I was bored while waiting for some other things to finish and I remembered I had some high voltage transformers pulled from some scrapped laser printers.
They have two coils, one at around 100uH, the other at over 40H. I’ve always wanted to try the MOSFET based push-pull resonant converters and they where the ideal candidates

Everything is mounted dead-bug style, using the transformer itself as a base

I’ve kept the gate bias supply separated from the main power, to be able to experiment how I can modulate the output voltage

At first power up everything looked OK. I have 6V for gate bias and around 7V for the main supply.
I get a nice spark and the waveforms look nice (I forgot to take a snapshot)

The next thing to try is raising the main power supply and setting the discharge wires further apart. The transformer secondary breaks down and it has enough time to melt itself:

Lesson learned: check transformer spacing before pushing it to the limit. Luckily I have another transformer to play with …

I’ve managed to get the other transformer going with a CCFL bulb as a load. This way the CCFL will limit the voltage and I have a decent load to test some control ideas.
I’ve attached some wave forms:

I’ve left if running with ~ 6W of input power and only the transformer gets warm: 600mA trough the primary is too much
It looks like the best way to control it is to keep the gate bias constant, above the threshold voltage, and to change the main voltage – a variable buck converter.