Cap AC voltage ratings

To: tesla-at-pupman-dot-com
Subject: Cap AC voltage ratings
From: Gary Lau 13-Mar-1999 2038 <lau-at-hdecad.ENET.dec-dot-com> (by way of Terry Fritz <twf-at-verinet-dot-com>)
Date: Sun, 14 Mar 1999 09:14:06 -0700
Approved: twf-at-verinet-dot-com

I have begun making a new cap using strings of 16 Wima FKP-1 .015uF/1600VDC
polypropylene caps.  The thing that concerns me is voltage ratings.
While the DC rating seems to be no problem (16 x 1600 = 25.6KV), I'm
concerned with how much the mfgr derates the DC rating for AC.  My caps
are AC rated for only 500VAC each (16 x 500 = 8KV).  I believe the two
reasons for the derating are 1) to allow for the 1.414X RMS to peak
conversion, and 2) to limit the temperature rise due to ESR dielectric
heating.  The graphs I have seen show a derating curve, being a function
of applied frequency, and I would expect this.  The unexpected thing
however is that the curves do not converge to a simple derating of 1.4X
at a sufficiently low frequency, but instead, never allow over the rated
AC voltage, regardless of frequency.

While the DC rating of the caps may not be in danger, the AC rating
appears to be a problem.  Or is the AC rating that low only if given a
continuous duty cycle?  While the tank oscillations in TC use occur at
very low duty cycles, the 60Hz component is continuous.  But then, we
typically don't run our coils for more than a few minutes at a time.

So, does anyone have a comforting explanation that will make be believe I
won't be blowing by caps due to violating their AC ratings?

Gary Lau
Waltham, MA USA