the circuit board.
On the printed circuit the various

connections have been marked in the
same way as the switch in the parts list.
Pushbuttons used to interrupt the base
drive of the TUT should be of the
digitast type. Below S2 are the connec-
tions for the PNP and below SI those
for the NPN. The pin assignment code is
C = collector, B = base and E = emitter.

The 1C op-amps are the popular (and
inexpensive) 741 type. There is however
one minor disadvantage to this, six IC's
are necessary. By avoiding the use of 1C
sockets (not needed in this case) costs
can be kept to a minimum. It is however
advisable to use a socket for the display.
The transistors to be tested should,
ideally, be connected to the board by
means of clip leads. If this proves
impossible, a transistor socket may be
used, but this has shown disadvantages
in practice. K

timbugll

The initial design requirements for this 'bug' were that it be able to
'see' objects in its path and take avoiding action. It should also be
as inexpensive and as simple as possible. As the circuit was designed
around the ever popular 555 timer 1C, and the circuit shown here is the
Mk II version, it is not difficult to realise how it came to be christened
Timbug II. By using just four timer ICs and a handful of other
components a quite 'intelligent' and lifelike animal can be made.

Rather like a bat, the bug transmits a
'radar beam' of ultrasonic sound which
will be reflected by any obstacle in its
path. Once this reflected signal has
been detected the bug will alter its
course. It does this simply by reversing
a short distance while turning to the
left or right at the same time. If the
path in front of the bug is now clear
it will move straight ahead thereby
avoiding any obstacle. If, however,
another object is detected the bug will
continue to 'wriggle its way out' by
turning to the left and right alternately.

Circuit diagram

The complete circuit diagram of the bug
is shown in figure 1 . An oscillator with
a frequency of approximately 40 kHz
is formed by the circuit around I Cl .
The output of this oscillator is fed
directly to an ultrasonic transducer to
provide the 'radar beam' mentioned
earlier. Preset potentiometer PI is used
to adjust the oscillator frequency to suit
the particular transducers used.

Any reflected ultrasonic signal is picked
up by the circuit around IC2. The
internal comparators of this 1C are
biased so that any significant change at

elektor june 1980 — 6-19

the input (pin 2) is detected and, as the
1C functions as a window discriminator,
a large voltage swing is produced at both
outputs. One of the outputs is con-
nected to an LED (D1) which will
appear to be 'on' when a signal is
being received — it will of course be
turned on and off at the same rate as
the input frequency. The 'sensibility'
of the detector circuit is determined by
the setting of P2.

Resistor R7 and capacitor C4 provide
a simple filter for the second output
of the detector circuit. As soon as a
reflected signal is detected this output
will go low thereby discharging C4 and
turning off transistor T1 which, in turn,
will turn on transistor T2 to activate
relay Rel . This relay has two sets of
changeover contacts which are wired so
that when operated they will reverse the
voltage polarity to the drive motors.
Rel will remain activated until the
voltage on C4 reaches a level high
enough to turn on T1 and so turn off
T2. Due to the time constant of R7/C4,
the relay will remain activated for about
two seconds after the detected signal
has gone, that is when there is no longer
an obstacle in the path of the bug.

The circuit configuration of IC3 is
similar to that of the transmitter (IC1)
but, as the values of the components
are much larger, the frequency of
oscillation is much lower. With the
values shown, the period of the oscil-
lator is approximately 9.8 seconds.
The output of this oscillator is fed to
yet another timer 1C (IC4) which is
connected simply as an inverter. This
means that the outputs of IC3 and
IC4 provide two low frequency signals
which are 180° out of phase with each
other. These two outputs control the
'left' and 'right' relays (Re2 and Re3
respectively) each of whose normally
closed set of contacts are wired in series
with one of the drive motors. Diodes D5

2

and D6 are included so that there
can be no feedback between the two
outputs which could cause both relays
to be activated at the same time.

As it is, the relays can only be activated
when transistor T4 is conducting
which in turn is controlled by T3 and
T2. The end result of all this is that,
when a reflected ultrasonic signal is re-
ceived by IC2, relay Rel is activated and
at the same time either Re2 or Re3 is
also activated. Therefore, the bug will
reverse and turn in the direction dic-
tated by the state of the low frequency
oscillator IC3. If, of course, IC3 changes
state while T2 is still conducting, the
direction of turn will also change —
making for more interesting and life-
like results.

Construction and setting up

The circuit for Timbug II can be in-
corporated into virtually any model
which has two drive motors - one for
each wheel - and a single castor type
front (or rear) wheel. As direction is
controlled by the two drive motors, the
circuit may even be built into a tank.

As can be seen from figure 2, the
contacts of relay Rel are wired so that
when the relay is activated the voltage
polarity to the drive motors is reversed.
The normally closed contacts of relays
Re2 and Re3 are wired in series with
the motors. Thus, when none of the
relays are activated both motors will
run in the forward direction. When Rel
is activated the motors will reverse but
as only one motor will run — Re2 or
Re3 will also be activated — the bug will
turn away from the obstacle.

The setting up procedure for the unit
could hardly be simpler and requires
no special test equipment whatsoever -
not even to adjust the transmitter
frequency! Initially P2 is adjusted so
that LED D1 turns ON and then care-
fully readjusted so that the LED turns
OFF but is close to the point of coming
ON. With an object placed a few centi-
meters in front of the bug's 'eyes', PI
is adjusted until LED D1 lights. P2 may
require some readjustment but normally
the object can be moved further away
and PI adjusted until the required
'seeing distance' is obtained. M

coming

soon

summer

circuits

The next Elektor is the July/August
'Summer Circuits 80' issue. It contains
over 100 projects and design ideas. This
means that our design staff has to dream
up as many new circuits for one issue as
would otherwise suffice for the whole
year. Some are based on application
notes, others on ideas sent in from
readers, but all are interesting or excep-
tional in one way or another. The editor
demands that the circuits should be
'new', 'original', and/or 'different'. The
head of design demands that they work,
and the deputy editors for the various
editions demand that the components
should be available.

From missile attack games to melody
makers. From video pattern generators
to wind detectors.

Some are basic design ideas, others are
completed circuits. Some come with
printed circuit layouts too. All promise
to be interesting!

more Ilian
MK) circuits
in

one issue !

6-20 - elektor ju

mnM cube...

Having a somewhat weird sense of
humour, a certain member of the Elektor
design team (who shall be nameless) has
come up with this rather novel circuit.
When friends come round for drinks and
a chat it is no longer necessary to stare
at each other in total silence while one
of you thinks of something to say.
With the musical cube sitting on the
coffee table you have a ready-made
conversation piece. 'What's that?' they
say. 'What's what? you say. That thing
there!' they say. That's my pet musical
cube' you say, 'he sings!'. Say no more,
the evening is off to a flying start. You
then go on to explain that you are the
only person able to control its rather
nasty temper. To prove it you talk
quietly to the cube and it will 'sing' its
reply. You then pick the cube up and
move it to a different place in total
silence. The guest is then invited to move
the cube back in a similar manner. The
cube, of course, not being used to the
new scent, will complain bitterly.

How is it done? Easy, four of the five
sides (not the base - even a cube has
to have something to sit on!.) are
touched in a certain sequence. As the
owner/trainer of the cube knows the
sequence there is no problem. As
each side is touched the cube will
produce a tone and when all four sides
have been touched in the correct order
the cube will remain silent. If, however,
one of the sides is touched out of
sequence the cube will produce a
horrible noise to show its disapproval.
The odds against a newcomer hitting
upon the correct sequence at the first
attempt are, of course, very high.

Circuit

The circuit diagram of the musical
cube is shown in figure 1. It may seem
a bit complex at first sight but its
operation is fairly straightforward. It
works on the 'vicious circle' principle.
Initially everything is reset. The outputs
of N1, N3, N5 and N7 are all low, while
the outputs of their counterparts (N2,
N4, N6 and N8) are all high. As IC1 is
reset each of its outputs is low — note
output '0' is not used. Because all inputs
to the EX-NOR gates (N10 . . . N13) are
the same (low initially) all of their
outputs will be high. This in turn
means that the output on N14 will
be low. The Q output of IC3a is low
so that the oscillator formed by N19
will be inhibited. The output of N19

will of course be high so that N20 is
enabled.

