[TCML] The difference btwn a SIDAC and a SIDACtor.

[bartb]

Finn Hammer wrote:
> In my SISG scheme, where I bypass most of the SIDAC/SIDACtor's with a 
> SCR, the current can easily be supplied from the tank capacitor, (When 
> the SCR is trigged, the action is much like kicking the stool away 
> from under the feet of a poor sod with a rope around his neck) and 
> since the break over voltage rating of the one remaining SIDACtor in 
> each section is very low, a successfully triggered section cannot be 
> over volted.
Now that's an "analogy"! I like this triggering scheme Finn. It makes 
sense to use a fractional sidac value on the top end keeping the voltage 
across the IGBT string nearly identical.

> But I am ultimately depending on the crowbar action of the 
> SIDAC/SIDACtor string: Some of the stacked sections are bound to get 
> triggered later than others, and these sections rely on the breakdown 
> of the full string of SIDACtor's to protest the IGBT from over 
> voltage. At least that is my cautious point of view.
This is what makes the SISG successful and why we can run several IGBT's 
in series (24 in my case). The full string is required to fire before 
the 25V TVS rings the doorbell. How close the TVS components are to one 
another ultimately is the cause of why all the IGBT's fire nearly at the 
same time (and turn off similarly). The SISG design is really amazing 
when you compare schemes attempting dynamic gate balance. From what I've 
read, there is the use of feedback charge (similar to SISG, but still 
very different) and RC snubber passives across CE to balance the rate of 
rise across the IGBT. None that I've seen use the TVS scheme that Terry 
brilliantly conjured up. The 25V TVS waits until the sidac string fires 
(bang), then it "quickly" performs it's doorbell ring at the gate of the 
IGBT, and thus, they all turn on nearly at the same time (Big Bang 
Theory). Turn off is just as cool!

For my purposes, the Sidactor will not be used. I'll stay with the Sidac.

Take care,
Bart