"Franc Zabkar" wrote in message
...
On Mon, 19 Nov 2007 08:50:51 +0000, Eeyore
put finger to keyboard and
composed:

Arfa Daily wrote:

they have tiny 85 deg caps fitted that are rated about 2 volts above
what they

have to work at

There's nothing wrong with that voltage rating. Why do you think they have
a
working voltage rating ?

Graham

I'm aware that solid electros are not relevant to this discussion, but
I found the following datasheet useful because it discusses failure
rates for aluminium electrolytic capacitors:

NSP SERIES - LOW ESR SOLID ALUMINUM ELECTROLYTIC CAPACITOR
http://www.niccomp.com/NSPAppGuide0503.pdf

There is a table at the end which represents a failure rate model as
outlined in MIL-HDBK-217, the Military Handbook for "Reliability
Prediction of Electronic Equipment".

The formula for failure rate (FR) is ...

FR = (BFR) x (temperature factor) x (voltage factor) x (capacitance
value factor)

BFR = Basic Failure Rate for Aluminum Electrolytic Capacitors (per
MIL-HDBK-217F)

FR is influenced by:
- temperature of operation
- applied voltage to voltage rating (voltage ratio factor)
- capacitance value factor

For example, the FR for a 1uF 100V electrolytic capacitor operating at
25C 30V is 1 x BFR.

The same cap operating at its rated voltage would have a 14x greater
failure rate. The voltage factors at voltage ratios of 0.6, 0.7, 0.8
and 0.9 are 2, 3.2, 5.2 and 8.6, respectively. This suggests to me
that any design that allowed for a 50% margin would be 5 times more
reliable than one that operated at rated voltages.

If typical "wet" electros follow the above pattern, then Arfa's
observations would be right on the money.

By comparison, solid electros (as used in current motherboards) appear
to be a *lot* more reliable and according to the datasheet suffer no
degradation in ESR, capacitance or leakage current after being
operated for 7000 hours at rated voltage and temperature (105C).

- Franc Zabkar
--

That goes along with what Jeff Liebermann said on the subject as well, and
goes a long way towards explaining what I daily observe. It should also
knock on the head, any and all future discussion as to whether electrolytics
should be rated at +30% of applied voltage, for long-term reliability ...
Thanks for your interesting input on the subject, Franc.

Arfa