Novel s.s.t.c.--utter bafflement (fwd)

["Tesla list" <tesla@xxxxxxxxxx>]

---------- Forwarded message ----------
Date: Sat, 06 Oct 2007 11:59:13 -0700
From: Ken or Doris Herrick <kchdlh@xxxxxxxxx>
To: Tesla List <tesla@xxxxxxxxxx>
Subject: Novel s.s.t.c.--utter bafflement

A few of you coilers may recall my sporadic efforts at reviving my prior 
(now defunct) s.s.t.c.  I've now reached the stage where I think it 
should again work...but it does not, exhibiting instead a truly baffling 
phenomenon.

The redesign incorporates somewhat of the notion outlined in my "vanity" 
U.S. patent 6,069,413--now inactive.  I have presently configured a 
novel voltage-doubling bridge-driver using 4, FGA50N100BNTD IGBTs and 
feeding an untuned primary, with secondary-current feedback to build & 
sustain oscillation.  The bridge's supply-capacitors are charged 
directly off-mains with half-wave rectifiers.  The circuit effectively 
doubles the +/- 150V or so obtainable from my 117VAC mains.

A 5V Schmitt oscillator running at ~100 KHz first provides a signal that 
is amplified and then drives a 5-winding toroidal transformer whose 
outputs charge up 4, floating +/- 30V power supplies.  Those supplies 
power 4, NPN/PNP totem-pole drivers for delivering +/- 25V pulse-bursts 
to the 4 IGBTs.  At the start of a given pulse-burst for generating a 
spark, the oscillator's amplified output is gated into other drivers ( a 
UCC37321/37322 pair) which feed 4, 1:1 toroidal transformers whose 
outputs drive the NPN/PNP pairs in the correct phase relation.  As soon 
as the secondary starts to respond resonantly, its return-current 
overcomes the Schmitt-oscillator and is amplified instead of the 
oscillator's signal, providing the Fr-feedback to sustain and build up 
oscillation.  An IC counter counts the cycles and shuts off the 
pulse-excitation after a chosen duration.

After much travail involving sundry design errors, redesignings, 
rebuildings, and rebuildings of the rebuildings, all now seems well & 
good except for this:  With no mains connection at all (the plug pulled 
out) but with everything else connected, I get nice pulse-bursts feeding 
the IGBTs for the first half-dozen or so cycles.  But then, the +25V 
portions feeding 2 of them decline rapidly to below 0V and stay there 
for the duration of the pulse-burst.  If I just keep the pulse-bursts 
coming, at 20/second or so, eventually the voltages rise up again toward 
+25.  But if I let up on the pushbutton for a few seconds and then try 
again, the phenomenon repeats.

This is, mind you, with a) floating drivers for the IGBTs, b) no supply 
voltages applied to the bridge, and c) all IGBTs seemingly OK.  And what 
really mystifies me is that when I disconnect the secondary-return wire 
from the circuit, the trouble seems to go away: all the IGBT gate 
signals then look fine.  But by all that's good & holy, the secondary 
shouldn't have any excitation at that time since the bridge has no power 
applied to it.  Furthermore, I isolate the secondary's return-end from 
the Schmitt circuit to the extent that it has to generate some 25 mA of 
current before it becomes connected to that circuit.

This is all very weird & is what drives me away to a good book or 
whatever for extended periods of recuperation.  Anyone local to Oakland, 
CA who's a) interested in the details of the design, b) is competent in 
s.s., and c) would like to spend an interesting & perhaps frustrating  
few hours in head-scratching would be welcome to visit!  In the 
meantime, I may just put my feet up & try to forget it, at least over 
the weekend.

Ken Herrick