Original poster: "Lau, Gary" <Gary.Lau@xxxxxx>
I agree that coronal rectification is the most likely explanation, as a
non-linear process is needed and that's the only plausible one that
comes to mind. But I don't think it is due to differences in component
layout.
Reviewing my experimental results posted at
http://www.pupman.com/listarchives/2004/January/msg01310.html, I
miss-spoke - the cap value differed between trials with different cap
types. But to summarize the results:
Charge? Cap type/value
No (4) CD942C 0.1uF/2000V No (4) BC Components 383-MMKP 0.1uF/1600V
Yes! (4) BC Components 376-MMKP 0.047uF/1600V
The CD caps are the axial units we all know so well, and no residual
charge was observed. The two different BC caps were radial-lead box
caps with identical case sizes and assembly geometries
(http://users.rcn.com/laushaus/tesla/minicoil_files/mmc.jpg).
The only other non-linear process I can think of is dielectric breakdown
and puncture, but I hate to think that it happens so regularly!
Regards, Gary Lau
MA, USA
> Original poster: "Antonio Carlos M. de Queiroz" <acmdq@xxxxxxxxxx>
>
> Tesla list wrote:
>
> >Original poster: "Lau, Gary" <Gary.Lau@xxxxxx>
> >Bleeding and balancing are two unrelated things.
> >The mechanism by which unequal charges remain on individual MMC caps
is
> >not understood. It is not due to unequal capacitance values. In my
> >experiments, some capacitor types are prone to developing these
charges,
> >and others are not, using exactly the same cap values and test
circuit.
> >I have no idea what causes this difference.
>
> Probably corona at the interconnections, effectively adding rectifiers
> across the capacitors at random positions (or systematically,
depending
> on how asymmetrical the construction of the connections around each
> capacitor is).
>
> Antonio Carlos M. de Queiroz
>
>