Drifting MMC Cap Biopsy

Hi All,

	I opened my EMMC cap that seems to be drifting lower in capacitance value
and tested each individual cap.  All the caps were very even and there were
certainly no unusually low ones.  I did not find anything unusual except
that the average of values was very slightly low I thought.  I removed the
lowest value cap for inspection and replaced it with a new one and put the
EMMC back together.

	I snapped the epoxy case off the removed cap and a brand new cap for
comparison.  It was fairly easy to unroll both.  

	I did not see any signs of internal arcing or lost or vaporized film.
Both parts looked pretty much identical with tiny flaws and imperfections
here or there but nothing remarkable at all.  I certainly didn't see 0.7%
of the metal missing or anything like that.  The end metal looked the same
on both parts.  I carefully examined the lead attach area on the new and
removed parts but they looked identical (I think they should extend the
lead further into the cap for greater strength).  I did not particularly
like the way they attached the leads but it looked like it was doing fine
in this case.  I carefully went over the majority of the plastic and
metalized layers with a magnifier and held it up to the light looking for
little holes and such but saw nothing at all unusual.

	However, I do think I found the cause of the capacitance reduction.  After
about an hour of scratching my head and looking at the layers I noticed
that the removed cap seemed to have more little air voids in the many (4)
layers of poly (those layers are far less than a mil thick!).  The new cap
had less voiding and the voids looked just like areas that had separated
during the unwinding of the cap.  However, the removed cap had more voids
and many had a sort of rainbow colored cast or sheen to them.  The voided
areas also seemed to be more crinkled are stressed than in the new cap.  In
some cases the voids were a little cloudy.  The voids seemed to occur at
the edges of the plates were they would pass over a plate of opposite
polarity.  I posted a diagram of this at:


	I think there may be ionization or heating occurring in these high field
stress areas that is causing the layers of poly to separate and cause the
slight reduction in capacitance.  The areas do not look especially damaged
but they do seem to have some kind of nasty process occurring in these areas.

	Perhaps this is the "ion inception level" mentioned in some data sheets.
Perhaps those low AC voltage ratings do have some other reasons besides RMS
current behind them.  This problem seems to not be related to the metalized
film and may be worse in true metal layer parts.  I see no reason that the
Panasonic parts would be the only types affected by such a thing.

	I would suggest that anyone using MMCs far past the AC voltage rating (I
went far past the DC rating too) should monitor the caps value with use.
Perhaps we have a problem.  My EMMC cap is purposely run over voltage just
to check for this stuff.  Looks like I may have found something.  I would
think this drift would tend to slow down over time and perhaps the poly
would continue to hold up well.  However, the warning signs are definitely
there...  In all other respects, the cap looked fine.  No doubt the MMC can
be made to be practically bullet proof.  But we seem to be finding every
known failure mode of poly caps in the learning process...  Of course,
that's what makes it so much fun! :-)

Much to ponder...