[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: A s.s. head-scratcher



Original poster: Ken or Doris Herrick <kchdlh@xxxxxxxxx>

Awaiting replacement IGBT bricks (used, of course, and perhaps already no-good as were some of the others, I'm sure), I've simulated something interesting. See Single res..._p-p4c.pdf in http://www.pupman.com/current/kcherrick --giving Chip a bit of time to post it.

Q1 & Q4 are driven out of phase by +/- 25 V square waves at the secondary's Fr. They're in push-pull with the two main, untuned, primaries of TX1. D6/C3 and D9/C5 conventionally snub the collectors' overshoots--and at the same time slow the collectors' voltage rises during turn-off. The "snub" primaries act to discharge the snubbing capacitors on alternate half-cycles. Notice that each of them is returned to the opposite transistor's collector.

The V4 & V5 circuits simulate MOVs, added as insurance to clip any overshoots exceeding ~900 V.

I haven't fully doped it out, but I found it a little startling that in this simulation I realize half-again the voltage rise at the output--and that, occurring in half the time--as I had obtained with my prior simulation of the push-pull "snub-primary" idea. Of course, the mains current is too much--for my mains, at the 1% duty-cycle--but the scheme does seem worth pursuing. I'll try another turn on each of the primaries, to cut down that current.

The waveforms are of the 4 primary currents plus the C4 mains-current.
I manually added up the amplitudes of the primary currents and found that the sum did, indeed, match the shape and amplitude of the C4 current, as I think one should expect. So all that wiggle-waggle produces some 800 A p-p of net primary current.

Any comments on this will be most welcome; and I repeat: if anyone has SiMetrix and would like to play with this, I'll send along the file. I intend to re-connect my hardware to try this out within the next few days--assuming the "new" bricks work.

Ken Herrick