Re: [TCML] New SISG failure mode

[FIFTYGUY@xxxxxxx]

In a message dated 6/13/08 8:25:54 P.M. Eastern Daylight Time,  
bartb@xxxxxxxxxxxxxxxx writes:

>Have you found any heating of the Sidac's since our last  discussion? 
 
    Haven't run the coil since.
 
>I decided to leave them "unsunk" but staggered the Sidac fins for  more air 

>gap between them.


    I staggered the SIDAC tabs for more voltage  standoff between them. I 
know it's only a 300V theoretical difference between  the tabs, but they're all 
floating at several kV and passing lots of RF. Guess  it can't hurt...
 
    If the current is high enough, one single initial  pulse from the first 
bang could kill any given IGBT (or any semi). Heatsinking  wouldn't matter at a 
high enough current level, because it still takes a finite  time to conduct 
the heat out from the die. That's the "brick wall" I worry  about. I think it's 
more likely that the IGBT I recently found that died from a  secondary 
breakdown was probably suffering from steady-state overheating. If it  was on the 
ragged edge of the max pulsed current I probably would've killed more  than just 
one. OTOH, as more sections burn out in the SISG, instantaneous  current 
should go down because the cap will charge to progressively lower  voltages before 
firing. But the firing rate will go up, increasing the  steady-state heating. 
    I guess you could test it by firing a single bang,  and checking if any 
SISG modules died. Terry tested them to an extreme, but I  don't think we've 
found that precise level yet.
    The temp coefficient might help a little with  steady-state heating for 
the IGBTs, but I think we need to force-ventilate if  we're pushing the 
Super-247s in Pig use. My last run was with .185 uF primary  cap with my pig cranked 
to 17kV, and I was drawing about 30 amps steady state.  About 8kVA, and the 
heatsinks got hot rapidly.
    The other thing is keeping the voltage carefully at  a point where the 
BPS doesn't run away. A fine line, and I've been intentionally  pushing it to 
see how bad I can hurt things...
 
-Phil LaBudde

Center for the Advanced Study of Ballistic  Improbabilities



**************Vote for your city's best dining and nightlife. City's Best 
2008.      (http://citysbest.aol.com?ncid=aolacg00050000000102)
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla