The following picture from Justin McMahan, who is using this circuit to power a Radio Shack plasma globe. He says "the plasma is supposed to stay inside the globe!"

**** Note added June 6, 2002: There is a maximum duty cycle that the circuit will work properly, above which really horrible MOSFET-eating things will happen. The maximum duty cycle for my circuit is approximately 70% (70% MOSFET on time, 30% off). Maximum duty cycles will change based on your flyback transformer, your primary coil, and input voltage. You will KNOW when you've hit the maximum. Most likely there will be oscillations, screeching noises, high input current and not much output, etc. So start at a low duty cycle and slowly increase it WHILE you pull arcs. At some point, the circuit will get highly unstable....so back off the duty cycle a little bit, and turn everything off. Take the duty cycle pot out of the circuit and measure its value with an ohmmeter. Now you have your minimum/maximum resistances for your circuit, and you can fine-tune them with series resistors and/or a bigger or smaller pot value. Ultimately, you want to go from minimum duty cycle to maximum with one turn of the pot....and you want it to go just far enough to give good output, but not so far that it get's unstable.

So far, there are two schematics on this page. One is a signal generator, the other is the power electronics stage of a flyback driver. Both work extremely well AS SEPARATE UNITS. As of today I have not tested the two circuits together, but I see absolutely no reason why they wouldn't be compatible.

There are two things you can do:

1). You can use your own signal generator or a commercial unit. In either case, I suggest keeping transistors Q2, Q3, Q4, and Q5 in the design. They make up buffer amplifier stages --- they clean up the input square wave, and provide a high impedance input with a nice solid low impedance output. As an option to the four (expensive) transistors, Jan Wagner suggests using a CD4049 hex inverter in their place. If you do this, tie the inputs and outputs of the chip together so you use all 6 inverters in the chip. Also another way to do this is use a gate drive IC. Maxim makes some, International Rectifier does, as well as TI and others. The LM7812 power supply circuitry consisting of T1, BR1, C1, C2, U1, and C3 is only used to power the transistors. It is a regulated 12V supply so it can also power a gate drive IC, a CD4049, or other method.

2). The other option is to use our signal generator. This guy will tune any flyback transformer, while allowing you to adjust the duty cycle from 0 to 100%. And what's really beautiful, is that if you change the duty cycle, the frequency will remain the same. And vice versa. I don't know of many circuits that will do this at 200kHz and still give a clean output. YOU DO NOT NEED THE FOUR FRONT-END TRANSISTORS (referring to the power electronics schematic) if you use our signal generator. The MAX4420 is the reason for that.

In all cases, make sure there is a 33 or 47 ohm resistor between the output of the signal stage (whether it be your own or our design) and the gate of the power MOSFET, an IRFP460 in this case. That resistor forms an RC time constant with the gate capacitance, and slows switching time. There is a balance between MOSFET heating/losses and parasitic oscillation. Somewhere in the 50 ohm range will do the trick.

I designed and tested this circuit late last night, really late last night so there may be some errors. I checked everything over again today and found none --- but two heads are better than one, so if someone spots a screwup please let me know!

Diode D2 is an IXYS 30-06A. Looking at the front, the cathode is the left-hand pin, the anode is on the right. The package is TO-247. The writing on the diode says "DSEI 30-06A" on the first line. The second line says "022G".

Just about all the enhancement-mode N-channel MOSFET's will have a pinout like this: G, D, S. Gate on the left, drain in middle, source on right....looking from the front of course. So the IRFP460 is GDS.

• T1 can be any transformer with 120V primary and 12V secondary @ 100mA or more.
• BR1 can be any small bridge rectifier, 30V or higher at a few hundred milliamps.
• Q1 and Q2 can be any generic small transistor, ex. 2N2222 (NPN) and 2N3906 (PNP)
• Q5 is a single IRFP460. Heatsink! Two paralleled IRF740's or IRF840's can be used also but this is untested.
• D1 is three IN5822 Shottky diodes in parallel. Could possibly be unecessary.
• I'm using IXYS 30-06A for rectifier D2. A 1kV, 70nS, 20A rectifier would work too. Heatsink!
• T1 is a TV flyback transformer, remove old windings from core and wind on 15 new turns for the primary.
• The external supply should be rated 0 to 30VDC at up to 5A. Current draw will vary with type of flyback transformer.
• The LM7812 circuitry can be replaced with a 12V battery or benchtop power supply. Put a 10uF capacitor directly across its terminals, watch polarity. Current draw on this supply is very little.
• The RC snubber network (10R and .01C across the flyback's primary) is probably unecessary.
• Voltage spikes on the MOSFET drain should not exceed 300V.
• A 250 or 300V MOV directly across the flyback's primary will help insure MOSFET protection.

Email me at [[ justin@hvguy.com ]] if you have questions or comments, this schematic was draw directly off of my circuit but there still could be errors!