[Prev][Next][Index][Thread]
RE. Many dry caps in series?
>Original Poster: Tesla Technology Australia Pty Ltd
<tesla-at-draal.apex-dot-net.au>
>> I'm considering a new capacitor design that uses 20 rolled poly
>> capacitors in series, each using a single 6 mil LDPE dielectric. I would
>> again use extended-foil construction with household aluminum foil. Would
>> such a construction reduce the voltage per cap enough that it wouldn't
>> create corona and be necessary to displace the air with oil? I'd love to
>> avoid that whole oil-thing!
>Hmmm, I've found LDPE to have a rough puncture voltage rating of around
>420v/mil -at- 100KHz (2KVA). For an example scenario, say, 20 caps each with
>a single dielectric layer of 6mil LDPE, means each cap (ideally) rated to
>2520 volts -at- 100KHz. 20 in series thus yeilding a cap rating up to 50KV.
The total proposed dielectric thickness is 120 mils. I've heard a few
(probably wrongly) suggest that 60-90 mils is enough, so I figured 120 should
be bombproof. I've also heard that the dielectric strength per mil goes
way up as the thickness goes down.
>Suggestions for open-air construction - make the edge-spacing for the foil
>plates much larger, anywhere from atleast 1.5". ie, ensure that there is
>atleast 1.5" outlay of LDPE around the edge of the plate.
Arn't edge-spacing considerations to prevent surface track arcovers and
unrelated to corona? I would have thought that as the number of units in
series goes up, the required arcover gap goes down proportionally. In my
ill fated 2-in-series capacitor, I used 1.5 inches in each and had no
arcover problems.
>Results from various experiments I've performed with LDPE caps has found
>that in air, corona has always propogated, but not necessarily caused a
>problem, ie breakdown, other than energy losses.
I think that any corona, in contact with LDPE, would be a bad thing,
given the very low melting point of LDPE.
>Original Poster: "Steve Young" <youngs-at-konnections-dot-com>
>
>Exactly my thinking also. I second your request for knowledge and
>experience by others with such a scheme--anyone done it? My added
>thought is the concern that if a couple of sections were to fail, then
>the rest of the sections would be voltage stressed and would likely also
>fail prematurely. I believe it would be a good idea to put safety gaps
>across each series cap. I also think it is risky using a single layer of
>dielectric, such as home depot 6 mil drop cloth whose manufacturer
>probably doesn't care if it has a few pinholes in it. Even at only a
>couple of KV per series cap, I believe it would be safer to go with, say,
>3 layers of 2 mil, which would virtually assure flaws in layers would not
>end up being aligned.
Your point about the single layer dielectric failing from pinholes is
valid. My problem is that with the extended-foil construction, one loses
sight and access to the poly edges when rolling, and things DO shift! With
a single layer of poly, the foil is taped to the poly so the poly's
position can be judged and corrected during rolling via the exposed foil's
edge. If there were multiple layers of poly, they would have to float
relative to each other, and any lateral shift would be undetectable. I'm
going to have to do some rolling experiments.
I also plan to make a small 20 segment series dry capacitor to evaluate
whether corona is still a problem with such a design. I'll run it in
parallel with my backup capacitor and it it doesn't fail, dissect it and
look for signs of corona damage to the poly.
Regards, Gary Lau
Waltham, MA USA