I have some
old photographic flashes that I'm trying to control digitally. Flashes control their brightness by controlling the duration the flash fires for. For example, at minimum brightness, a flash might fire for 10 μs, while maximum brightness is achieved by letting the flash fire for 1000 μs. I've figured out how to start the flash, but I'm having trouble with quenching (stopping) it. I've also done my best to reverse-engineer the analog control circuit (attached).
The flash is split into a flash head unit and a control unit. The analog control unit has four wires: ground, trigger, ready/quench, and something I don't understand. To start the flash, you put a high voltage (> 50 V) pulse on the trigger line. The ready/quench line appears to directly reflect the charge on the capacitors (up to 320 V), but is current-limited to about 1.5 mA by the flash head. To quench (stop) the flash pulse, the ready/quench line is pulled to ground.
I can get the flash to trigger just fine. (The analog unit generates its high voltage pulse using a transformer, so I copied that idea.) The trouble is that when I pull the ready/quench signal low, it somehow feeds back into the microcontroller (a Pi Pico) a sizeable voltage pulse, which sometimes crashes the microcontroller. I've tried blocking that with a diode and I've tried absorbing the pulse with a
3.3 V Zener diode, but I'm still getting voltage pulses approaching 20 V depending on the circuitry I use. Fortunately, the pulse only lasts a couple microseconds, and doesn't appear to have damaged the microcontroller.
I've attached my reverse-engineered schematic of the original analog circuitry, along with the circuit I'm currently trying to use. Probing my circuit, I can see that voltage pulses are feeding back into the microcontroller. In the trace, 3 (cyan) shows the FET (Q4) gate voltage, while 2 (magenta) shows the microcontroller output pin ("Stop") voltage. (1 (yellow) shows the ready/quench line, while 4 (green) shows the start signal.) You may notice that some
slight ringing is apparent. I have found that using a physically much larger gate drive transformer tamed the back feeding enough to prevent crashes, but it's still getting to nearly 5 V and I can't shake the feeling that I fundamentally don't understand something very important.
How is it possible for these (relatively)
high voltage pulses to travel back through the FET, through my crudely-wound gate drive transformer, through the drive transistor, through a diode in series with the gate of aforesaid transistor (not pictured in schematic),
and despite a 3.3 V Zener diode (also not pictured)
? Is it even actually happening or am I just bad at probing?
I've blown a few FETs (extracted from CFL ballasts) while experimenting with different circuits to produce the quench signal. I also tried using a
TRIAC, but the back feeding is actually much worse. I tried using a high voltage BJT, but it didn't work at all (not enough current gain?). I originally tried using an
optoisolator, but its response time was far too poor.
What would be a reliable and fast (< 20 μs variation)
way for a microcontroller to pull a 320 V signal down to ground?One thing I'm noticing looking at the original circuit is that the unknown line labeled "C" appears to be getting a lot of the voltage from the ready/quench line. Perhaps I should try dumping the ready/quench voltage in there? But I'm not sure how to design a circuit to do that and I'm running low on high voltage FETs. I really don't understand how to use a TRIAC because either (or neither??) terminal can be the ground reference.