Re: Capacitor voltage - AC or DC

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: "Gerry  Reynolds" <gerryreynolds@xxxxxxxxxxxxx>

Hi Terry,

Often, "test to failure" will yield a lot of info about a part (this 
you have already done).  Another approach is to run an accellerated 
test.  In this case, run your high AC voltage at less than instant 
failure and log the hours it takes to failure.  Start the test at 
close to the instant failure point and work down to a less stressful 
point and you might be able to extract a curve without a horrible 
amout of time.  The accelerated tests require an appropriate sample 
size to be meaningful, but I believe it is doable.

Gerry R.

> Original poster: Terry Fritz <vardin@xxxxxxxxxxxxxxxxxxxxxxx>
>
> Hi,
>
> Somehow, someday, somebody, has to do something... :-))
>
> Probably the "last unknown" of MMCs is how long the (CD 942C20P15K) 
> will tolerate say 60Hz at 6000Vp-p....
>
> Maybe do a bunch of different voltages to make a nice graph of 
> 'lifetime vs. over voltage'...
>
> That would tell us all we still want to know...
>
> Cheers,
>
>         Terry
>
>
>
> > One usually overrates the cap for AC use.  I suggest Erms x 2 as a 
> > correct DC rating, so with your 15 kV AC xmfr, for a near bullet 
> > proof cap bank, you want a cap bank rated at 30 kV DC.
> >
> > If you are using 0.15 uF 2 kV caps in your strings then each 15 cap 
> > string would be rated .01 uF at 30 kV DC.
> >
> > With a 30 mA xmfr you need approx .014 uF capacitance so you cheat 
> > a bit and use 11 caps in each string.  This gives you a single 
> > string rated .0136 uF at 22 kV DC which will work fine with your 30 
> > mA xmfr.  You try to avoid a .01 uF match with the xmfr as the cap 
> > bank might resonate with the xmfr and damage the xmfr.  Going a bit 
> > higher usually works very well. 1.2-1.3 x Cres.
> >
> > Dr. Resonance