[R-390] Orange Drop Caps vs Film Polyester Axial

[Charles P. Steinmetz]

Dave wrote:

>Can you put some meat on these bones, please?  How is "unacceptable 
>self-healing event" defined?  What were the test conditions (e.g. 
>test setup, capacitor value, etc.)?

The test protocol used a slowly ramping voltage on the cap, up to its 
rated voltage, from a source with several hundred ohms 
impedance.  Every self-healing event reset the voltage to 
zero.  Every several hours there was a simulated power cycle (drain 
to zero, charge to rated voltage in 1/12 second, drain to zero and 
start ramp).  The voltage vs. time was recorded, as well as the 
charge dumped to ground during each self-healing 
event.  "Unacceptable" self-healing was defined as: (1) one or more 
clearing charge dumps in excess of a trigger value; or (2) more than 
(n) self-healing events under the trigger value -- each measured over 
the course of a month.  I do not recall the precise charge dump 
trigger value or the value of "n" -- I think "n" was 3, but many 
metallized film caps failed with "n" much higher than that -- 
sometimes 5-20 per day.  A typical pattern was a series of 
self-healing events at lower and lower voltages, sometimes with a 
"reset" to a higher voltage followed by another series at lower and 
lower voltages.  The "reset" event was usually a "charge dump in 
excess of trigger value" event.

I tested PE and PP metallized-film and film-and-foil caps, as well as 
Class 2 disk ceramics.  The caps were all new stock from quality 
manufacturers, acquired through the primary supply chain.  Values 
ranged from 0.005 to 0.22 uF.  (I also tested some large film caps -- 
5 to 100 uF -- and some more esoteric dielectrics -- but that is not 
pertinent to BA bypass caps.)

This was an extension of a project to characterize capacitors in a 
production environment where we needed to get failure rates over the 
equipment lifetime down very, very close to zero.  We also had 
extremely limited space, so there was a lot of pressure to use 
metallized film caps.  In the production test, a single self-healing 
event over the course of a month disqualified a capacitor.  No 
metallized film cap ever passed that test.

Best regards,

Charles