Re: FET vs IGBT thoughts

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

Original poster: "Justin Hays by way of Terry Fritz <teslalist-at-qwest-dot-net>" <pyrotrons2000-at-yahoo-dot-com>

Hi All,

Yes, Thanks Terry. I didn't realize that IGBT's were that nonlinear!
I knew they were, but wow. That's a huge advantage (1/20th of the
dissipation I think you mentioned) over a MOSFET on the high end of
the current range.

 > IGBTs tend to turn on and off "relatively" slower than FETs.

Yah, I noticed. I made a half bridge induction heater with them once,
using a single primary on a gate transformer, and two secondaries. I
got a tremendous amount of shoot-through current spikes due to the
overlapping conduction times ; ). If an SSTC is built with IGBT's,
you would either have to run a very low frequency (50kHz) or add a
couple hundred nS of dead time (which is where the freewheeling
diodes would be *very* helpful).

 > Of course, if one really wants to speed the gates up, you resonant
 > the gate capacitance with an inductor for operation in the 10s of
 > MHz if you want.

Doing this restricts the circuit to "single band" operation due to
resonance at one frequency (ignoring harmonics). If you want a nice
range of frequencies, use a broadband matching transformer ; )

Instead of having a secondary with more turns, you use a secondary
with less turns which gives more current. I was able to get a single
IRF840 to produce a couple hundred watts of power, from 5 - 10Mhz
like this in an RF amplifier I made. The gate transformer had I think
7 turns on the primary, and two on the secondary. The impedance of
the gate is very very low at these frequencies ; ) For the IRF840, it
is 17 ohms -at- 7Mhz.

I also got a TC4420 to switch at 7Mhz. The output was so far from a
square wave it wasn't even funny...oh wait, yah it is lol.

 > I must wonder what in IGBT with a resonated gate could do...

Well, you can kill the gate from overvoltage ; ))). Nah, you'd just
have to be careful. Resonant gate circuits are all over the place in
the RF world, the technique works very very well.

 > It takes some real "power" to run the gates that fast, but it may
 > be a way to force IGBTs into the switching mode at 100's of KHz...

Yah, tens of watts straight into a gate. I've noticed that the part
needs a heatsink just from I^2R loss in the gate resistance! MOSFET's
meant for use at RF have metallic gate structures with very low
resistance, whereas SMPS/switching FET's are polysilicon.

Take care,

Justin Hays
KC5PNP
Email: justin-at-hvguy-dot-com
Website: www.hvguy-dot-com