Royer oscillator instability

[ezalys]

Hey all! I've tried building jim williams' high voltage CCFL supply... well, a very basic version of it where I just short the bases of the transistors to ground (no regulation), ground one terminal of the secondary and load the other with a 200kohm resistor. Just to see how it works. To be clear, the only parts involved are the switch transistors, one 1k base resistor, one 200k resistor as a load on the secondary, tank capacitor, and the transformer. When I include the the tank capacitor, everything works very smoothly and happily. When I don't, the system seems to just oscillate chaotically. Measuring a terminal of the primary, I observe a series of decreasing spikes. It seems to be switching back and forth but without completely switching all the way, but also with varying period. Listening to the transformer, I hear a squealing noise that doesn't have a definite frequency. I'm a big fan of the resonant royer design but I'm curious what causes this behavior and how you alleviate it in a Royer oscillator that doesn't involve a tank capacitor to obtain a relatively clean square wave.

[T3sl4co1l]

Sounds like squegging.  You need a low enough resonant tank impedance to dominate over other possible modes of operation, and enough base bias to power it.  You may or may not need/want base bypass capacitance (probably not; that's more relevant for single transistor (blocking oscillator) circuits), which causes squegging when the value is incorrect (again, a value of 0 might be most correct here).

Don't at all expect it to behave like a fixed mains supply.  The whole point of the oscillator is that you're sacrificing ease of use for sparsity of components.  Expect quirks!

Tim