Re. Partial Discharges.....A good reason for cap failure?

Hi Reinhard:

The phenomenon you describe is exactly what took out my last rolled poly
cap.  Scott Hanson did some microscopic analysis of a sample of my failed
cap's poly dielectric.  He sent me a photo of the poly just beyond the
foil edge towards the margin, and all along that foil edge was this
characteristic branching etching pattern.  

I've appended this photo to the capacitor page on my web site:

Regards, Gary

>Original Poster: "Reinhard Walter Buchner" <rw.buchner-at-verbund-dot-net>=20
>Hello All,
>A while ago, we were discussing the phenomenon of partial discharges
>within caps on the GTL. I have never seen this mentioned on the
>Pupman List (okay, I have only been on this list for 3/4 year), so I
>thought I would mention it:
>It seems as if all HV RF caps using PE, PS, PP, etc. (read: plastic)
>as dielectrica will experience the phenomenon of "partial discharge".
>Now, what exactly is this?
>Partial discharges are local corona discharges that occur to some
>extent in just about any HV isolation material. They occur because
>of non uniform material (thickness), minute dirt particles, entrapped
>air and so called "Lunker areas" (very, very small hollow areas).
>Corona and the free radicals (O3, NOx) it produces, will split up
>the long polymer chains, that the plastic is made of. This
>structural damage to the plastic is also called treeing, because
>the shape of the damage sort of looks like a tree. The insulation
>voltage (volts/mil) within these zones is sharply reduced. It takes
>a while for this to happen, but it DOES happen. As soon as the
>V/mil has been "eaten" low enough, the dielectrica will be punctured,
>of course, leading to cap failure. Elevated temperatures will increase
>the rate of destruction. This might explain, why even commercial
>PS caps (very lossy at RF frequencies) will die very quickly in HV/
>RF enviroments. I=B4m not quite sure how the self healing aspects of
>some commercial caps will counteract the phenomenon of partial
>discharges, but no homemade PE cap has any self healing
>capability at all.
>Any kind of disturbance (i.e. non uniformity) in the E-Field (sharp
>metal edges for example) will also increase the production of
>partial discharges (->corona formation). Important to know is that
>this phenomenon does NOT lead to instant failure, but rather, it
>accumulates over time and then leads to failure. I believe this to
>be one of the major reasons why commercial pulse caps are built
>with a very low volt per mil rating (lots of series caps). And now it
>should make sense why coilers like Bert Hickman (72-96 mils)
>or Ed Sonderman (90 mils) suggest such "gigantic" mil numbers
>for homebrew caps.  Oil DOES NOT prevent the formation of
>corona!! Of course a pulse cap always has a limited lifetime,
>simply because it is a highly stressed component.
>Interestingly enough, mica caps are totally insensitive to partial
>discharges. However, mica caps are VERY sensitive to voltage
>spikes exceeding their rated limit, so they really aren=B4t any
>better in coiling usage :o).