Now for the juicy bits . . . Each of the
points marked A . . . D are connected
to four of the sides of the cube (we
count the top of the cube as being one
side). The cube can be made from
single or double sided copper clad
board suitably etched to provide a
touch sensor. Each of the connected
sides has to be touched in the sequence
A, B, C, D. It is left to the constructor
to decide which of the sides will corre-
spond to A etc. When the first side is
touched the flipflop formed by N1/N2
will change state. The output of N2
will go low providing one of the inputs
of N9 with a negative going pulse via
C4. The output of N16 will therefore
go low for the same duration. This
output has a dual function. Firstly, it
triggers IC4a which, via N17, removes
the reset from the oscillator formed by
IC2. Secondly, via N20, it provides a
clock pulse for IC1. This means that
the first output of IC1 will go high
turning on transistor T1 so that IC2
oscillates at the frequency determined
by the values of R13 and C7. Both in-
puts of N10 will now be high so its
output is still high and the output of
N14 will still be low. The network
R10/C5 takes care of propagation
delay problems and ensures that IC3a
is not triggered at this time. The 555
oscillator (IC2) will produce a tone
via the loudspeaker for as long as the 0
output of IC4a remains low - just less
than half a second with the values
shown.

The same will happen when side B is
now touched with the addition that the
first flipflop is reset by the output of
N4. And so on along the line until side
0 is touched. This being the last side in
the sequence it is assumed that having
got this far everything is OK. The
output of N15 will now go low and
C6 will start to discharge. When C6 is
sufficiently discharged the output of
N18 will go high and trigger IC4b. This
slight delay is incorporated so that the
system will not reset before the last
tone is heard.

Both outputs of IC4b are used to reset
the entire works. The 0 output resets
all the flip-flops, while the Q output is
used to reset the counter (IC1). This
reset condition lasts for around 10
seconds — more than sufficient time for
the owner/trainer to move the cube.

So far so good. We have dealt with 1
correct operation, but what happens
when one of the sides is touched out of 1
sequence? The counter will still be
clocked and a tone will be produced
from the 555 oscillator — albeit very
briefly. However, the inputs to two of
the EX-NOR gates (it doesn't matter
which two) will now be different.
This means that their outputs will go
low taking, in turn, the output of N14
high. As soon as C5 is sufficiently
charged (a few /is) IC3a will be triggered.
The 0 output of this monostable
performs the same function as the 0
output of IC4a and that is to remove
the reset from the 555 oscillator via
N17. The Q output however removes
the inhibit from the oscillator formed
by N19 which means that IC1 is now
clocked rapidly. Transistors T1 . . . T4
are switched on in turn so that IC2
produces rather a horrendous noise.
The clock rate, and therefore the noise,
can be adjusted by means of the preset,
PI. When the delay time of IC3a runs
out, IC3b is triggered which in turn
triggers off IC4b via N18 to reset the
whole system once again. The circuit
can of course be extended so that all
sides have to be touched or indeed
some of the sides touched more than
once. However, we leave that to the
discretion of the constructor — enough
is enough!!

The inside story for
Owner/Trainers.

It will already be apparent that the
secret of successful fireside training
of the musical cube is knowing which
sides are to be touched and in which
order. Bearing in mind that, besides the
four sides, the top is also a 'side', this
making five sides in all (the side the
cube resides on is, of course, the er . .
bottom).

It is up to the constructor/owner/trainer
to decide the sequence of side touching
but some discrete visual aids might
prevent the O/T from getting side-
tracked. Patterns etched in the copper
sides will be the answer (and be an
added aside to the conversation). As
long as the side connections coincide
with the circuit inside, the cube should
stay on your side. Owners must be
warned — we know of one cube that
commited suicide by going up in a cube
of flame, possibly due to ill treatment. M

6-22 - elektor june 1980

Several things have to be taken care of
first. The IF signal, which is produced
by the front-end of the receiver, must
be 'purified' or 'cleansed' of as much
unwanted noise as possible so that only
the pure IF signal is left. This is done
with the aid of a filter circuit. The cir-
cuit here uses crystals; which is known
to be one of the best methods to achieve
high selectivity.

The filter circuit is designed for an IF of
9 MHz. An advantage is that use may be
made of popular 27 MHz '3rd octave'
crystals. These are easily obtainable and
what's more, at a reasonable price.

The pure 9 MHz signal could be ampli-
fied and then demodulated, but it is
better to have an intermediary stage and
derive a second IF with a much lower
frequency (130 kHz) from the 9 MHz
IF. In this way a 'double' super
heterodyne circuit is achieved with two

narrow band
IF strip

an IF amplifier/demodulator
for amateur receivers

An amateur receiver for the two metre (144 MHz) band or for the
citizens-band (CB; 27 MHz) is quite different in construction to that
used by the broadcasting industry. It is not the quality of the
reproduction that determines whether a particular receiver is good or
bad, but its selectivity, sensitivity, etc. In other words, how universally
applicable is it? This article describes part of a receiver and how a low
frequency signal may be obtained from a narrow-band AM or FM IF
signal.

1

important advantages. First, better
suppression of the various spurious
signals is obtained. A second mixer —
provided they occur at a frequency at
which the filter operates — will partially
remove them. A relatively narrow
bandwidth is obtained at low frequencies
using LC filters. Another advantage of
the 'double-super' is that the 'lion's
share’ of the signal can be amplified at a
relatively low frequency. This makes
amplifier design and construction much
simpler and less critical, because it is
less susceptible to oscillation and inter-
ference.

Would it be possible to derive an IF of
130 kHz right away? No, because the
image frequency would be very close to
that of the input signal required and
would therefore be difficult to filter
away.

Block diagram

The operation of the circuit is put into a
nutshell by the block diagram in fig. 1.
The 9 MHz signal is filtered in a
network employing crystals and is then
amplified slightly in a mixing circuit
(MIX). The 9 MHz signal is combined
with an oscillator signal of 8.87 MHz.
This produces a difference signal of
130 kHz. The signal is then amplified
and fed to both an AM and an FM
detector from which the actual low
frequency signals are derived.

When the 130 kHz signal is amplified,
a voltage is derived that is proportional
to the input signal and used to drive an
'S' or signal strength meter.

Although the 'double super' is usually
considered to be a luxury, costs have
been kept to a minimum by using
inexpensive 27 MHz crystals and by
reducing the number of active com-
ponents to a couple of uncomplicated
IC's and two transistors.

«R°

0*OH

if

I l

+\ FM K)

8.87

MHz

©

S

L_J

Figure 1. Block diagram of the Narrow Band IF receiver. It is a 'double-super heterodyne'
design. The IF signal of 9 MHz is mixed with an oscillator signal of 8.87 MHz to create a
difference signal of 130 kHz. It can accept AM and FM modulated inputs.

t

The 9 MHz Filter

Figure 2 shows the crystal filter for the
IF signal from the front-end. It is a
totally passive circuit, which only means
that it has no amplification of its own.
As has been shown before, it is better to
filter an IF signal thoroughly first and
then amplify it before using the 'old-
fashioned' method of an amplification
stage followed by a filtering stage
followed by an amplification stage and
so on. The main reason for this is that if
an RF signal is amplified too early all
the unwanted signals may overload the
amplifier stage. It is advisable to filter
them first, even though this may weaken
the desirable part of the signal. After all,
it can always be amplified later on.

The input and output of the crystal
filter are both adapted to the standard
high-frequency impedance of 50 ohms.
This is achieved with the aid of two very
simple home-made HF transformers.
Detailsare given alongside the schematic.
The input transformer is not critical. Its
purpose is to change the impedance to
that required by the following trans-
former which is selective. This trans-
former is really a 10.7 MHzFM IF filter,
but it has been converted for the
frequency desired here. This is done by
means of capacitor Cl. It is in parallel
with the capacitor built into the trans-
former. This increases the total
capacitance which in turn reduces the
resonant frequency. Because of the need
to adapt the transformer, no other type
of Trl, other than that indicated, may
be used.

The lion's share of the filtering is done
by crystals XI . . . X4, all of which are
'3rd harmonic' types, meaning that they
are supposed to resonate at the third
harmonic of their rated frequency. This
is 9 MHz. Between XI and X2 and
between X3 and X4 a network has been
inserted to prevent undesirable im-
pedance jumps. L3 and L4 are ordinary

Figure 3. These curves show the effect of the
crystal filter and the difference C2 and C8
make.

figure 3b shows filter operation with

C2 = 10 and C8 = 5p6; finally, for figure 3c

miniature coils with a value of lO/tH.
Tr2 has been converted from a
10.7 MHz transformer like Trl. Here C9
is in parallel with the built-in capacitor.
Do capacitors C2 and C8, which are in
parallel with XI and X4 respectively,
affect the incidence of the slope? They
can be left out if necessary. Ideally
speaking, they should be replaced by
trimmer capacitors with a value between
2 and 22 p. Then the filtering may be
trimmed until the slope is as steep as
possible. Unfortunately, quite a lot of
complicated and expensive equipment is
required to optimally set such a trimmer.
So the trimmer will not be optimally set,
but it comes quite close. Figure 3 shows
that this doesn't present any problems
because the filter continues to work
well, regardless of the slope. Even if the
worst comes to the worst (figure 3a) C2
and C8 are left out altogether, attenu-
ation will still be at least 50 dB. An
improvement for a start would be to
give capacitors C2 and C8 set values (see
figure 3b). The slope (a function of fre-
quency vs. U 0 ) will then be about 3.
Figure 3c shows how the crystal filter
works at its best, when C2 and C8 are
replaced by optimally set trimmers.

