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

Re: A new cap failure mode?

Tesla List wrote:
> 
> Original Poster: Terry Fritz <twf-at-verinet-dot-com>
> 
> Hi All,
> 
>         In an off list discussion, a topic came up that I have not seen
before but
> I always assumed may be a factor in capacitor failure.
> 
>         When one places a nice safety gap across a capacitor, and it is
firing for
> any reason, what is the current in the discharge pulse?  Since Tesla caps
> are designed to be very low inductance and low resistance, there is very
> little limiting the current when the safety gap fires.  If a cap is charged
> to say 20kV and the resistance of the mess is say 0.5 ohm - we get 40000
> amps!!  That is enough to do some real internal damage to any capacitor.
> The suggestion comes up that perhaps a safety gap placed directly across a
> primary cap needs a little resistance in the circuit to keep this current
> to a "safe" level.
> 
> Comments or suggestions are welcome...
> 
>         Terry

Terry and all,

So THAT's why it makes such a loud bang! :^)

While the instantaneous currents can be quite high, the duration is also
very short, and the physical inertia of the elements in either a rolled
or flat-plate cap should prevent damage from electromagnetic forces. An
exception might be the very thin metalization in a ceramic doorknob cap,
but even here, the dielectric itself will generally limit the maximum
rate that energy can be "extracted". 

However, when the protection gap fires, the circuit will ring at a
frequency limited only by the cap's internal inductance and parasitic
inductance of the protection gap wiring. I'd suspect that these high
amplitude, high frequency (multi-megahertz) voltage reversals would
place great stress on the dielectric, especially if the gap fired with
any regularity. Adding a series resistance in the gap circuit should
help damp these oscillations, limit the current peaks, and even knock
down the magnitude of the voltage reversal with minimal impact on the
overvoltage protection function... sounds like a pretty good idea! The
resistor must be physically large enough to withstand worst-case 
voltage however. 

-- Bert --