Re: OLTC Maggy modelling

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

Original poster: "Stephen Conner by way of Terry Fritz <teslalist-at-qwest-dot-net>" <steve-at-scopeboy-dot-com>

At 11:34 13/04/03 -0600, you wrote:
>Original poster: "boris petkovic by way of Terry Fritz 
><teslalist-at-qwest-dot-net>" <petkovic7-at-yahoo-dot-com>
>
>
>---
>Something is fishy again..

To make a switch that is "open" for both polarities, you need two IGBTs, 
one with emitter grounded, the other with collector grounded. IGBTs either 
have an antiparallel diode, or can't tolerate reverse voltage. Therefore 
you have to put an extra diode in series with each IGBT to protect them. 
The resulting circuit uses two IGBTs, but has more losses (because of the 
diodes) than a single IGBT carrying the same current. I estimate 1.5x more.

If you put the same two IGBTs in parallel (as in an ordinary OLTC) you 
would have one half the losses of a single IGBT. So I suppose the bipolar 
switch has about three times the losses, not six. Whoops!

My point still holds: It's an inefficient use of costly silicon.

>Never mind.In contest of
>IGBT brick ,I was thinking that circuit with IGBT
>brick can provide you with much higher bang size
>without going in 1000 BPS something rates for higher
>power input.

In theory it could, since some bricks are good for 2000A, which would 
translate to 20000A peak in OLTC duty. That's about 10 Joules bang size, 
equivalent to about 50 of the IGBTs we use just now.

Terry Fritz is of the opinion that IGBT bricks are useless for an OLTC, 
because the package inductances are too high, causing uneven current 
sharing between the dice. (A brick typically has about 10-20 IGBT dice 
inside connected together by relatively long/thin busbars.) Greg Leyh 
trashed a number of 2000 amp bricks while working on the klystron 
modulator. However he was hard switching thousands of amps with a very fast 
di/dt. The OLTC is slow in comparison. If the emitter circuit stray 
inductances in the brick were no bigger than about 50nH it would probably 
work OK. (How many inches of busbar is that?)

>How high K could be increased in present 2-coil system
>for primary circuit still be able to
>quench.K=0.28,more?
>regards,
>boris

Quenching can happen at any time in the OLTC. But there are other reasons 
not to use very high coupling: flashover between primary and secondary 
caused by close spacing, racing sparks caused by exciting higher modes of 
the secondary, and excessive voltage coupled back into the primary after 
quenching. With the charging circuit that I use, this last one is by far 
the biggest problem.

Steve C.