From 9 MHz to 130 kHz

As can be seen, the circuit in figure 4
bears a close resemblance to the block
diagram. It amplifies the 9 MHz signal,
is amplified and mixed with the
8.87 MHz oscillator signal to produce a
difference signal of 130 kHz. This is
filtered and then amplified. In the
mixing process a signal is derived to
drive the S meter (to indicate relative
strength of the aerial signal). Further-
more, the actual AM detection also
occurs in this part of the circuit.

The circuit has been constructed around
a single 1C; the TCA 440. This chip

made it possible to design an inexpens- 1C. These are the inputs of an amplifier
ive, and simple, single 1C mediumwave stage. The amount of amplification is
receiver. determined by the voltage at pin 3.

By means of a home-made transformer. Later it will become apparent where
(coil data given with the schematic) the that comes from. The amplified 9 MHz
9 MHz signal reaches pins 1 and 2 of the signal arrives at a multiplier used as a

Figure 5. The one transistor amplifier/filter for the AM output.

mixer. The other input signal of the
mixer originates from an oscillator
which generates a signal of 8.87 MHz.
This is done with the aid of crystal X5.
This may be an 8.87 MHz crystal but a
3rd harmonic crystal of 26.600 MHz
may also be used, as was done for the
crystals in figure 2.

The signal of one of the mixer's outputs
is used to derive the second IF signal of
130 kHz. This is done with the aid of
transformer Tr3. Like the other two
transformers in figure 2, it will be a
'converted' transformer. This time it's
a 455 kHz IF transformer of which the
resonant frequency has been reduced
to 130 kHz by adding C15.

The signal across the resonant circuit
(the secondary coil is not used) is fed to
a second amplifier section in the
TCA 440. This really consists of three
amplifiers in parallel, thereby consider-
ably increasing the signal strength. The
signal must not be limited, especially
where AM detection is concerned,
because the low frequency information
is in the amplitude. For this reason an
automatic amplifier gain control has
been incorporated. This works as
follows: the output signal of the second
amplifier section is rectified by D1 and
C18; this produces a direct voltage
which has three purposes. They are:

- To control amplification in the first
9 MHz amplifier stage.

— To control amplification in the
second amplifier stage.

- To provide a signal for the S meter.
The 130 kHz signal must now be fed to
the FM detector. Before this however, it
will have to be filtered once more using
the earlier mentioned 455 kHz trans-

' former Tr4. Potentiometer PI is re-
quired to adjust the amplitude of the
signal to a level which the FM detector
can handle.

Since the 130 kHz signal is being
rectified, (for the automatic gain
control) we already have a detected
AM signal at our disposal. This can be
derived from Dl's cathode; as given in
the schematic. Attention should be paid
to the fact that D1 is a germanium
diode, (for instance an AA 119 type)
not a silicon diode.

The detected AM signal is amplified in
the single transistor amplifier stage
shown in figure 5. Any remaining
130 kHz signal is filtered out using the
simple RC network R7/C21. Poten-
tiometer P2 controls the level of the low
frequency output signal.

FM demodulator

One of the best ways to demodulate an
FM signal is to use a phase lock loop

(PLL). What it really comes to is having
a voltage controlled oscillator (VCO)
make an accurate copy of the IF signal.
A phase detector then checks whether
the VCO is doing it properly and sends a
control voltage to the VCO as soon as
any change is detected in its frequency.
This is the demodulated FM signal.

The adapted 1C in figure 6's FM detector
also contains a PLL. By means of the
IC's pin 2, the phase detector is fed with
the 130 kHz signal. By way of the other
input (Pin 5) another signal reaches a
phase detector after originating from
the VCO. The phase detector makes
sure (by means of an amplifier) that
both its input signals have the same
phase and frequency, and its output is
at the same time, the low frequency
signal desired. C28, R17 and C30 form
the loop filter. Its dimensions are deter-
mined by the characteristics of the PLL.
Although the 9 MHz crystal filter has a
bandwidth of approximately 1 0 kHz, a
frequency deviation of 4.5 kHz can be
processed. The PLL operates well at a
deviation of up to 6 kHz, in other
words, for all the signals which pass
through the crystal filter.

The low frequency signal is amplified,
like the detected AM signal, in a single
transistor stage. With the aid of R18 and
C31 any spurious, left over 130 kHz

signal is suppressed. The output level
may be preset with P4.

The FM demodulator operates optimally
at an input voltage of approximately
200 mV. That is why potentiometer PI
has been introduced into the circuit. It
presets the optimal level of the signal
for the demodulator.

When building the FM demodulator, a
high quality capacitor must be chosen
for C26. This is one of the components
of the VCO which determines its fre-
quency.

The only trimming point of the FM
demodulator is the preset potentiometer
P3. Trimming is best accomplished by
setting PI (figure 4) at a maximum, so
that the input signal of the PLL is as
large as possible. Now it should be poss-
ible to adjust P3 so that an FM signal is
demodulated (which should, of course
be available!). Usually, demodulation
will be possible over a fairly large part
of the range of P3. Set P3 somewhere in
the middle of its range, and increase PI
slightly so that the signal level at the
detector's output is reduced. P3's range
(in which FM demodulation occurs) will
now be smaller; then set P3 again some-
where in the middle of that range. This
may be repeated until no noticeable
change occurs when P3 is varied. At this
point P3 is optimally tuned. M

6-26 - elektor june 1 980

First, the joysticks. Readers' reactions
and data were highly varied, to say the
least. Unfortunately, there was one
misprint in the 'test program' given
(Table 17 in the November issue): at
address 097A, the instruction should
read 0E427B instead of 0E4278. Until
this is corrected, the text on the screen
will be nonsense.

However, a sufficient number of readers
discovered this (or worked out their
own alternative program), so that we
received a large amount of 'joystick
data'. And widely varied it is, too . . .

on TV

games

In earlier articles ('I played TV games', Elektor October and November
1979) we asked for our readers experiences with the joysticks - and
promised to come back on the subject when we had enough data. We
received several interesting reactions to this request, often including
other suggestions and comments.

By now, we feel that it is high time to bring all other interested readers
up to date!

promises,
promises .

more

The minimum values found vary
between 05 and 28; the maxima were
anywhere between 25 and FA. The mid-
range could be anything between 15 and
7E. Help! What do you do when one
person's minimum is more than
someone elses maximum? The only
result that was consistent (not surpris-
ingly) was the value obtained without
any joystick connected: 0D in all cases.
Against all odds, we think we have a
solution that should satisfy everyone. It
is based on two conclusions from the
results given above:

• If joysticks are to be used, (auto-
matic) calibration is essential.

• Wherever possible, the joysticks are
best used as four-way switches

(signalling 'up', 'down', 'left' or 'right').
Trying to obtain data that corresponds
to all possible positions is virtually
doomed to failure, insofar as it is to be
compatible with other computers. For
strictly 'personal' programs it is no
problem, of course.

Before describing our solution, there is
one other point that must be made clear.
As several readers have pointed out, our
'definition' of the joystick connections
is not ideal. Furthermore, it does not
correspond to that used in the program
given as File 1 on the ESS003 record,
nor to that used in a commercial TV
games computer' based on the same
CPU and PVI. For these reasons, we
have decided to specify the following
'standards' (see figure 1 ) :

• Left joystick = address 1 FCC;
right = 1FCD.

more on TV games

• Horizontal movement = flag off;
vertical = flag on.

• Low data value = left or up; high
value = right or down.

Obviously, modifying an existing TV
games computer to conform with these
'standards' will require some re-
soldering. Not much, however, and
Table 17 in the November issue (with
the correction given above) provides an
adequate test procedure.

Now, we come to our 'solution'. An
automatic calibration routine and
'joystick scan' that can be incorporated
in any program that uses joysticks. The
complete routine is given in Table 1. As
given here, the actual initial calibration
routine starts at address 0F94. A
program can therefore be started in two
ways: 1F0F94 (BCTA.UN) or 3F0F94
(BSTA.UN). In the latter case, the
calibration routine is concluded at
address 0FAF with 16, C0, C0, as
shown; in the former, a branch to any
desired address can be inserted as
1 Exxxx, at the same address. In either
case, the calibration routine is run once,
at the start of the program. The joysticks
are assumed to be in their mid positions,
and switching points relative to these
positions are calculated and stored from
address 0FC0 on.

