Conductor failure mechanism (was: Re: Bubbles)
Subject: Conductor failure mechanism (was: Re: Bubbles)
Date: Sat, 14 Jun 1997 09:21:31 +0300 (EET DST)
From: Harri Suomalainen <haba-at-cc.hut.fi>
To: Tesla List <tesla-at-pupman-dot-com>
On Fri, 13 Jun 1997, Tesla List wrote:
> Speaking from experience with run-in-air LDPE rolled caps, main thing
> is heating at any favourable corona spots (foil edges) wnich seems
> to actually "crack" or decompose the plastic. Shortly before the caps
That's surely the hardest spot usually. The actual mechanism depends
on electric field strenght. Say you have a strenght of 10kV/mm at some
point due to the electric field. If the insulator cannot handle that it
will break down. Usually high field strenght points (in caps) are near
the edges of foils. However, in different setups they can appear in
That's why they use sometimes toroids and other objects in power
transformers etc to get the desired field distribution. Simulations
are nice for studying those effects..
> that I described in that article fail, the sheeting inside them is
> noticeably oily and has a strange pong to it. Appears the plastic
> degrades to the point where a spark eventually punches through.
In a rolled cap you have a bigger surface area on the outer side than
on the inner side. Say you have two foils and insulator: inner foil
and outer foil point have same amount of total flux (coming from one
foil goes to the other foil). I'm now talking about the situation
far enough from the edges. Flux for some unit area is therefore bigger
on the inner side where total flux is distributed on a smaller surface.
That will lead to higher field strenght (kV/mm) on the inner side.
Therefore there will appear small discharges (so called partial
on or near the inner side probably. Every tiny discharge will distroy
some plastic. The spot will become more conducting and the next
will be bigger. This will continue untill ultimately it breaks down from
one foil to the other. Cap has been destroyed.
Air bubbles are related to this phenomena. If you have some flux at some
area the field strenght will be dependent on flux/permittivity. If your
insulator has permittivity e=2 and air bubble has permittivity e=1 the
field strenght in the bubble will be twice the strenght in the
No wonder there will appear a partial discharge ("a tiny spark") in the
bubble. Then like previously described it will eventually eat its way
through the insulator and cap will fail.
Commersial caps are usually measured for partial discharges. Measuring
them is actually quite easy (in theory at least :). After every tiny
discharge the cap will draw a small current pulse which can be measured.
Usually acceptable discharged are in the range of 10-100pQ or so.
Depending on the type of apparatus of couse.
Unfortunately there is another nasty effect. If you use cap at some
voltage there may not appear any partial discharges (PD). However, if
momentarily apply more voltage there may occur one PD and the "channel
formed" will conduct thereafter even at a lowered voltage. That's why
equipment is usually measured with applying a few seconds of overvoltage
(for onsetting the PD's) and then measured at rated voltage for a minute
Another nasty feature is due to the capasitive background behing the
Say you have a mains freq. cap. It will have a PD multiple times in a
cycle. That's just too bad. It also causes rf emi. It can and sometimes
detected by measuring the emi.
In dc caps the situation is somewhat different due to difference in
capasitive effects etc. I'll not go into that becouse ac is definately
more interesting to coilers.
All in all: if caps are made to fail even a bit they will wear out
finally. Measure caps if possible. Unfortunately almost no average
coiler is probably able to measure home-made caps. Keep things simple
and fold the edges of foils (edges less sharp). Keep the worse points
in the inside of the foil in a rolled cap. Field strenght will be
smaller there. The difference may be vanishingly small if the insulator
is quite thin. It will not hurt in any case.
We have phone numbers already, why would we need IP-numbers! -unknown
Harri.Suomalainen-at-hut.fi - PGP key available by fingering