[TCML] SISG heating at high freqencies
FIFTYGUY at aol.com
FIFTYGUY at aol.com
Thu Apr 17 15:26:58 MDT 2008
In a message dated 4/17/08 9:56:26 A.M. Eastern Daylight Time,
pterren at iinet.net.au writes:
>I was quite surprised to find that the SIDAC's were scorching hot and the
>IGBT's were uncomfortably hot, particularly one of them after a very short
>run. Considering that these have been used by others on a pole pig with
>only mild heating, that surprised me.
Well, I haven't *thoroughly* abused the SISG yet. If the darn weather would
coincide with my schedule, we'll know pretty soon what it takes to vaporize
>It sounds like they had some frequency dependent problem because the SISG
>works OK (ie cold) with more primary inductance/capacitance.
Terry estimated things would be fine for <125kHz or so. Couple that with
his estimate of ~800A peak pulse rate. For a given number of SISG sections,
that kind of locks you into a somewhat narrow range of medium-sized
If you make the primary low voltage, you need a big cap to get energy
into the system. But the upper limit on frequency forces you to use a smaller
primary inductance, The smaller inductance causes the primary current to soar,
until you hit the 800A limit. Also invokes the annoying problems of tuning,
losses, and coupling that small-turn primaries cause.
If you make the primary high voltage, you can use a smaller cap to get
the same energy. Then you can use a bigger primary to keep the frequency down.
But once you get to ~30kV on the primary cap, corona problems are gonna
start to get troublesome. And then the only way to go is a bigger cap to get
longer sparks, but the bigger cap will make more peak current unless you make the
primary bigger - thus lowering the frequency. The lower the freq, the bigger
the secondary components.
Either run fewer sections with bricks to handle the current, or more
sections with cheaper IGBT's to handle the voltage. Do both for more "cubic
inches"! But it seems there is a *lower limit* on overall SISG coil size because
of the maximum frequency.
> Beats me why the SIDAC's were getting hot though. Perhaps the IGBT's
weren't turning on
>fast or fully so the SIDAC's spend more of the time conducting throughout
the cycle relatively.
That'd be my guess, as well. How much power did you get through your
purely SIDAC coil?
>Perhaps a lower gate resistor may have helped (I was using 100 ohms).
I think everybody's gone to 50 Ohm gate resistors. IIRC there were
indications that even less was better? One of the areas open for
>I was planning to use a smaller 4 stage SISG for a demo mini coil but would
>have to sort out the frequency issues first.
I've been working on the same thing. The way I'm doing it is to run a
twin - the double-inductance of the two primaries in series helps lower the
freq and keeps the peak current within limits. Still, it's tough doing it with
only 3600 Volts. That's why I'm buying another board from Mark to make it a
reasonable 7200 Volts. Allows much more design flexibility. Twins also solve
the problem of RF ground-return for the streamers, so long as you just let the
two terminals arc to each other. Any secondary that runs around 125 kHz isn't
gonna be "mini", but if you're gonna wind one, you might as wind two..
Center for the Advanced Study of Ballistic Improbabilities
**************Need a new ride? Check out the largest site for U.S. used car
listings at AOL Autos.
More information about the Tesla