Re: Cap AC voltage ratings
Hello Gary, all,
>Original Poster: Gary Lau 15-Mar-1999 2241 <lau-at-hdecad.ENET.dec-dot-com>
>I think I'm gonna be sorry I dug into this...
Naw, Iīve been through this on the GTL, so no problem.
>The catalog explains this. There are two issues here.
>1) The AC voltage rating seems to be a fixed maximum AC level that is
>independant of frequency (though it's not clear why), to avoid
> "Furthermore the RMS voltage derived from the peak voltage shall
> not be greater than the nomimal AC voltage rating of the
> capacitor to avoid the ionization inception level:
>2) The dielectric strength is a direct function of operating frequency.
>The catalog has a graph plotting the normalized dielectric strength of
>polypropylene film vs. frequency.
>At 10Hz, the strength is 100%.
>At 100Hz, the strength is 81%.
>At 1KHz, the strength is 65%.
>At 10KHz, the strength is 48%.
>At 100KHz, the strength is 32%.
>At 200KHz, the strength is 30%.
>So, if I use a 1000VDC cap at 200KHz, it's only good for 300V, peak to
>peak! The low tank duty cycle doesn't enter into this.
>This dielectric derating part really worries me, despite the apparent
>success of several list members using these parts. I'm just gonna have
>to pretend I didn't see this...
What you shouldnīt forget is the fact that a cap manufacturer will
NEVER publish the max values a cap can take, because he wants to
keep his (good) reputation. Secondly, the frequency derating is for
EQUAL cap lifetime (at 50/60Hz). Now you have to make a
compromise. But on the other hand, will you be running your TC for
10^5hrs (typical cap lifetime cycle)? :o). I run my MKP caps at DC
= peak AC voltage rating and they survive. There is NO change in
temperature during a 5-8 min run. I donīt use a safety gap or
equalizing resistors across the caps (at the moment). During tuning
I had the xformer safety gap ignite many a time, but the caps
didnīt seem to care. I will be adding a 30% safety margin for my
high power setup. A friend of mine (Stefan Kluge) made an XLS
spreadsheet for cap derating using manufacturer specs. Real cap
killers are ultra high bps (>1000bps). His factory (!!) derating
factor for MKP was 4-5. (For MKT it is ~30!!) However, I found
this isnīt necessary (for MKP caps). Another advantage these
commercial caps have is their self healing ability.
Of great importance is to find caps that use metal foil and NOT
plastic with an ultra thin metallic plating. These caps usually
fail very quickly. I tried a set of Philips X2 MKP surpressor caps
(35 x 330nF-at-250Vac each). They just gurgled and that was it!
It seems they canīt take the high currents involved (just like the
cheapo ceramic caps).
Your best bet is to start out with a string (just add it to your
existing setup and retune) that has enough safety factor (4-5) built
in. Run your coil and start shorting out the caps one by one. As soon
as you reach peak AC = DC rating, quit (or they might fail). If they
fail before that, you now know the max limit your caps can take.
This will allow you to design a MMC that fits your needs without
blowing too many. A 30-40% safety margin should be well suited
for TC use (kickbacks, strikes, etc). You will need to retune the
coil for every shorted cap of course. Check and note the
capacitance of each cap before a run. After the run, but before
going to the next step (shorting an additional cap), re-measure
the capacitance. If the capacitance starts to change more than
0.5nF I would start being careful about shorting out more caps.
(This could save a string from destruction). After you find the
optimum construction for your MMC measure the total
capacitance and write it down (include the date). Every now
and then re-measure your capacitance, to see if it starts
changing big time. This will let you keep tabs on cap
BTW: Someone asked about a box of Philips caps he aquired.
They will have a number printed on them. For a complete spec
sheet, have a look at the Philipsī website. If you canīt find them,
send me the numbers and I will try to get a spec sheet for them.
Coiler greets from germany,