Re: gEDA-user: electrolytic capacitors

[Gene Heskett <gene.heskett@xxxxxxxxxxx>]

On Friday 26 December 2008, Peter Clifton wrote:
>On Fri, 2008-12-26 at 21:00 -0500, gene wrote:
>> What does it mean when manufactures specify 'Endurance @ temp' or 'life
>> time @ temp'?  For example, one company claims:
>> Endurance : 105 C 1000 h to 5000h
>> Then they further state that after testing at 105C with maximum ripple
>> current applied, that when the part is cooled to 20C, it will then meet
>> original specs.
>
>It will still have had its life-time shortened.
>
>The spec you quote above, AIUI, means that after somewhere between 1000
>and 5000 hours of operation at 105 degrees centigrade, the cap (as
>measured at 20 degrees) will be outside of its quoted specifications.
>
>I can't remember the exact numbers, but derating to lower temperatures
>has a stupid factor of lifetime increase. It might even be as much as
>double the life-time for every 10 degrees drop in working temperature.
>
>I guess that merely being at elevated temperature is also lifetime
>degrading. You circuit wouldn't need to be on for the electrolyte to be
>degrading.

As long as the seals hold, the electrolyte won't degrade.  Its not the 
electrolyte that causes most caps to go out of tolerance.  The most often 
seen failure in recent years has not been that of reduced capacity, but sky 
high ESR.  I have seen caps marked 220 uf at 10 volts read 240 uf on a std 
cap tester, but toss a "capacitor wizard" at them, which tests them with 
little or no bias, measuring instead their series resistance using a 100khz 
low level signal, and they will be 5+ ohms at 100khz.  Digital circuits go 
crazy when their supply rails are being bypassed with such a capacitor.

What happens when you have such a capacitor in a switching power supply?  
Easy, those things can put 10 or more amps of ripple currents through such a 
capacitor, and 5 ohms with 10 amps rms flowing through it equals an explosion 
as that is 50 watts of power being dissipated in a little can the size of 
half a joint of your little finger.  But it normally doesn't come to that, 
the voltage surges they allow as the ESR rises above the 2 ohm range usually 
blow one or more of the switching transistors, followed 50ms later by the 
line fuse.

The culprit of course is the connection between the terminals on the bottom of 
the can, and the microscopicly thin alu foil the capacitor is made of.  A 
truly 'gas tight' joint simply cannot be done when the foils are so thin you 
can literally see through them.

For best life, keep the caps as cool as you can, and if the capacitance is to 
be maintained over an extended time period, then the applied bias should be 
at least 80% of the nameplate rating.  Reduced below that, the alu will 
chemically 'deform', meaning the acid like etching that creates the 
dielectric film at 10 to 50 times the measured square inches, will eventually 
smooth enough to cause a loss of the dielectric area, and hence a loss of 
capacity.

The real plates of an electrolytic capacitor are not the two separated alu 
foils, but the alu oxide coating of the etched foil.  The glycol essentially 
is there for the ride (it has a high dielectric constant and isn't really a 
great insulator) and its etch restoring properties under sufficient bias.

If any of you do service work involving such parts, one of the first things to 
look for is an alu can with a bulged top (replace it, then measure to 
confirm :), which tells you the ESR is high and that cap is running hot 
enough to boil the glycol, possibly blowing the seal in the bottom, but 
occasionally will split the top open on the scores in the alu can put there 
for exactly that purpose.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Q: What do you call 50 Microsoft products at the bottom of the ocean?
A: A darned good start. 


_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user