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

Re: Re. Cheap 20nF 20kV pulse caps - If it would work??

Hi Bert,

At 08:08 AM 11/1/98 -0600, you wrote:
>Tesla List wrote:
>snip
>
>Gary and Terry,
>
>For Tesla Coil application, it isn't. Equalizing resistors are necessary
>if the series of capacitors are subjected to extended operation at HVDC.
>Each cap may have significant variation in leakage resistance, and under
>extended DC excitation, the voltage distribution can be quite uneven,
>particularly if the capacitors came from different manufacturing runs.
>Adding equalizing resistors prevents this from happening - the large
>value equalizing resistors are still much lower than the leakage
>resistance of the caps, swamping out the unequal leakage resistances.
>Because of the very long effective Rleakage*C time contants, this effect
>only shows after extended DC operation. For Tesla Coil operation, each
>cap have about the same equivalent series resistance (ESR) so that tank
>current heating effects are shared relatively equally. Using identical
>caps from the same vendor should pretty much guarantee this as long as
>the peak currents are not extreme. 
>
>Offhand, I don't see any real problem in going with the proposed design
>approach for Gary's homebrew or Terry's commercial series cap, since the
>voltage stresses should be below the point of internal corona formation.
>Terry's approach has the added benefit of sharing the current stress,
>and that a cascade failure would be constrained only to a single chain.
>And Terry's design should not require an oil bath for cooling, since 320
>individual caps would provide excellent power dissipation to free air. 
>
>Good luck to both of you!
>
>-- Bert --
>
>

	Thanks for the input.  I will go ahead with this project in a limited way
at first.  I am going to leave the balance resistors in since they don't
hurt anything and I want to run them in air where corona, finger prints,
etc. can add to leakage significantly.  I may try leaving them off a string
to see if it helps the string to blow up :-)
	It is interesting that Richard suggest running them at one third their
rated voltage.  Since measurements indicate that they will stay within
their ratings they should live.  However, the "3x rule" is very popular.
It will be easy to experiment with blowing this cap up since it is very
cheap and easy to repair.  Perhaps I can gain some insight as to why
running caps at their rated voltage tends to lead to failure in TC
applications.  
	It is interesting that the two nice caps Chip blew last night were of
floating plate construction.  This could theoretically allow the floating
plate to obtain a net DC charge which could result in failure.  Hopefully,
I will be able to examine some of the remains.  We use high voltage caps at
work that have floating plates that have all kinds of "fun" failure modes.
I personally am quite concerned that series caps may fail due to net DC
charges building up even under AC conditions.  The work caps run at 13MHz.
	I suspect most cap are blowing due to either resonant charging causing
over-voltage breakdown or deterioration of the dielectric.  My little test
capacitor setup should be able to supply me with the raw dead bodies needed
to analysis this problem.  This should be very easy to determine upon
examination of the cap's internal parts.  Over voltage will appear as a
clean explosion.  Over dissipation will show as damage in the surviving
parts.  In Chip's case, If the other three cap sections look brand new with
no dielectric deterioration then the cause is very likely to be over
voltage.  However, if the other sections are discolored, delaminating, etc.
then over dissipation is the cause.

An interesting and valuable area of study... 

	Terry