[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Geek .15uF caps Was: "plate" capacitors



Original poster: "Terry Fritz" <twftesla-at-uswest-dot-net>

Hi David,

	Just for the record I'll give the long answer to you short question :-)

Metal foil/film caps can self heal after a short over voltage spike which
is a tremendous benefit to us.  They can absorb thousands and probably
hundreds of thousands of sort over voltage punches.  It appears the only
way to kill them with voltage is to get the voltage so high that they
constantly arc internally and burn up.  That is at about 3X their DC
voltage rating.  Hard to do in our case unless you really try.  The
transformer or something else will blow up first.

Current is our main worry and it comes in two flavors dV/dT and RMS.

The first concern is that the cap must handle short ~1% duty cycle bursts
at hundreds of amps at say 120BPS.  Even though there is not much heating,
those giant spikes will easily blow poor connections inside the cap.  That
is why we want caps that have a high dV/dT rating.  The problem area is
were the leads attach to the end plates (easy to fix) and where the plates
attached to the end plates (not so easy).  Cheaper film caps use evaporated
film for the electrode plates which have a very poor attach that will fail
under very high current.  The few millionths inch thick metal film on
plastic is just hard to connect to.  The foil/film caps have solid metal
foil for the end electrodes that is easy to connect to and will not easily
burn up.  That is why we always want "metal/film" high dV/dT types.

The RMS current is the "real" limit.  All caps have some loss which is in
effect like an internal resistor.  Polypropylene has the lowest loss for
common capacitor materials at about 5X better than anything else at TC
frequencies.  Thus, they run 5X cooler.  In Tesla coils, the RMS currents
can be very high and we push it to the limit since caps are expensive and
we want to run as much current as we safely can through the minimum number
of cap strings.  The power lost to heating the cap is simply proportional
to the RMS current squared as in a simple resistor.  

Polypropylene is a soft low temperature plastic that does not like heat at
all.  85C is the limit although they play a few games to get that as high
as 105C.  One should not run them with the outside temperature getting much
over 10C above the air temperature.  Even thought the cap is say 40C on the
outside, the poly film is a very good insulator so the heat generated in
the center gets traps and the middle may be cooking at 110C where it melts
and fails.  So the thermal design of the things could stand a ton of
improvement to even out the temperature.  The metal foil plates help draw
heat out a little but it could be far far better.  "i" would make them
hollow so the cap is more of a tube.  Then the heat would not build up in
the center and I think you could get far higher currents through them
without overheating.  Most of the capacitance is in the outer layers anyway
so making them hollow would not reduce the capacitance/volume that much.  I
would not be surprised if you could get 4X the current through one.  Maybe
Chris could mention this to his pals at CD ;-))  You need fancy stuff to
make poly film/foil caps so it is not a home project...

Unlike tcp/ip DNS server network stuff, I do know a bunch about poly caps
:-))  The hollow cap idea seems to me like a really promising thing to try.
 Maybe it could be patented buy some big cap manufacturer and make them
rich.  Of course, we could use them too.  If it were a big deal idea, I
would not care about the patents and all that crap.  I just want the caps
for my coil :-))

Cheers,

	Terry


At 08:59 AM 4/20/2001 -0500, you wrote:
>> Original poster: "Terry Fritz" <twftesla-at-uswest-dot-net>
>> Personally, I would go by RMS current vs. center temperature rise.  A
>> hollow capacitor, sort of like a tube, should have much higher dissipation
>> since the highly thermally insulated center is removed
>
>Is the power handling primarily thermally limited?
>
>DK
>