Thyratron Pulse Generator
9th November 2004.
Another project that started with an idle conversation!
The project is to build a simple pulse generator which will provide a 1Hz pulse to drive a Nixe tube clock. The clock is intended to use Thyratrons (gas triodes) in its divider circuits, so it would be nice if the pulse generator could do the same!
The designer of the clock is Sam Peck, and he has a website for the project here:
Sam Peck's Thyratron Clock Page.
The initial thought is to use a thyratron in a relaxation oscillator, running at 1Hz. This will provide a ramp waveform, this can then be differentiated to provide a pulse ouput. The basic circuit can be seen here.
The timing of the circuit is governed by the charging of capacitor C1 via resistor R1, until the breakdown voltage of the thyratron (V1) is reached, at which point the capacitor discharges via V1, and the cycle starts again. The period of the oscillation can be determined from:
vc is the voltage across the capacitor
V is the applied voltage across the resistor and capacitor combination
C is the value of capacitance in Farads
R is the value of resistance in Ohms
t is time in seconds
e is the base of natural logarythms.
To suit us, this rearranges as:
In our case, vc is the breakdown voltage of the thyratron (measured as 16V), C is 0.5uF, and R is 1M, this gives a calculated period of 33mS, as opposed to a measured period of 40mS. Grid bias of the thyratron was set at -1.4V.
These values of R and C should easily give a period of 0.5S, but this will require an increase of thyratron bias, which in turn increases the breakdown voltage.
The main reason for running the experimental circuit at 25Hz is that this is the lowest frequency at which my Philips 'scope will reliably trigger!
The ouput of the circuit is a negative going pulse with an amplitude of 10V.
10 th November 2004.
After a little more playing, I have settled on the following circuit (component values may change a little in the future):
Thyratron Pulse generator
I tried different bias arrangements, in an effort to make the setting of the bias voltage less critical (the basic circuit period can change by 100mS with a bias change of about 0.2V). In the end, it seems best to go with a simple bias arrangement, but vary the timing resistor to give accurate setting. The 100K variable will change the period by about 90mS from end to end, and the 1K variable gives about a 2mS fine adjustment.
Lowering the supply voltage to 150V allows the circuit to run at 1Hz, and improves the setting of the bias voltage.
The value of the bias pot has been reduced to 5K, as the thyratron can draw a couple of mA through the control grid.
I have been observing the waveforms on my Tektronix 535A oscilloscope, as this will trigger reliably at very low frequencies.
12th November 2004.
The following circuit provides a 1Hz pulse, and is probably about as far as this idea can go:
Thyratron Pulse Generator with Ouput Stage.
The ouput stage, V2(half of a 6SN7 double triode) is run nominally with zero bias, driving the valve hard on - the anode load (R8) is made large to limit the current to acceptable limits, and so that the load is very large with respect to the anode resistance (about 6k Ohms at this operating point), consequently, almost all of the voltage is dropped across the load resistor, making the ouput voltage close to zero (actually about 1V). When the Thyratron triggers, the anode voltage drops almost to zero. This sharp pulse edge is differentiated by C2 and R5, providing a very sharp negative going pulse of about 100V amplitude. This is applied to the grid of V2, which cuts off the valve. The anode voltage then rises to almost the full HT voltage, providing a sharp positive pulse at the ouput pin, the pulse has an amplitude of approx 220V, and a pulse width of around 15mS, with a sharp leading edge, and slight slope to the trailing edge (driving a 'scope with a 10M ohm input impedance, roughly equivalent to the grid circuit of another valve).
Other suitable valves for V2 are the 6J5, which has the same characteristics as one of the triodes in a 6SN7, or you could use a 6SL7, which is a double triode, but with different characteristics. As the 6SL7 has a higher anode resistance, using a single section will increase the voltage of the lower part of the pulse to around 9V, and using both sections in paralell will reduce this to about 4.5V. An ECC81 should give the same results as a 6SL7 as well, with the advantage of a smaller base and envelope.
V3 is a voltage stabiliser (VR150/30 or 0D3), with a running voltage of 150V, in conjunction with R7, it stabilises the oscillator voltage to approx. 150V with a HT input of between 220V and 300V. An 0A2 is also a possibilty in this position. R7 really needs to be a 5W device.
The oscillator will free run at 1Hz, with coarse frequency control via the bias pot, and fine frequency control by altering the total charging resistance (R1 + R4). To set up the oscillator, set R1 to its midpoint, then adjust R6 for as close to 1Hz as possible. R1 can then be used as a fine frequency control. The timing capacitor is C3.
Even with the voltage stabilised by V3, the oscillator suffers from some drift (about 10 mS, but has jitter of around 20mS), which means if it was used as the bassis of a clock, it could drift by as much as 20 minutes a day, which is not acceptable. The circuit can be easily synchronised to an external 1Hz pulse by applying the pulse to the Thyratron grid via C1, an amplitude of around 5V is required for accurate synchronisation. In this case, the oscillator frequency should be adjusted to a little below 1Hz.
The bias voltage also need to be stabillised, but due to the very low current consumption, a dry cell could be used to supply the bias, and the value of R6 adjusted to suite. As long as G1 stays negative with respect to the cathode, the EN91 does not draw grid current.
This circuit makes a convenient "quick and dirty" pulse generator for testing valve counter circuits, but is not really accurate enough for its intended purpose, so back to the drawing board!
Copyright J.Beacon & M.Wroe-Parker 2004