In passing, it may be noted that the
‘wait for VRLE' subroutine (starting at
address 0F80) may well prove useful at
various points in the main program.
Having calibrated the joysticks, control
returns to the main program. At any
point in this program, a joystick scan
can be requested by branching to the
subroutine that starts at address 0FC8.
For correct operation, this 'branch to
subroutine' must occur at frame end -
after a 'wait for VRLE' loop, for
instance. Depending on the program,
several variations on the routine given
may prove useful:

• at address 0FC8, the upper register
bank can be selected (to protect

existing data in R1 . . . R3) by modifying
the instruction to 7712.

• from address 0FF8 on, additions can
be incorporated: either resetting the

register bank (7510 = CPSL, RS) or
storing the data found: R2 contains the
left-hand joystick data, R3 that of the
right-hand joystick.

• the instruction at address 0FD5
depends on the point at which the

flag is set or reset. Obviously, to scan
both horizontal and vertical joystick
positions the flag must be set and reset
on alternate frames. The routine given
assumes that the flag is modified after
the joystick scan routine has been run.
In some cases, however, it may be
preferable to modify the flag first; the
instruction at address 0FD5 must then
become 1802.

• The complete routine can be situated
at any other point in memory, if

required. Since most of the instructions
use relative addressing, they can remain
unchanged. The only exceptions are the
absolute-indexed instructions at ad-

more on TV gar

elektc

1980 -6-27

0F80 0881 LODR.R0,lnd

0F82 r»0C1FCB LODA.R0

0F85 F440 TMI.R0

0F87 1—9379 BCFR

0F89 17 RETC.UN

SUBROUTINE:
WAIT FOR VRLE

0F8A Cl STRZ.R1

0F8B 51 RRR.R1

0F8C 51 RRR, R1

0F8D 453F ANDI.R1

0F8F A1 SUBZ.R1

0F90 C2 STRZ.R2

0F91 81 AODZ.R1

0F92 81 ADDZ.R1

0F93 1 7 RETC.UN

SUBROUTINE:
CALCULATE LIMITS

CALIBRATION ROUTINE

0F94

0F96

0F98

0F9A

0F9C

0F9E

0FA0

0FA2

0FA4

0FA6

0FA7

0FA9

0FAB

0FAO

0FAF

0FB0

0FB2

0FB4

0FB6

0FB9

0FBC

0FBE

7660

7518

♦3B66

3B66

08B0

0BB1

3B68

C81D

3B61
C818
CA1 5
B440
16

C0.C0

7440

0504

r*0D4FC0
CD6FC4
1—5978
— 1 B58

PPSU.II/Flag

CPSL.RS/WC

BSTR.UN

BSTR.UN

LODR.R0.lnd

L0DR.R3.lnd

BSTR.UN

STRR.R0

STRR.R2

LODZ.R3

BSTR.UN

STRR.R0

STRR.R2

TPSU.flag

RETC

2xN0P

CPSU.flag

L0DI.R1

L0DA.I-R1

STRA.I/R1

BRNR.R1

BCTR.UN

(flag on =

vertical)

clear VRLE.

1 FCC = left
1FCD = right
calculate and store
lower and upper
limits left

calculate and store

limits right

set. Note: absolute

(flag off ■ horizontal)

0FC0

0FC4

low high

00 00
Taft""

0FC8 7702

0FCA 20

0FCB Cl

0FCC C2

0FCD 0C1FCC

0FD0 0F1FCD

0FD3 B440

0FD5 p9802

0FD7 0504

0FD9 l*ED2FBF

0FDC p9A02

0FDE A601

0FE0 1*ED2FBF

0FE3 p 9902

0FE5 8601

0FE7 1*03

0FE8 0700

0FE A E02FBF

0FED r 9A02

0FEF A701

0FF1 l*ED2FBF

0FF4 p 9902

0FF6 8701

0FF8 1*17

0FF9 C0

SCAN

PPSL.COM

EORZ.R0

STRZ.R1

STRZ.R2

LODA.R0

L0DA.R3

TPSU.flag

BCFR

L0DI.R1

C0MA.I+R1

8CFR

SUBI.R2

C0MA.I+R1

BCFR

ADDI.R2

L0DZ.R3

L0DI.R3

COMA.I+R1

BCFR

SUBI.R3

C0MA.I+R1

BCFR

A0DI.R3

RETC.UN

0FFF C0

(or 7712= PPSL.RS/COM)

clear R1, R2 and
load joystick data

1 preset for R1 . if
> vertical. Note: sometimes
' 1802= BCTR is required

room for 7510 = CPSL.RS.
and/or CExxxx.

CFxxxx for data
transfer

Table 1. Jo

dresses 0FB6, 0FB9, 0FD9, 0FEO,
0FEA and 0FF1. These all depend on
the position of the data, stored here
from address 0FC0 on.

To illustrate how this routine works, a
very simple demonstration program is
given in Table 2. Once both programs
(Tables 1 and 2) have been loaded, the
main program is started at 0900. The
first two 'objects', corresponding to the
displays 'PC' and '=' respectively, will
jump to the centre of the screen ('0900'
remains at the top right of the screen).
The positions of these two objects can
now be modified by means of the left-
hand and right-hand joysticks, respect-
ively.

Hopefully, the suggestions given here
should be sufficient for those readers
who are developing their own programs.
As far as all others are concerned, the
only thing that is important is that all
programs that will be supplied as
'Elektor Software' in future should run
on their computer — provided the
joysticks are wired as outlined above.

Interrupt!

It is now time to acknowledge an
interrupt request ... A very welcome
one, at that. As mentioned earlier, we
received several reactions from readers.
One of the subjects that was often
mentioned was the 'interrupt' facility.
One reader in particular, Mr. Norman,
sent us a long letter in which he offers
the following tips:

'When using interrupts, you demonstrate
the method of looping the main
program and leaving everything to the
interrupt routine(s). This is a little
'wasteful' in processing time and I find
it preferable to 'share the workload' —
object movement and collision detec-
tion, say, to the interrupt routines and
score updates, off-screen travel, key
scan etc. to the main program.

To run both main and interrupts, it is
essential that registers and condition
codes do not ’clash’ and you do not
describe techniques in any great detail.

'If the interrupt routine uses, say, the
upper register bank whilst the main
program uses the lower register bank,
then a typical interrupt routine may
commence as follows:

7710 PPSL.RS

CC08FE STRA.R0

it cpci

CC08FF STRA.R0
and end :

0C08FF LODA.R0

0C08FE LODA.R0

7510 CPSL.RS

37 RETE.UN
'It is vital that PSL is preserved, other-
wise the main program may make
decisions on a condition code set by the
interrupt program!'

Very true. However, as another reader
has pointed out, the above routine is
not quite correct: after restoring the
PSL data, R0 is re-loaded - altering the
Condition Code!

6-28 - ale

0900

0903

0905

0908

090B

090D

090E

0910

0913

0915

0916
0918
091 B
091 D
0920
0923

0925

0926
0928
092B
0920
092E
0930
0933
0935

3F0F94
-* 7440
3F0F82
3F0FC8
0828
82

C825
CC1 F0A
0821
83

C81E

CC1F1A

7640

3F0F82

3F0FCA

0812

82

C80F
CC1F0C
080 B
83

C808

CC1F1C

1B4E

44 55 77

BSTA.UN

CPSU.flag

BSTA.UN

BSTA.UN

LODR.R0

ADDZ.R2

STRR.R0

STRA.R0

LODR.R0

ADDZ.R3

STRR.R0

STRA.R0

PPSU.flag

BSTA.UN

BSTA.UN

LODR.R0

AD0Z.R2

STRR.R0

STRA.R0

LODR.R0

ADDZ.R3

STRR.R0

STRA.R0

BCTR.UN

71

wait for VRLE

vert. pos. obj. 2

a the new 'joystick st

A routine that seems to meet all require-
ments is the following:

Start the interrupt routine with:

7710 PPSL, RS

CC09F1 STRA.R0

13 SPSL

CC09F3 STRA.R0

24 EF EORI.R0

CC09F5 STRA.R0

and end, at 09F0 say, with:

09F0 04XX LODI.R0
09F2 77XX PPSL
09F4 75XX CPSL (including RS!)
09F6 37 RETE.UN
Obviously, the three absolute addresses
in the 'save' routine will depend on the
position of the 'restore' scratch bytes.

