|
for +WTFcircuit. this is on my breadboard at the moment. LM324 opamp is two flexible PWM controllers with adjustable frequency. I set the upper PWM to the optimal frequency and duty for this particular transformer, and the bottom PWM can be played with to make different sounds and effects; it sounds a lot better at very low frequencies and also dramatically reduces the power needed to maintain a big arc. you can still wind it right up to 100% duty if you want but it squeals.
PWM output is digitally mixed and inverted by bc945 transistors, the bottom transistor is an enable line which I use for safety - I'm working on a thermistor circuit to pull this low when the transformer or fet gets too hot.
output of this goes to a bc337 drive transistor. this drives a push pull pair of pn100/pn200 transistors which rams current hard onto the mosfet gate.
the mosfet I'm using here is an IRFP250. it has a lot of capacitance across the drain and gate junction. add the capacitance of the transistors and leads and it adds up to a substantial amount. this means that the gate signal won't be perfectly square, so the fet doesn't switch on and off sharply. this means the fet drain-source resistance increases, turning it into a effective heater instead of a switch. it has to be hard on or hard off to not waste power. and if you don't have a sharp switch off, you lose the high voltage back EMF effect on the transformer.
another problem here; when the fet switches off, the primary coil and this capacitance makes the gate signal oscillate wildly; the peaks can be high enough to turn the gate back on, or even blow it out. using this very low impedance push pull stage reduces both of these problems so the drive signal is powerful enough to charge and discharge the gate junction at the switch speed with an almost perfect square. I've tested this setup up to 100kHz with 40nF capacitors.
i found the main power rail needs at least 10mF worth of electrolytic capacitors on it or back emf spikes from the primary go back across the rest of the circuit and start destroying the opamps and control transistors. when the fet is on it will also drop the rail voltage a lot too, which drops the gate voltage and also screws up the duty cycle and frequency. caps hold it up here.
the transformer I'm playing with now is from a trinitron 21" monitor. it's not very big but it's putting out the biggest arcs I've seen yet. I'm using the primary windings in the transformer, no other windings added. The longest sparks happen with a switch frequency of about 7.5kHz at around 70% duty with 12V input and drawing about 5A. faster switching doesn't make the voltage any bigger; the arc gets hotter and won't form by itself, a struck arc can be drawn to about the same length, but the current draw on the supply increases exponentially. all sorts of bad stuff starts to happen as voltage starts to break over internally and across the pins, and even onto the core itself until the whole thing turns into a high voltage conductor. these transformers aren't intended to driven at these currents or up to these voltages. the ferrite core slowly heats up until it starts to melt the case, and once it gets hot, it stays hot for up to an hour. I've potted this transformer with pvc and sunk it in mineral oil but you can still hear it flashing over inside. nasty
anyway have fun and don't kill yourself or anyone else with it
|
