[TCML] MMC failure / question or advice.

[Barton B. Anderson]

Hi Vinnie,

Sounds like your over volting the caps for whatever reason. Eight caps 
in series is very weak at reducing the voltage stress across each cap. 
Maybe the gap is firing at a higher voltage these days for some reason? 
I personally would double up on the caps per string to stop the failures 
once and for all. Regarding the resistor, it's probably because the 
voltage across that individual cap was high enough to arc across the 
resistor and promote it's failure.

De-rating the cap standoff voltage is "always" a good idea. 2x is 
acceptable for most applications. There's also running time with caps. 
Once enough internal holes are blasted through, it's just a matter of 
time before they just can't take the abuse any longer. I think your caps 
have maybe reached that point. A good idea is to just rebuild with more 
caps per string and use new caps.

Take care,
Bart

Vinnie wrote:
> Hello people
>
>  Recently I have been running into some trouble with my coil and what I believe to be an MMC failure
> but not sure whats causing it.
>  I am running 8 series CDE 942C20P15-F .15MFD 2000VDC caps in my MMC. First failure burned a hole
> through two of these caps. I replaced the two bad caps with the same type and did two more 20 second
> runs which resulted in a loud snap. No visible damage this time but I could still hear a loud snap or arc 
> even after the coil stopped functioning. One bleeder resistor on one of my caps looks to have heated quite
> a bit but it doesn't look burnt. Primary power input is 2 12/30 NST's and a RQ style gap. This functioned
> great for years but it seems somewhere something is breaking down now. 
> Does anyone have an ideas what could be happening. I am thinking maybe it's time to build complete new MMC.
> Would anyone know if there is a better more robust MMC solution for my setup?
>
> Thanks
>
> Vinnie
> _______________________________________________
> Tesla mailing list
> Tesla at www.pupman.com
> http://www.pupman.com/mailman/listinfo/tesla
>
>
>
>