This drawing is of three different oscillator ideas. They look pretty unbelievable but all work. They rely on little
known features of the active components. Do be careful if you experiment with 2 and 3. The voltages are high enough to
hurt ... a lot!


Fig 1

This looks as if the transistors are the wrong way round, and there are no bias resistors either!

Well with most ordinary transistors - that is ones without special internal diodes, resistors etc. - the collector and
emitter can be interchanged. It gives you a device with pretty horrible characteristics, but that is exactly what we
want here.

Ever wondered why VBE is always quoted as 6V for silicon transistors? The reason is that a zenner effect takes place at
around 8V. Now, if you actually use a significantly higher voltage, but current limited via a resistor, then the zenner
effect will allow enough base leakage current to pass to make the device conduct as if a bias had been applied. This
conduction holds the emitter (now collector) at this zenner voltage.

We now have a curious self biasing transistor, and two of these cross coupled with capacitors will oscillate happily as
a multivibrator. Needless to say performance is unpredictable, as is frequency!

With the components shown it will probably run at around 300Hz but this will vary a lot depending on the transistors.
Also, unlike an ordinary multivibrator the amplitude is virtually independent of supply voltage, but the frequency is
very voltage dependent, rising with increasing voltage.

The slope at the bottom of the downward stroke of the waveform doesn't always appear. I don't know why, but I suspect it
is a feature of different manufacturing processes. In any given batch of transistors they either all exhibit this
characteristic or they all don't.


Fig 2

Old timers will recognise this circuit immediately. It was very common as a basic timebase oscillator in valve (tube)
driven oscilloscopes. These neons (N1) are easily obtainable from hobby shops, both as wire ended and MES screw in
types.

Neons typically strike (start to conduct) at about 50V and extinguish again when the voltage drops by about 20V. When
used as a normal indicator, the series resistor is chosen so that the current is high enough to keep the neon 'on'.

In our circuit we do something quite different. We chose a very high value resistor, then have a capacitor shunting the
neon. What happens now is that the resistor charges the capacitor until the neon strikes. This then discharges the
capacitor until the extinguishing voltage has been reached (remember the resistor can't supply enough current to keep it
conducting). The neon goes out and the cycle repeats. 

Electrolytics can't be used for C1 as their leakage would be too high, but with a high enough value for R1 and a
polyester capacitor of, say, 1uF and rated at better than 100V quite a slow flash rate is possible.


Fig 3

This is the most obscure idea of the lot and although interesting I have yet to find a practical application for it!

Germanium point-contact diodes have an curious self healing feature, and this is what we make use of here.

If you reverse bias such a diode until it's breakdown voltage is reached the diode will go short-circuit. If you then
IMMEDIATELY remove the supply and also remove heat from it as fast as possible, the diode will recover and the process
can be repeated.

For this circuit to work it is very important that the diode is placed into a good copper heat clip otherwise you'll get
a click and a dead diode. Also, the two types I've listed are the only ones I've found that will work for any length of
time, and the OA81 although obsolete is the better of the two.

To run the circuit set the potentiometer for maximum resistance, and ideally put a milliammeter in series to monitor
current. As the pot is turned the current will slowly increase until it suddenly jumps at about 5mA and the circuit
oscillates. Back the pot off a little straight away or the diode will overheat. This circuit will oscillate at around
500Hz.

If you can get hold of an old valve audio output transformer and speaker, try connecting one end to 0V and the other via
a capacitor of about 22n to the 'hot' end of the diode. You should get a really loud squeal.

The capacitors need to be rated at 200V or better.