El Supremo & OLTC "II"--not so good? (fwd)

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: Tesla List Moderator <mod1-at-poodle.pupman-dot-com>



---------- Forwarded message ----------
Date: Thu, 19 Sep 2002 11:40:49 -0700
From: K. C. Herrick <kchdlh-at-juno-dot-com>
To: tesla-at-pupman-dot-com
Subject: El Supremo & OLTC "II"--not so good?

Terry Fritz raised a question with me off-List regarding the notion of
using series-ed & paralleled transistors in making a Marx-like
primary-capacitor arrangement--in which capacitor groups are charged in
parallel from the mains and then discharged in series, through the
primary coil, via transistor switches.  I had pointed out that, if one
were to attempt that in a conventional L-C resonant circuit, then
problematic voltage-balancing circuits would have to be employed to
overcome the potential for avalanche breakdown due to differential
transistor leakages and gains.  Terry, I believe, agreed and also
indicated reservation as to the transistors' ESRs (equivalent series
resistances) adding up to unacceptable amounts.

Let me say this about all that, and about s.s. coils in general, it
seems--perhaps repeating some earlier observations by others (but
inviting comments, surely):

1.  In a conventional resonant-primary circuit e.g. the OLTC, potential
avalanche breakdown would be problematic.  But in my existing
Entertainment Machine and my prospective Supremo (see El Supremo's basic
schematic at  http://hot-streamer-dot-com/temp/tch2.gif), it's not a problem
because no transistor sees a voltage in excess of twice a capacitor
voltage.  The capacitor voltages are only uni-polar and they will never
exceed approximately the peak of the mains voltage.  

2.  Switching losses in the transistors differ, though, in an OLTC-type
design vs. mine.  In the former, the transistors are only switched on
once per spark-event whereas in my design they must be switched on and
off synchronously with the Fr of the secondary.  That's a marked
difference in total switching dissipation over each spark event--but
manageable, I would think, even using IGBTs (fast ones, that is), in
great part because the capacitor voltages decline during the spark event.
 Note that, in both Supremo and OLTC, switching off at the end of a spark
event is also much less of a problem since the capacitor voltages will
have markedly declined.  And another point:  In the E.M. and in El
Supremo, at each instance of all-transistors-off (i.e., between
half-cycles), the primary circuit is going to attempt to resonate at the
(series-C):L frequency.  But little resonant-rise will take place before
the transistors again conduct after the very brief interval.  

3.  As to the ESRs adding up, I looked at Fig. 2 of IR's IRG4PF50W data
sheet (which I assume is the same as for the "D" version with internal
diode).  From Fig. 2, I note a change in Vce of about 8 V over a change
in Ic of about 600 A.  This, I conclude, yields an ESR of approximately
8/600 or 0.013 ohms (and I may be repeating an observation of Terry's
here).  

3.1  Now, the inductance of my proposed 1-equivalent-turn, 7/8" x 16"
dia., copper-tubing primary is about 0.6 uH, yielding a reactance, at 140
KHz (my Fr), of about 0.53 ohms.  So already the ESR of just one
IGBT--never mind perhaps 5 in parallel--is negligible compared to the
reactance of just the tubing.  And although I have not calculated it, I
suspect that the tubing's equivalent resistance, at 140 KHz, would also
be negligible with respect to the reactance.

3.2  Maybe it's not the case as regards the tubing's Fr-resistance but
that's not the point:  Putting transistor-groups in series in order
thereby to multiply the applied voltage would seem to be advantageous
from the resistance standpoint.  Of course, they cost money...but who
counts?  (OK: a lot of us do)

3.3  As a matter of interest, compare a MOSFETs' ESR to that of the IGBT:
 I looked at Fig. 1 of IR's data sheet on the IRFPS43N50K.  It shows a
voltage drop of about 0.8 V at 10 A current and about 30 V at 300 A.  As
for the IGBT, I take this to mean (30 - 0.8)/(200 - 10) = 0.16 ohms
ESR--and only up to 300 A, not 600.  Over 10x the ESR of the IGBT! 
That's clearly why I couldn't use 1 (equivalent) primary turn in my
existing E. M.

4.  Lastly--and here's a kicker and perhaps Terry was making this point
as well--I make a rough comparison of primary current, during each first
cycle of excitation, in a conventional spark-gap coil vs. that in a
Supremo- or OLTC-type system:  If the spark-gap primary has 4 turns, say,
then its reactance at 140 KHz might be about 4^2 x 0.53 ohms = 8.5 ohms. 
If 10 KV gets applied initially through the spark gap, the current
(neglecting circuit resistance and considering the "C" to be a voltage
source during that first cycle) will be 1180 A.  In contrast, in El
Supremo or in OLTC, the "coil", of 1 turn, has a reactance of 0.53 ohms
(I posit similar 16" diameters).  Applying about 4 x 150 = 600 V as I
propose (and roughly that in OLTC), the current similarly will be about
1130 A.  So the first-cycle current is much the same.  >But< (the
kicker)...the flux available for exciting the secondary will be 4x
greater in the spark-gap system because there are 4x the quantity of
turns.  And as I have suggested before, in the first few cycles is where
the "punch" of a spark-producing Tesla coil should occur.  Tesla himself
would not have wanted that since he didn't want sparks; he would have
loved my invention: full power for as long as you want and can pay for. 
But we do want sparks.

4.1  So how to improve this?  It's not easy:  Doubling, say, the voltage
will double the current all right--but then each of, say, 5 IGBTs in
parallel will have to conduct ~400 A rather than 200--and they will have
to have >600 V voltage ratings, in El Supremo.  That would get one to
roughly half the performance of the good-old 19th century spark-gap coil.


So hmmm...any ideas, anyone?

At any rate, I'm going to try pressing on with at least a simulation of
El Supremo in SiMetrix--but pretty much as an exercise since I've already
"reduced to practice" my circuit-invention in the Entertainment Machine. 
After that--what?  Pushing 75, I'm inclined more & more to just loaf...

Comments are welcome.

Ken Herrick