In around 2000 I built a Tesla coil for producing high voltage electric discharges. Rather than use the conventional spark-gap circuit with all its noise and low-tech behavior, I decided to build one using vacuum tubes. I have been obsessed about high voltage phenomena and Tesla coils since my teens when I was experimenting with taking TVs apart to see how they generated their high voltages.
I went through around three iterations of this circuit to improve the performance. It is still not as good as it could be for a variety of reasons. Maybe one day I will re-visit the design with the benefit of some computer simulations to really nail the working parameters.
I’ll give you a description of the function of this circuit now. The power supply consists of a microwave oven transformer that gives 2kV at around 1kW power and this is doubled to 4kV by a voltage doubler circuit. Actually this is too much voltage. I did not run it typically at 100% on the variac. The 572B tubes are not specified up to 4kV, but they will just about take it. They did arc over once or twice, but it was a rare occurrence and they were not damaged. The arcing was probably because of too much negative grid voltage in actual fact (the grid swings extremely negative during cutoff). If I rebuild the circuit I will use a lower high voltage supply and try to get the efficiency higher.
The circuit consists of two Svetlana 572B tubes in parallel. These drive an air core transformer and there is a feedback wiring that goes to the grids to create oscillation. The tank capacitor is a huge 15kV 1nF doorknob capacitor, which seems indestructible. The grids get feedback using a self biasing scheme where the positive feedback voltage causes grid current to flow which via the 4nF capacitor biases the working point below ground. The RF circuit consists of the tank circuit connected to the anodes, then the 20nF 10kV HV bypass capacitor to ground, then from ground through the heater bypass capacitors at the cathode and then via the tubes back to the anode. It is important that good quality RF capacitors are used which can take a high voltage and RF current. The parallel arrangement of HV bypass capacitors that I used was not very good, and some of them have blown from too much RF. I would replace them with a single high value high current high voltage low loss capacitor.
The main difficulty with this circuit is getting the right turns ration, oscillation frequency, coupling and number of secondary turns, as well as the size of the top capacitance. This all has to match at the operational frequency and also it has to be optimal for the tube output impedance so that it maximizes the power that is delivered to the spark load. I would not say that this circuit was well optimized, even if it did produce quite nice 10 inch sparks as a result.