[Soekris] Solid state capacitors

Poul-Henning Kamp phk at phk.freebsd.dk
Thu Jan 11 17:00:58 UTC 2007

In message <45d561bc0701110843h7cfed9a5l71b5a735c413b7f7 at mail.gmail.com>, "Mats
 Lundqvist" writes:

>ahaa....I did not know that, I always figured the caps to burst from the
>electrolyte expanding when frozen.

I've never seen that.

>But isn't thermal stress only an issue when it's in rapid concession?


Silicon has a thermal expansion of approx 4.7e-6/K, so a silicon
chip 5 mm square, will be 23.5 nanometer shorter for each degree
the temperature drops.

Epoxy is generally in the 40 to 50e-6/K range, so the 5mm epoxy the
silicon chip is bonded to on the other hand will be 235 nanometer
shorter per degree.

Drop the temperature 40C and the differential is 8.5 micrometer
and the resulting stress is quite sizeable.

It's much less of an issue when you raise the temperature, because
most plastics get softer with rising temperature.

Surface mounted devices suffer from the same problem relative to
the PCB (this is belived to be a major factor in the decreased
reliability in temperature-stress environments).

So if you want cold-water electronics, you want through-hole
components (because the leads can bend and absorb the stresses
and you want ceramic chips.

Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

More information about the Soekris-tech mailing list