Dual 2n3055 Flyback Driver
Posted by aonomus on July 20, 2008
So for my first real flyback driver, I built a dual 2n3055 transistor based flyback driver. The gist of it is that while one 2n3055 overheats very quickly, 2 paralleled 2n3055s will be able to split the current load and avoid overheating too rapidly.
I used a long terminal strip to arrange the resistor components, had a fan on the heatsink to cool it down, and used some alligator clip leads to make a chicken stick for the HV side.
Once I had the driver setup I was able to push up to 30V at 3-4A into the driver with only moderate heating of the transistors. Short hot arcs or long-duration arcs cause the transistors to heat up significantly, and a short cooldown time is required for the heat to transfer over to the heatsink where it is quickly dissipated. After I figured this out, I was able to push up to 5A (max capabilities of my PSU), where arc over of the flyback pins started to occur, melting charring one of the alligator clip insulating boots slightly.
After I solved the arcover problem with some extra plastic insulation, the fun began, I strung together 3, then 4 disposable camera flash tubes. When you connect a flash tube directly to the flyback, it has a dim arc inside the tube, but I suppose the slight capacitance of the wiring creates a much brighter flash whenever a momentary arc forms.
Another interesting discovery I made was that at these voltages, everything behaves differently, particularly electrostatic attraction/repulsion, and magnetism. At HV, tissue paper flies away from the negative lead, and camera flash tubes roll away, you’ll have noticed that I had to tape down the tubes earlier because they continued to roll away. Whenever an arc formed, the HV leads would jump slightly as the arc formed too, I can only imagine that this would be even more near a magnet.
The schematic (source: wiki.4hv.org) of the driver is fairly straightforward, the only issue I had was remembering that the feedback coil had to be reversed. The basis for this circuit is that there is always current flowing into the base of the 2n3055, and that when the primary coil energizes, the magnetic field of the transformer core cancels out the voltage in the feedback coil, turning off the transistor, which allows the feedback coil to come back up to voltage, and the cycle repeats. A similar driver exists where 2 2n3055’s are set in a push-pull configuration for full-wave AC input, which is my next step, followed by a Mazilli ZVS driver with IRFP250’s. Until then, this driver should be useful, right up until I burn out the secondary that is 😀
Procrastinatus 
MadsKaizer said
How big sparks could you get to jump from this thing? I only got to build a single 2n3055 driver for a flyback to supply my mini SGTC from 12v supply giving about 20Kv out.
Then I went on to build bigger TCs and recently I tried out the ZVS driver, it really works wonders compared to the 2n3055, but it will also blow some fets 🙂
aonomus said
I managed to get sparks just about 1.5″ long, mainly limited by my power supply. When I build a more robust power supply I’ll probably be able to draw longer arcs.
In the grand scheme of things, the 2n3055 is really inefficient, there are newer transistors which work better, and ZVS drivers are like night and day. I’m just waiting for a heatsinks and machining tools before I start on the next project.
federico said
Make a flyback driver is very easy, my flyback driver works at 24v DC with a single 2N3055 transistor and give 20kv out.
It give sparks of 15cm.
aonomus said
Personally I doubt the 15cm claim, as the 2n3055 system is a highly inefficient flyback driver. The inductance in the primary windings avalanches through collector-emitter as the circuit is turned off and the voltage rises, generating a ton of waste heat. Ontop of it the 2n3055 doesn’t have that much current capability at the frequencies discussed. From other experiments driving inductive loads at high frequencies, the 2n3055 easily overheats and destroys itself.
The die itself on the 2n3055 is tiny and simply isn’t designed to handle the level of power at the frequency (ie: dead short across an inductive load).
Image: (2n3055 insides)
For 15cm long sparks a ZVS or zero-volt-switching flyback driver would be needed, and those have been known to provide the power needed for 15cm sparks. As a spark is drawn longer the voltage required to sustain it doesn’t necessarily increase (as the plasma is already conductive), but the current required to sustain the arc increases. ZVS drivers can give enough power to destroy regular modern TV flybacks by overheating, burning out the secondaries, etc.