More on the PVI.

Another topic that has elicited several
comments is the PVI. Two points, in

particular, have been raised quite
frequently.

• As can be derived from the documen-
tation supplied with the p.c. board,

several addresses in the PVI are available
as 'scratch'. So far, we have never
actually used them ourselves, but several
readers have pointed out that they can
indeed be used in the same way as the
'normal' RAM.

• Also shown in the documentation is
the fact that the 'I/O and control'

field is actually repeated four times:
1 FC0 ... 1 FCD, 1 FD0 ... 1 FDD,
1 FE0 ... 1 FED and 1FF0...1FFD.
This proves of particular interest for the
data stored at addresses 1FCA and
1FCB (collisions, VRLE, etc.). Both of
these bytes are cleared when read,
which can be a nuisance. However, one
reader has pointed out that reading

1FCA. say, only clears this one byte —
it does not clear 1FDA, 1FEA or 1FFA!
This means that a different address can
be used for retrieving data for each
object, as required, without affecting
the information required later on for
one of the other objects. Useful!

Questions and errors.

We are often asked why some programs
store 04 at address 1 E80. This was news
to us, but since then we have found the
reason. Apparently, a commercial
version of the TV games computer
exists, with the same CPU and PVI.
However, there is a difference: when
04 is stored in 1E80, the sound effects
are reproduced through the TV receiver!
We don't know how this works — it
certainly doesn't apply to our version -
but maybe someone can enlighten us?
Another regular query concerns R58. In
the circuit, this resistor is shown con-
nected to the video output - and
rightly so. However, observant readers
have found that it is connected to
positive supply on the p.c. board. When
this mistake was discovered, we im-
mediately tested several of our
prototypes to see what the effect was.
To our surprise and relief, it doesn't
make a scrap of difference! Which is
why we didn't mention it earlier.

A few readers have run into a minor
'problem' concerning the 'vertical offset'
for the duplicates. It was perhaps not
made sufficiently clear that a vertical
offset 'FF' is used as 'minus one': the
gap between duplicates becomes zero.
To delete the duplicates entirely, an
offset 'FE' must be stored.

Finally, it was stated in the original
article that only file numbers up to 9
are permissible. This is an unneccessary
limitation: any single-digit number
(1 ... F) can be used.

Interrupt! Mr. Norman again: 'In Elektor
48, page 28, figure 2a you show IC6
connections incorrectly. The A7 lead
should be connected to pin 13 and A6
should connect to pin 14. Furthermore,
the labels for pins 13 and 15 are trans-
posed: pin 15 is the '2c' input and vice

A useful tip

Have you ever tried to develop a
program? And discovered, after the first
trial run, that you had omitted a few
essential steps somewhere? Welcome to
the club!

Inserting the necessary steps can be
accomplished by replacing three of the
original instruction bytes by an uncon-
ditional branch to an empty memory
space, restoring the delected instruc-
tion(s) there and adding the missing
steps at that point before branching
back. This system works, but it is
anything but elegant. Those readers
who have 'decoded' the space shoot-out
program on the second ESS record for

TV gar

elektorjune 1980 -6-29

the TV games computer will know what
the end result looks like: a mess.

The only alternative is to move up the
remainder of the program to make room.
Unfortunately, this may well involve
moving up several handwritten pages of
perfectly good program, by laboriously
keying them in again at the new ad-
dresses. This is a nuisance, to say the
least. A so-called 'block transfer' routine
is much easier — in effect, you make the
computer do the bulk of the work.

The basic principle is quite simple. Let's
assume that one additional 'store
absolute' instruction (3 bytes) is to be
inserted at address 0A00. The remainder
of the program, say from 0A00 to
0AFE, will have to be moved up three
places in memory. This can be ac-
complished as follows:

08C0 05FF LODI.R1

08C2 | — *■ 0D4A00 LODA.I-R1

08C5 CD6A03 STRA, I/R1

08C8 1 5978 BRNR.R1

08CA 1F0000 BCTA, UN

When this program is started, at address
08C0, all instruction bytes are moved
up three address positions - one at a
time, starting at the 'top'. In practice
the computer does the job so fast that
the display on the screen hardly flickers.
The only modifications that must then
be entered by hand are all positions
containing absolute address instructions
that refer to the program section that
has been moved, and any relative
addresses that operate 'across the gap'.

What of the future?

More programs, in particular. We already
have an updated version of the 'space
shoot-out' that includes joystick cali-
bration and a few more 'gimmicks', a
'Mastermind' program and an 'Amazone'
game (man v, machine!). We're working
on a random number generator for
Bingo and a helicopter maze. In the near
future, we hope to introduce a new
ESS record — or maybe tape? — with
these programs. We've also got stacks of
other basic ideas, but developing
programs takes time. Come on, readers,
surely you've got some more ideas?!
We're quite prepared to help you out if
you've run into problems. The more
programs, the merrier!

On the hardware side, we have some
ideas. Memory extension? So far, we
haven't needed it — but we have an
extension circuit ready and tested. If
enough readers want it, we can design
a p.c. board. A random numbers
generator? No problem - a few ICswill
do the job. Basically, as far as we're
concerned, you name it and we can
provide it. However, we don't intend to
'waste' valuable magazine pages on
circuits that only appeal to one or two
readers. For this reason, we would very
much appreciate reactions from
interested readers: if several readers ask
for an extension, that gives us a good
reason to take it into consideration.
Over to you! M

A little program

Mr. M. Saliger sent us a small program
for automatically scanning existing
software. After some drastic shortening,
the result is as shown here. The first
address of the program (section) to be
displayed is stored at address 08C0,
after which this routine is started at
address 08C2. The addresses and in-
structions will now roll up the screen
automatically; the display can be
'frozen' and re-started by holding the
'start' button down for a moment. If
this key is held down continuously,

the 'scroll up' occurs at half speed. The
speed is further determined by the data
at address 1F9C.

When plotting long tables, it can be
useful to modify the instruction at
08 F0 to 0604, 0608 or 060C. Note
that the first few instructions after a
series of data values may be misinter-
preted.

Return to monitor via the 'reset' key;
don't operate the 'start' key in this
mode, or the program from 1F80
on will be erased!

08C0 0000

08C2 7620

08C4 0502

08C6 r*0D48C0

08C9 CD68A4

08CC 1-5978

08CE 3F02CF

08D1 3F042B

08D4 0706

08D6 r*CF488A

08D9 I-5B78

08DB 3BF2

08DO 0E88A4

08E0 F608

08E2 | 180C

08E4 F6A0

2 scratch bytes for address
PPSU.II

LODI.R1 ,

LODA.I-R1 (

STRA.I/R1 l ,rans

BRNR.R1 )

BSTA.UN scroll

BSTA.UN I

LODI.R3 I addre

STRA.I-R3 MLIf

BRNR.R3 lands

BSTR,UN,lnd.(02CF)l
LODA.Ind.l
TMI.R2
BCTR
TMI.R2
BCTR
EORI.R2
TMI.R2
BCTR
ADDI.R2
ANDI.R2
RRR.R2
RRR.R2
BSTA.UN

-, 2- or 3-byte
l instruction?
/Result in R2as
| 01 . 02 or 03

1 F80
1 F82
1F85
1F88
1 F8B
1 F8D
1 F90
1F93
1 F94
1F96
1F98
1F9B
1F9D
1 F9F
1FA0
1FA3
1FA5
1 FA6
1FA7
1FA9
1 FAB

[ 8704
0D88A4
3F0354
3F039F
FA73
3F020E

rrrr cB

9A7A
7702
3F0O55
8704
l— 5B71
12

0D1E8B

ADDI.R3

LODA.R1.lnd

BSTA.UN

BSTA.UN

BDRR.R2

BSTA.UN

LODA.R0

RRL.R0

BCFR

PPSL.COM

BSTA.UN

ADDI.R3

BRNR.R3

SPSU

LODA.R1

ANDI.R1

EORZ.R1

LPSU

TPSU.flag

BCTR

BSTA.UN

1 . 2 or 3 databytes
to display;

wait for VRLE;
display 6 lines

•SPEED

the monitor software should note that the

latter contains data at the following addresses:

0006 . . . 0009

00 AD . . . 00BC

0122 . . .013D

0177 ...0180

027B ...02CE

02F5 . . .031C

0537 . . . 053 F

The RAM scratch starts at 0800. The start

• Initiate : 0023

• ’Reg’ : 03B0

• 'Mem' : 040C

• 'BK' : 04A9 and 0594/0582

• -PC' : 050 E

• Wcas' : 05E8

• ’Reas' : 0758

6-30 - elektor ji

variable
logic gale

M. van Kerkwijk

The variable logic gate, which was
published inthejune 1979 issue,
generated more interest with
readers than expected.
Unfortunately, the 1C required
appeared to be difficult to obtain.
This gave many of our imaginative
readers food for thought. One of
the designs consists of two
inexpensive and readily available
IC's and is presented here.

