Board only built on mostly local parts

[Bayley]

Depends on your buscap size, the speed of your fuse, and bus voltage you were running at - the IGBT's are tough little guys (the datasheet rating is 120A for a thermally-limited pulse) and can handle some abuse before failing.

[Artlav]

Actually, the culprit turned out to be dead batteries on the logic part
I was still powering it from the batteries instead of bothering to adapt some sort of a PSU to it, and left them connected over a few hours - they lasted a few notes, then something bad apparently happened.

But there was no damage (IGBTs are indeed tough little guys), and the Bumblebee flies at last:
http://www.youtube.com/watch?v=ydonb1FKehw

Had to revert to the 480p camera - the HD one don't have manual focus and won't focus automatically.
The sphere is 12cm in diametre, and the sparks look to me to be about 30 cm long.
Is that a lot?

About 200W input power, same 30us on time per note, at full mains voltage.
Power consumption was about 0.5A on the low notes, 1A on the high ones, averaging about 0.7A.
Peaks were up to 11A!
Guess the capacitors need to be a little larger. Not long enough to trip any fuses, however.

I did wrap the interrupter in a metal cocoon, just in case, and cleaned up all the grounding.

I wonder what if i add more power?
Not tonight, most likely - the thing is too damn loud for this time of the evening.

[Artlav]

Meanwhile, wrapped up my build into a presentable shape with mostly 3D printed case.
No IGBTs blown so far, and i have 3 sets of spares (and two more sets of irgp50b60pd1), so higher power tests might happen soon.


Also, my two kopeck to the tuning and secondary coil parameters problems raging now:
My secondary inductance is 28.6mH, 424 Ohm resistance.
It's 50mm in diametre and 280 mm long, with 1900+-10 turns of 0.125 mm diametre wire (AWG36).
Top load is a 12 cm diametre sphere.
It resonates at about 350kHz, with the primary matched up at a little lower than that.
Performance appears normal, but no idea how good it is in terms of % of Onetesla power.

[jeffcrilly]

That's very nice.

Makes me want to build another one...

[stegu]

It resonates at about 350kHz, with the primary matched up at a little lower than that.

More exactly how much is "a little lower"? 10% or 5%, or even less? In any case, your results make me confident that I won't blow anything up by raising my primary resonance to match the 330 kHz I measure from the probably strongly out of spec secondary I received from oneTesla. Thanks for the input.

[DJKOR]

I must say, I'm quite impressed with a custom build like that. There is something about the homebrew look it has that makes it look even cooler. Good work.

On the note of running the primary at a higher frequency to match a higher resonant frequency of a secondary, is there any disadvantage to this? Like extra load on the components or more stress on the primary? Also, after a good 2.5 minute song that contains a fair few notes, should the secondary be completely cool indicating things are spot on, or are there reasonable losses in the primary that will result in it being warm to the touch? I need to get an IR thermometer, but after a good song, my secondary is a little warm, but not crazy. Like my old Galaxy SIII has gotten hotter under heavy load.

[Artlav]

More exactly how much is "a little lower"? 10% or 5%, or even less?

Primary is about 310-320kHz. The secondary is 340-350kHz, with counterpoise and breakout point.
So, it's 5 to 10%.

I guess the difference is consumed by the streamers - i tuned it based on the performance, varying the shape and size of the primary, not the precise calculations and measurements.

If anything, the secondary might need to go higher - i tried adding a small cylinder as a breakout point to get crown-shaped streamers, but the performance got noticeably worse.
Even though it lowers the frequency by only 10kHz.

On the note of running the primary at a higher frequency to match a higher resonant frequency of a secondary, is there any disadvantage to this? Like extra load on the components or more stress on the primary

I expect more load - more switching losses, less accurate ZCS, and so on. IGBTs are slow beasts, they can't switch on a whim.
So, the on times and off times are getting closer together at higher frequency, increasing the ratio between the time IGBTs spend in the lossy region and the near-lossless region.

Also, after a good 2.5 minute song that contains a fair few notes, should the secondary be completely cool indicating things are spot on, or are there reasonable losses in the primary that will result in it being warm to the touch?

After 1 minute of operation at full power (260VAC input, 30us notes) the bridge's radiator gets to 60*C (no fan), the primary is about 45*C, the secondary stays at ambient (28*C), give or take IR thermometre's allergy to metals.