David Trimmell <dwt-at-efn-dot-org> writes:
> I'm still working on the design) and then it died! Today I dismantaled the
> cap (a messy job) and located a hole the size of my pinky, it was allong
> the edge of the flashing threw the plastic. I was very carefull in sanding
> and deburring the alluminum and inspected the poly prior to assembly.
> I would like to say that obviously the standard schools of thought do
> not apply in TC
> and especially resonator practise. I asumed 600 DC volts/mil breakthrew
> voltage and then divided by three for the TC voltage for 16KV. Now
> I would sugest dividing the DC breakthrew voltage by four, but would
> wonder about that?
Your comment that the point of failure was along the edge of the metal
brings up a possibility that I hadn't heard of before. I know you said
that you sanded the edge carefully, but a tiny missed burr, or even a
particle of metal falling back onto the sheet might do some interesting
electric field tricks. A sharp point, even one too tiny to be an
abraision or incision problem, can cause a voltage gradient right near
it that's much, much higher than the voltage across the plates, divided
by the plastic thickness, and might easily exceed the breakdown limit
for the plastic in just a microscopic region. Once a tiny dot of the
plastic breaks down, it should contain a lot of conductive carbon,
which could get hot, and break down further. The breakdown would then
avalanche into a short.
This is just idle speculation at this point, but it might suggest
a possible new breakdown mechanism for the homemade caps. (Or it
might represent an old, well-known breakdown mechanism that others
here have already discussed, and I was too sleepy to notice :) If so,
it also points to some ways to eliminate the problem. Do those who've
had unexplained failures see any way that a burr or leftover flakes
from the sanding step could have gotten into their cap?