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

Re: Resonance - what is it? and OLTC stuff



Original poster: "Terry Fritz" <twftesla-at-qwest-dot-net>

Hi Jonathon,

At 12:29 AM 8/19/2002 -0400, you wrote:
...
>
>I also noticed, that SAY one of the IGBTs was to let out its smoke, the coil 
>would continue to operate safely, correct?  Each IGBT/cap
> pair is independent of the others, so no harm to the others. Sure, It would 
>be de-tuned, but whats that matter, when you got a crowd? Am I right?
>

First of all, we try not to let the smoke out ;-)  There are three ways to
kill an IGBT - heat, voltage, and current.  In our OLTC application, heat
is simply not an issue.  If the voltage on the IGBT gate goes above 50
volts or the voltage on the collector goes about 950 it will fail.  That is
why we use MOVs and transorbs that clamp and limit these voltages:

http://hot-streamer-dot-com/temp/OLTC08-18-10.gif

We do have to be careful not to destroy these parts too, but they can take
hits far better than the IGBTs can.  They also fail shorted which basically
shuts things down to a safe level. 

Current will kill our IGBTs if anything does.  It is easy to drive more
than 200 amps through them.  So we have to watch current very carefully.

Streamer hits to the gate, and thermal cycling are other possible failure
modes but they are minor.  Very fast switching of the IGBT is stopped by
the 100 ohm resistor.  We are in no hurry to turn it on or off in this case.


So, if one IGBT fails, what happens.  It is almost certain that the IGBT
will fail to a dead shorted condition.  This will drop the voltage and
current to the other IGBTs as the bad one "eats it".  This effectively
protects the others from damage.  There is a chance the collector voltage
will find it's way onto the gate which could cause damage to the other
IGBTs, but a 100 ohm resistor and 20V transorb stop that path.  Soon the
shorted IGBT will start drawing lots of line current.  The 5 ohm charging
resistor limits that current to about 100 amps long enough for the main
fuses to blow.  So the damage is pretty limited to the IGBT and two cheap
resistors.

If The IGBT fails open, the frequency of the primary may shift.  In that
case, the energy of the caps is not going into the secondary.  But, it has
to go somewhere.  We may be able to carry the oscillation and just end up
with the caps almost fully charged like nothing happened.  But really, the
IGBTs and the caps are going to have to dissipate the energy.  Unlike the
"old days" when the spark gap would just get loud and bright to dissipate
the heat, the IGBTs will now take the load.  In general they will have to
take 1/2 the system power or about 450 watts of heat.  That is 45 watts per
device which is no problem (the heat sink is really there to take such a
condition).  There is plenty of time to shut the coil down since it will
not shut down by itself.  As long as the coil is not run so long that the
heat sink gets over 75C, it will be fine.  Of course, this is all the
"idea".  Unforeseen things may happen too, but at least we have caught the
obvious ones.

Cheers,

	Terry


>Good luck everyone,
>
>---------------------------------------
>Jonathon Reinhart
>hot-streamer-dot-com/jonathon
>