For those readers who are not familiar
with digital technology and thinking,
a variable logic gate provides an excellent
opportunity to 'get into' it. To make
things easier, a truth table of all the logic
functions is given in table 1. All the sym-
bols representing the various logic func-
tions are also included.

The noughts and ones indicate the logic
level. A '0' means 0 volts and a '1' means
there is a voltage (for TTL it is +5 V).
The truth tables indicate how the output
(Q) behaves when various logic states are
fed to the input (A) or inputs (A and B).

The simplest logic gate is the buffer. The
logic state applied to the gate's input also
appears at its output. This can also be
shown in (Boolean) algebraic form (see
the second column in table 2); Q=A. The
buffer's purpose is, as its name suggests,
to increase the current driving ability of
a given logic line.

The inverter does more than the buffer.
As well as buffering the input to the out-
put, it also inverts the logic state of the
input. A 'V at the input will produce a
'0' at the output and vice-versa. The
algebraic formula for this function is:
Q = A, where the line over the 'A' indi-
cates its inversion.

The AND gate is a gate with at least two
inputs. Output (Q) is a logic 'V only if
both the inputs are also at a logic 'V.
The algebraic equation for this function
is: Q = A • B where the point must be
read as 'and'.

The NAND works almost exactly the
same except that there is an inverter on

the output. The truth tables show this
clearly . The equation for this function
is: Q = A • B.

The OR gate does something quite dif-
ferent. It produces a logic '1' on the out-
put if input A or input B has a '1' fed to
it. It also has a '1' on the output if both
the inputs are '1'. Expressed in algebraic
terms, this is: Q= A + B where '+' is to be
read as 'or'.

The OR gate also has an inverted type,
the NOR gate. The truth tables clearly
show the inverting o f the O R states. The
formula here is: Q = A + B. Two gates are
left: the EXOR and the EXNOR gates.
The EXOR (exclusive OR) behaves in the
same way as the OR gate with the exep-
tion that if both inputs are logic '1', the
output is not logic 'V but 'O'. To express
this difference in algebraic terms the sign
© is used. It then reads: Q = A ® B.

The EXNOR gate is, as you might ex-
pect, the inverse of the EXOR. This can
be seen from the tru th tab les. The for-
mula then reads: Q = A © B.

It all sounds very impressive, but what is
the point of these logic gates? you may
well ask. Let us take a practical example
in which a logic function is used. To cut
a piece of metal an automatic cutting
machine is used. It can of course be a
rather sticky business if, while opera-
ting the machine with one hand, you
happen to forget to remove the other
from under the blade. To prevent this
sort of thing from happening, opera-
tion may be made to be two-handed, in
other words, two pushbuttons need to
be depressed before cutting begins. Here
an AND circuit with two inputs may be
used. The inputs receive their informa-
tion from the pushbuttons and the out-
put operates the machine by means of
a relay. Only when both buttons are de-
pressed is there a 'V at each input, thus
a '1' at the output. A relay clicks and
the machine cuts the metal plate. A
safety measure may be included for the
sake of those dare devils who try to
push the metal plate forward with their
feet, thereby accidentally starting the
machine. Then an AND gate with four
inputs and four operation buttons needs
to be used, so that the person in ques-
tion has his hands and feet tied during

el elite

1980 -6-31

AND i[3 i

” 0 ^

operation. Admittedly, in practice no
use will be made of a real AND gate in
such a case, but the four pushbuttons
would be placed in series to the engine.
Even then it is still an AND function.
This is a simple yet useful way in which
digital technology may be applied. And
there are many other examples like it.

The variable logic gate

In figure 1 the layout of the gate is
given. This circuit is capable of fulfil-

that they form a single input. Input C
serves as a second input. If one wishes
to consult the truth table for the EXOR
gate (table 1) while experimenting, then
inputs A and B will need to be read as
A/B and C respectively.

Construction

The logic variable gate is best built with
TTL or with low power schottky TTL
IC's. The circuit needs to be fed with a
voltage of 5 V. A 4.5 V battery is there-
fore not suitable here. The power sup-

Figure 2. Power supply for the veriat

Figure 3. With the aid of this logic probe the
logic state at the output of the variable logic
gate may be read directly.

"O

EX NOR A\t

Input | output

A I B O

ling all the digital tasks hitherto men-
tioned. As illustrated, the variable gate
consists of three EXOR gates and a
NAND gate. How the circuit may be
programmed to carry out a particular
function, is shown in table 2. Supposing
we wish to turn the variable gate into
an OR function. Input C will then be
connected to the + of the supply (logic
'V) and input D to ground (logic '0').
This creates an OR gate with inputs
A and B and an output Q.

An EXOR function is obtained accor-
ding to table 2 by connecting input D
to the supply voltage (logic '1'). Inputs
A and B are connected to each other, so

ply drawn in figure 2 is however suita-
ble. The logic levels which appear at the
output Q can be 'seen' by means of a
voltmeter. A more elegant solution is
to read them with the use of an LED.
Figure 3 shows how this may be done.
Logic input probe Q needs to be con-
nected to the Q output of the variable
logic gate. If the LED lights, this means
there is a logic 'V level at the output.
Those of you who are interested in fin-
ding out more about digital technology
are advised to read digibook 1. This
home study course includes an experi-
mental circuit board, so that the theory
may be put into practice immediately. M

programmini

Inverter

AND

NAND

OR

NOR

EXOR

EXNOR C

1

smoke

detector

1 . When smoke is present, infrared light is

id detected by the photo-

The MEM 4963, recently introduced by
General Instrument Microelectronics, is
an 1C specifically designed for use in
smoke detector circuits. It is an
improved version of the older MEM
4962.

Many professional smoke detection
installations use an ionisation chamber
and, since this device is radio active, an
exclusive permit must first be obtained.
In this article we will discuss, what is for
us, a more familiar method - infrared
light.

The drawing in figure 1 illustrates how
the presence of smoke can be detected
by means of infrared light. An infrared
LED and photo diode have been placed
at an angle of 45 to each other. Smoke
has to be present before the photo
diode will detect infrared light. It is
really that simple.

However, an LED and a photo diode
alone do not make for a reliable fire
alarm. In figure 2 the complete circuit
for a suitable alarm using the MEM
4963 is shown. The circuit can be used
as it stands, or, together with any
number of identical circuits, can create
a larger fire alarm system. The circuits
are then connected by only two wires
(pin 8 of IC2 and supply common) as
indicated in the diagram and will detect
smoke both individually and collec-
tively. Furthermore, each smoke detec-
tor will test its own battery state
at regular intervals.

Four states

The circuit in figure 2 makes a distinc-

tion between four states and will react
differently in every case.

1. The circuit detects smoke itself
and sounds an alarm continuously. It

also sends a signal to the other circuits
connected to it.

2. If one of the other circuits detects
smoke, the alarm will not sound

continuously but in pulses during 30 ms
for every 100 ms.

3. The circuit 'notices' that its battery
is running out. The alarm then

sounds 3 ms for every 40 seconds.

4. Standby state.

Conditions 1 ... 4 have a decreasing
priority. That is to say that when, for
instance, the circuit detects smoke
around it and at the same time its
battery appears to be running out, it is
considered more important to report
the smoke than the low battery state.
There is something to be said for this, of
course.

The infrared LED is not continuously
supplied with current, but is pulsed. The
repetition time of the 150ps long
current pulses is about 10 seconds for
conditions 3 and 4, when there is no
sign of smoke. If smoke has been
detected, however, either by the cir-
cuit itself or by one of its kind, pulse
repetition rate goes up to 0.4 seconds. It
is better not to regulate the LED
continuously to save battery power. At
rest, the current consumed by the
circuit will be no more than approxi-
mately 10 pA.

The LED is not only controlled to test
for smoke, but also to check the
battery's state. After all, the voltage

during a loaded condition is important.
As soon as this drops below the value of
the zener diode plus approximately
0.2 V the battery alarm (state 3) is
sounded. The battery voltage is
measured by means of pin 13 of the 1C;
pin 14 is the connection for the supply
voltage.

