Motherboard backup battery usage

[kladner]

Originally Posted by Xyzzy

Another interesting data point WRT CMOS/RTC batteries: We once worked on a pretty large (boat anchor) Compaq or Dell server that had 2×AA NiCd cells in a little shrink-wrapped holder wired up as the CMOS/RTC battery, and it would charge continuously, at a very very low maH rate. Unfortunately, over (a long) time, the NiCd cells lost their ability to hold a charge and failed.

Since we had piles of Sanyo Sub-C NiCd cells for our remote control cars we just slapped two together with shrink-wrap and some wire and used double-sided sticky tape to secure it in the chassis.

Up until that point the organization was going to scrap the server (!) because nobody could figure out why the clock would slew continuously. (And there was no facility in the OS to compensate for the slew.) The BIOS defaults were enough for it to boot. (It was a real simple BIOS and we think it had to be accessed via a serial port.)

So we were a hero for a day.

NiCd is notorious for its "memory effect". It works best if fully depleted and promptly recharged. Recharging a nearly full cell will tend to give it a shallow cycle. One of my brothers would overcome this problem by giving the NiCd battery a jolt at higher than normal voltage (read: charger plugged into a variac transformer).

NiMH are not supposed to be as vulnerable to memory effect, and Lithium Ion less still. Older NiMH do have a problem of loosing charge over time. Newer ("Precharged") cells hold most of their charge for a year.

[LaurV]

Everything sounds reasonable except the part of plugging the adapters to a higher input voltage. The older adapters (read: 10 years ago or more) had separation transformer and capacitor drop (they were heavy, etc) and they gave out an output voltage proportional with the input voltage. This is not true anymore, for years. The newer switching power supplies will give you a constant voltage no matter what to plug it into, they even work between 100V and 250V, etc. Giving them higher voltage will only damage the switching ICs inside. Especially for charger adapters, the clever one will give you a constant current according with the battery type and depletion level, again, no matter what voltage you put in. I think what your brother was doing was to use a separate variable power supply (directly, with no adapter between) to give a small current shock to a depleted accumulator to clear its "memory" effect. This is quite common, but don't do this at home unless you know what you are doing, the accu can easily explode in your face, that is a small bomb if you expose it to excessive current or heat.

[xilman]

Originally Posted by LaurV

This is quite common, but don't do this at home unless you know what you are doing, the accu can easily explode in your face, that is a small bomb if you expose it to excessive current or heat.

A colleague, who shall remain nameless, once gave a demo to the head of MSFT Research in an office two doors down from mine. Unfortunately the battery had been inserted backwards. It exploded, bounced off first a wall, then the ceiling and then another wall. It finally came to rest on the floor where it burnt a hole in the carpet and filled the office with a foul smell.

An impressive demonstration, but not quite what was intended.