A different s.s. simulation

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: "K. C. Herrick" <kchdlh@xxxxxxx>

Biding time awaiting replacement IGBT bricks, I've developed an 
interesting secondary-feedback simulation.  I'll ask Chip to post my 
screen-captures, 
<http://www.pupman.com/current/kcherrick/2007-01-07_135435.jpg>http://www.pupman.com/current/kcherrick/2007-01-08_114712.jpg 
and 
<http://www.pupman.com/current/kcherrick/2007-01-07_153223.jpg>http://www.pupman.com/current/kcherrick/2007-01-08_115534.jpg, 
waveforms and schematic.

In the schematic, secondary-current feedback entrains the 
pilot-oscillator's frequency.  Its output is amplified and gated, at 
100 us-on, to drive IGBTs Q4 and Q6 phase-oppositely.  Overshoot from 
the tapped primary is snubbed by the D6/D10/C5 circuit, and C7 
resonates with TX3's primary.  In an identical circuit but without 
C7, D7 & D8, I find the expected big inconvenience--of each 
transistor turning off while still drawing a substantial amount of 
current.  But here it seems to be different...

Looking at the waveforms, notice Q6 turning on at ~85.5 us, with Q6's 
current following closely as would be expected.  But then, at 89.4 
us, Q6's current goes to zero before(!) Q6's collector voltage starts 
to rise.  The voltage rises slightly between 89.4 and 89.6, then 
rises to full value between 90 and 90.3 us.  The same effect obtains 
for Q4, as can be seen.

On the face of it, this would seem to be ideal: each transistor fully 
on before the current rises significantly and essentially off only 
after the current has diminished to zero.

Incorporating D7 & D8 greatly reduces the mains current and loading 
on the primary; I've since moved them over to the right of D6 & D10 
for even better effect.

The circuit tempts me since a) it requires only two power transistors 
(although, admittedly in addition, 4 high-current diodes); b) the (2 
instead of 4) drives to the transistors are ground-referred; c) 
there's no pesky double-resonance, with its "notches", to be 
concerned about; and d) it's always "in tune".

But perhaps, of course, an H-bridge would deliver more power to the 
secondary, more quickly.  This circuit produces, from the secondary, 
650 KV in 100 us.  A similar half-bridge 
resonant-primary/primary-feedback simulation--with +/-150V mains 
supplies--produces 400 KV in the same 100 us, implying 800 KV from a 
full bridge.  But the latter exhibits those dread notches--and, more 
complexity in power components and in the drive and the 
tuning.  Further, the push-pull drive scheme would reach the same 800 
KV only shortly after the 100 us interval.  And with any s.s. drive, 
that doesn't deliver the mighty initial wallop to the primary of a 
"disruptive" drive, who cares whether it takes a few more us to reach 
spark-breakout?

So who will comment on this?  I'll be happy to email the SiMetrix 
schematic-file to anyone who has that simulation program, so they can 
fiddle with it.  And I highly recommend that freebie, by the 
way.  It's such a deal!

KCH