The infrared LED D1 draws a lot of
current. This is necessary in order to
give the detector diode D2 a clear signal
which deviates from the infrared
environmental light which is almost
always present everywhere. The appli-
cations schematic uses fairly unknown
types for D1 as well as D2, but no
doubt the circuit works just as well with
the more common numbers given in
brackets. PI can be preset to reference
the input of the opamp within its
modulation range. Potentiometer P2
serves to regulate the circuit in such a
way that the alarm will only sound
when smoke is present. First place P2 in
the centre of its travel and then preset
PI.

The MEM 4963 has another CMOS
compatible output at pin 4. Normally
this will be a logic 0 but whenthealarm
goes off, it changes to a logic 1 . The
current capability of the pin 5 output is
approximately 240 mA. H

6-34 - 1

ligh fre

noise at high
fireqiioides

(an important factor)

Noise in UHF/VHF receivers can
be determined by using extensive
and expensive test equipment.
However, tests with a noise
generator can give usable results
at a much lower cost. Such a noise
generator can, of course, be
constructed by the amateur.

1

What is noise?

Noise is caused by highly complicated
physical and thermodynamic processes.
Briefly, it is the random movement of
electrical charge carriers. Noise increases
with rise in temperature: at absolute
zero (-273°C = 0 K) noise is zero, for
at this temperature all movement is
frozen. This is why during certain
critical processes, cryogenic techniques
are used to attenuate the noise factor

to obtain a certain signal-to-noise ratio
at the output.

How to determine the noise factor

The noise factor in receivers can be
calculated in two different ways, either
from a sensitivity or from a noise
measurement. In order to test sensitivity
a signal generator is required: however,
good quality HF signal generators tend
to be very expensive . . . Instead of
measuring the sensitivity with only
one frequency, we can apply many
frequencies at once: use a noise signal,
in other words.

This is how it works. First the basic
noise N of the receiver is measured
when the noise generator is switched
off. Then the noise generator is
switched on and the noise level is set
(by means of an attenuator) in such a
way that twice the input level can be
measured at the output. This corre-
sponds to a S/N ratio of 3 dB. The nice
thing about using noise methods is that
the S/N ratio is not dependent on
temperature or bandwidth.

Circuit

A small generator can be built with
inexpensive and readily available com-
ponents as shown in figure 1. A high
frequency transistor (T2) is connected

111 **/ xae..w imafik

Figure 2. Part of the frequency spectrum of

the circuit of figure 1. On the upper trace

the frequency is shown horizontally

(100 MHz per cm) and the amplitude
vertically (2 dB per cm). The lower trace
represents the noise produced by the
spectrum analyser (—97 dBm: 0 dBm is
1 mW with an impedance of 50 ft).

with very low temperatures. However, as a zener diode. It is fed by a DC
it is not always practical to go to these voltage source (T1). The noise voltage
extremes. and therefore output level is determined

The signal-to-noise ratio is the best by the setting of potentiometer PI
known method for determining to what which controls the amount of current
extent noise (N) affects the signal (S). that flows through the zener diode. The
This can be done by expressing the output impedance of the circuit is
signal-to-noise ratio in dB:

S/N = 10 log S/N dB

Taking a certain point in the receiver (temperature coefficient of the voltage
(after the demodulator for instance), source T1) achieved in the long run
it can be determined how many micro- is not ideal, but for comparative (short
volts are required at the input in order term) noise tests it is quite adequate. K

approximately 50 Si. The photograph in
figure 2 shows part of the generator's
noise spectrum.

Obviously, the circuit cannot be ex-
pected to perform miracles. The stability

For those who don't know, Morse
is a little like binary without the
logic. Understandably, learning
the Morse code is a long process.

In practical use one has to know
all the signals by heart, there is no
time to even think about it when
listening to an actual transmission.

Learning them is therefore very
much like reciting multiplication
tables in school. This is the idea
behind the morse trainer.

morse trainer

The morse trainer constantly repeats a
certain signal which has been chosen by
a few switches. A letter is represented in
morse code by a series of dots and
dashes, a dash lasting three times as long
as a dot. The interval between two dots
(and the dashes too) is determined by
the clock generator (N 1 ) in figure 1. The
clock frequency can be varied for differ-
ent difficulty factors (DF's?l.

When S5 is depressed, the outputs 'O',
T, '2', '3', etc. of IC1 (a decade coun-
ter) are high in series according to the
clock frequency. (The counter switches
on the positive slope of the clock
squarewave.)

By using the outputs at '1' (pin 2),
'3' (pin 7), '5' (pin 1 ) and '7' (pin 6)
only, an equally long logic 0 follows
every logic 1 of IC1.

If all the switches SI . . . S4 are on 'c',
four short signals are given which enable
the low frequency oscillator/amplifier
to produce four 'dots' through the
loudspeaker. This is the morse code for
the letter 'H\ As long as switch S5 is
depressed, the decade counter (through
the low frequency oscillator) will
repeat this signal over and over with
short pauses in between.

If a switch is in the 'a' position however,
the output of the corresponding pin of
IC1 will be connected to an extra
diode and an electrolytic capacitor C2.
This prevents the clock signal from
reaching the counter clock input
(pin 14). The capacitor is discharged
by R2 and Plb. The setting of Plb
determines the time it takes to dis-
charge C2. A dash is the result.

elaktorjune 1980 - 6-37

Cl - 1 (i/10V
C2 = 2 m2/10V

Semiconductors:
IC1 =4017
IC2 = 4093
T1 = BC 516
01 D13-DUS

SI...

. S4 = Switch SPDT with

:h N/0

sr off

55 = Pushbutton s»
single pole

56 = Switch SPST

57 = Morse key

LS = speaker, miniature 8 SI

There are four SPDT switches (S1...S4).
Table 1 shows the various positions of
the switches in order to generate morse
coded letters.

All the morse signals may be created
with various combinations of these
switches. This excludes figures and
other signs.

The printed circuit board is shown in
figure 2, along with the parts layout.
The four switches can be mounted on
the board or at a remote location. If
the audio output is too loud for your
liking, a 50 pot may be wired in series
with the speaker. The output may also
be wired to a set of headphones. This
will allow for increased concentration,
while reducing public irritation!

By connecting a sign key for S7, it is
possible to use the trainer for trans-
mission practice. If this feature is used,
S5 is not depressed and IC1 is not
used at all. By depressing the sign key,
the supply voltage will be fed to the
low frequency generator. This is ac-
complished via R8 and D13, causing it
to generate a tone through the loud-
speaker. M

die SC/MP
as a mini organ

H.W. Wyes

Elektor's SC/MP system may be used
to produce musical notes in the form
of a two octave software organ. The
hex keyboard functions as the manual.
As there are not more than sixteen keys
available, only whole tones may be
produced, usually enough to play simple
melodies. The software ensures that a
squarewave is produced at the flag 1
output as soon as a key is depressed.
This signal can then be amplified and
reproduced through a loudspeaker inter-
face (figure 1). This is the same system
used for the 'Kojak Siren' and the
'Singing SC/MP'. As opposed to the
'Singing SC/MP', which sounded mono-
tonous, the squarewave is modulated
here. This produces a much more
pleasant and interesting sound. The
duty cycle (the ratio between positive
and negative swings in a waveform) is
varied.

The table (from 0F53 on) determines
the pitch of the sound. At S3 the hex-
keyboard is continuously scanned to
ascertain whether any keys are
depressed. Using the extension register,
the hexadecimal number concerned is
added to the address indicated by
pointer 3. At SO, pointer 3 is loaded
with the address of the table (0F53).
Then at SI and $2 the program for the
tone generation actually begins. For the
section of the waveform marked DOWN
in figure 2, the program beginning at
SI ensures that the required phase-
shifting and frequency generation for
the desired tones, is created.

The program in section $1 runs until
the counter has reached zero. Then

section S2 is run. This controls the
phase as shown in figure 2 (UP section).
As you might have guessed, $2 will run
until the phaseshift returns to its
original value.

