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

Re: Solid state spark length record?

 Well, from my experiments with SCR based coils IGBTs look like the hot
ticket. IGBTs like SCRs are four layer devices and have "fixed" voltage
drops. This means the power lost in the switching device goes up linearly
with increasing current rather than I^2R as does a mosfet. The IGBTs have
some nice advantages over SCRs namely they turn on fast and they can be
switched off.
   I tried using saturable reactors to effectively speed up the turn on time
of the SCRs which works but causes other problems ( like holding the SCR on
for an excessive period of time.) So I ditched the saturable reactors and
went to a 365A 1800V 60uS Tq SCR. This sucker is rated for 8000A peak. This
works OK except that the SCR doesn't turn off until the primary cap is
almost empty. No first notch quenching:-(. In order to get first notch with
a 100KHz coil a Tq of 5uS would be necessary. Unfortunately 5uS SCR are kind
of rare and don't usually exceed 600v.  As you head towards lower voltages
and higher currents and more capacitance keeping the inductance low enough
to hit the target resonance frequency becomes difficult. Not to mention how
absurdly careful you need to be in order to minimize parasitic inductance.
If your cap has as much inductance as your primary then your primary will
only see half the voltage. This sort of thing can happen when your primary
has only one turn.  So higher voltage is better! 
  A 1200V IGBT is sufficiently fast and can be turned off, but there remains
one question. How far can they be pushed past their current rating for a
pulse? A 100A SCR can do 1000A for a brief pulse. Could a 45A IGBT make it
to 450A? The ratings on IRG4PH50U say 45A cont. 180A peak but there is no
indication in the data sheet as to how the max peak current relates to the
with of a pulse (for infrequent pulses). If They can hit 450A look out!
  With my last experiments in solid state coiling I could pull about 18"
with a 1 joule cap fired 400 times per second. The SCR was barely warm.
quality caps are an absolute necessity! I started with some CDE SCRN222s.
Their internal connections were so poor the terminals got hot although the
rest of the cap was cool. With 10 joules per bang an IGBT coil should start
to really perform!

To see my outdated web page with some of my early work look at:

http://home.hiwaay-dot-net/~eburwell


Oh yeah.. IGBTs typically have internal anti-parallel diodes rated for the
same current as the IGBT. Convenient..

Eddie Burwell