0F22

C40F IDI 0F
37 XPAH 3
C453 IDI 53
33 XPAL 3

C108 LD 08 (1)
943A JP S 3
C402 IDI 02

07 CAS

C100 ID 00 (1)

08 tCP

C201 ID 01 (2)
8F00 DLV 00
C400 IDI 00
07 CAS
C202 ID 02 (2)
03 SCL

FA01 CAD 01 (2)
8F00 DLV 00
C100 ID 00 (1)
BA01 DID 01 (2)
9CE2 JNZ S 1

0F24 C108 ID 08 (1)
0F26 941C JP $ 3
0F28 C402 IDI 02

0F2A 07 CAS
0F28 C100 ID 00 (1|
0F2D 08 N3P
0F2E C201 ID 01 (2)
0F30 8F00 DLY 00

0F32 C400 IDI 00

0F34 07 CAS
0F35 C202 ID 02 (2)
0F37 03 SCL
0F38 FA01 CAD 01 (2)

0F3A 8F00 DLY 00

0F3C AA01 I ID 01 (2)

0F3E E202 XOR 02 (2)

0F40 9CE2 JNZ S 2

0F42 90C2 JMP S 1

S 3

0F44 C108 ID 08 (1)
0F46 94FC JP S 3
0F48 D40F ANI 0F
0F4A 01 XAE
0F4B C380 ID 80 (3)
0F4D CA01 ST 01 (2)
0F4F CA02 ST 02 (2)
0F51 90B3 JMP $ 1

T^lJTJTJTJTJXnjTJTJ^^

se Width Modulated Squarewave

6-38 -elektor june 1980

le elek terminal width extender

die

< k k k kl<i*iniiial
wfcNi extender

It's getting wider all the time . . .

W. Menzel

The elekterminal published in December
1978 meets the criteria set at that
time. Users may, however, wish to
adapt the row length to the screen
width of the TV. This involves moving
the beginning of a row to the left and
adjusting the length of the row. With
the elekterminal the dot clock generator
is synchronized from the CRTC with
the aid of the (NT-signal. Between the
sync and the (NT signals there is a
fixed delay of 1 1 ns. By reducing this
period the picture may be moved
towards the left. In figure 1 a simple
circuit has been drawn in which the
picture can be moved in this way. The

circuit is no more than a delay line with
which the sync signal is delayed by 4 ps
on its way to the video combiner. As a
result, the (NT and the sync signals are
separated by about 7 /is.

The screen width is adjusted by
lowering the frequency of the dot clock
generator with the aid of capacitor C2
on the elekterminal. This modification
enables the width of the letters to be
slightly increased, so that they are
easier to read. Furthermore, the video
bandwidth is then also reduced, thereby
bringing the letters in better focus.

The values of capacitors C given in
figure 1 need to be empirically establ-
ished. The indicated value of 1n5
proved to work well in most cases.

74LS14

MPG meter

Due to unforseeable circumstances
we have not been able to publish the
fuel consumption meter as announced
on the contents page in the last issue.
The type of transducers required by
the circuit are not available at present —
contrary to assurances made to us —
and we are therefore endeavouring to
modify the circuit, so that a more
readily obtainable type may be used.
The finished result will be published
as soon as possible.

junior

computer

slicker

A special sticker is available for the
Junior Computer to increase display
readability. It is made from red, trans-
parent celluloid and can be affixed onto
a perspex plate. This may then be fitted
on top of the display printed circuit
board with the aid of four bolts and
spacers.

The words 'ADDRESSES' (left four
displays) and 'DATA' belong to the
normal use of the JC; 'OP-CODE' (left
two displays) and 'OPERAND' (right
four displays) are for editing and
assembling.

The sticker is delivered free with the
EPS printed circuit boards for the JC.
Those of you who have already got a JC
board may apply for the sticker in
writing (please enclose a stamped
addressed envelope).

We would also like to point out that a
complete kit of parts (including the
programmed 2708) is available from
Technomatic Ltd. Details are given in
their advertisement on page UK 27.

Modifications to
Additions to
Improvements on
Corrections in

Circuits published in Elektor

Elektor 56, December 1979 page 12-13. The
negative voltage regulator. IC5, is shown the
wrong way round on the component layout
and the printed circuit board. The flat side of
the regulator should face towards the centre

19

—

milliamps for 15 hours and at. for example,
1 milliamp discharge rate, voltage is main-

At 100 microamp discharge rate, voltage is
maintained for over 40 days (at 20°CI or
over 25 days (at 45° C». This means that the
memory mount battery unit is especially
suitable for memory unit transportation
on relatively long journeys.

22a Reading Road,

Henley on Thames,

Oxon RG9 I AG.

Telephone: 1049 12 1 2663.

1C pin headers

The full AP range of male and female headers
based on 0.10" centres and are available in

People vs Elek terminal (December 1978. page
12-161. Resistor R7 of the RS232 interface
input stands accused oi attempted arson, on
a number of separate occasions, due to
overheating. Under cross-examination, Elektor
design staff have admitted that it was their
responsibility and have satisfied the court that

promising to increase the rating of R7 to
1 Watt. This', they say 'will enable R7 to
conduct its current duties more easily without
getting hot under the collar'.

- £ -

extended to include a 34.5 MHz
Active car aerial applications as the first filter in H

Elektor 60. April 1980 page 4-24. In the " 8 - pole , "°" 0 ' i,h :

drawing (figured, one half of switch SI ^w.dth SSB f. ter. centred on 1
(thus. Sla or Sib) is shown in the wrong The la,,er 15 liable w,th the a F
direction. As shown in the schematic, the vpper and lower sideband crystals.

^

polyester and the dual row headers are
ultrasonically welded together. Built-in stand-
offs facilitate wave soldering and board
cleaning.

Female headers have 'tuning fork' contacts to
ensure excellent electrical connection and
mate with matrices of 0.025 square or round
posts on 0.10 centres. The solder tails are

right angle configurai
suitable for either
wire-wrapping.
Elektrokit LTD,

ions. G1 and G2 Telephone: 102771 23909.

Sealed nicad battery for
computer memory back up

The value of information stored ir
state memories often means that loss
power supply failure would be catast

nalysis. while the instruments's dy-
nge of 80 dB enables very small signal
s to be examined.

Modular input/output system

New from Hamlin Electronics is a range of
solid-state input/output modules and systems
designed to provide a simple interface be-
tween microprocessors and similar logic
devices and the circuits being controlled. The
total Hamlin system, consisting of four basic
optically coupled, colour-coded modules
and three plug-compatible printed-circuit
modules with barrier strips, replaceable fuses
and status indicators, is designed to interface
with any 5 V logic family, and provides the
user with maximum flexibility during instal-

Touch activated keyboard sealed
in plastic

A fully solidstate ASCII keyboard measuring
just 0.325 inches thick has been introduced
into the UK by Interface Components Ltd.
Known as the TASA Micro-Proximity Key-
board and designed by TASA Inc. of
California. USA, the touch activated keyboard

cTT ? 1

For those who do a lot of work in the home,

every conceivable purpose with which to save
time and avoid inconvenience in hunting high
and low for springs that cannot be found
when they are needed, or rushing off to the
local shops in search of springs which they

Ltd. costs £25,00.

Lee Spring Ltd..
Cornwallis Estate , At

Piezo/acoustic resonators

Toko has recently introduced a range of
miniature piezo-ceramic 'sound-transducers'
for all types of electronic equipment requiring
acoustic information — such as alarm buzzers,
keyboard entry verification etc.

These buzzers are available as mounted discs,
or encapsulated in a small plastic enclosure
las illustrated). The type PB-2720 produces
> 85 dB SPL with as little as 1 mA drive
current, at the resonant peak of 4.5 kHz.

on all the standard Hamlin ranges of liquid-
crystal display devices in both reflective and
transf lective configurations.

Hamlin Electronics Europe Ltd..

Self-adjusting wire stripper

Developed as a result of 12 years expertise
and three years of field tests, AB Engineering's
new MK 2 FC wire stripper marks a new step
in tool design and operation.

Ergonomically designed to multiply the force
exterted by the hand and light in weight -

structed in glass fibre reinforced polyamide
with high tensile steel moving components.
The MK 2 FC features a novel 'floating cam'
which automatically adjusts the jaws to the
correct stripping depth and simultaneously
adjusts the gripping pressure on the insulation.

However, these units are fully specified with
regard to broadband response characteristics
- e.g. the SPL at 1.5 kHz is typically 77 dB
with a 10 V p-p squarewave drive. A wide

New hobbyist catalogue

Designed to a new formant, a 52 paged
Hobbyist Catalogue has recently been released
by Vero Electronics.

The brochure contains a selection of products
that are particulary useful to the home con-
structor.

Several new products are illustrated including
Verobloc; a new prototyping method of
building and testing circuits; SI 00 bussing
system; a rack mountable development kit
for evaluation of micro-processor based
systems to the S100 format and Low Profile
DIP Sockets.

Send 40p to cover post and packing and this
new catalogue is yours.

Vero Electronics Limited.

Industrial Estate.

Chandler's Ford.

Eastleigh.

Hampshire. S05 3ZR
Telephone: 1042 151 69911.

You've heard about it
Read about it — HERE IT IS

AVAILABLE EX-STOCK %

COMPLETE KIT AS PER ^CK 0 *

INCLUDES FREE 16K EXPANSION

NASCOM-2

ON DEMONSTRATION NOW

\ NEW