Some of you will be old enough to remember this...back in the the early days of PC's the motherboard had a rechargeable NiCd battery to keep bios settings & time...and the damn things were soldered into the motherboard! A real pain to replace.
I happen to have a couple of those old machines that now need batteries replaced. The battery is a 3.6v NiCd .. anyone know if I could just use a holder with 3x AAA NiCd wired up in place of it? (so same voltage, but the AAA's would have a higher current(basically more storage capacity)...I know that extra wouldn't harm anything, but the question comes in as: would the motherboard even be able to charge them fully?)

First I guess you won't be able to find any NiCd that isn't on its last minutes till it fails anymore. Don't forget NiCd's, mainly in the common "household" formats (R03, R6,...) were stopped being made more than 15 years ago, so even if you find some, they will be pretty much garbage now.

But you may safely use NiMh, the larger cells made the thing even easier (provided the charging current is still way greater than the cell selfdischarge; so use some low self discharge types like Eneloop,...). The thing is, the cell is being constantly "overcharged", which is not a problem, when the rate is low enough to not lead to overheating and too high pressure. And larger cell means less pressure would be needed to recycle the gas back to water, so less stress when the cells are larger (be aware, it is the physical size, not the Ah rating what counts here). Better choose even the type with lower rated capacity: These have more empty space in the cylinder, so are able to transfer the gasses back to the recombination places (areas where the excess electrode area for that sits) with less pressure, so even less stress for mainly the seal.

But generally I would consider replacing the charging resistor with a diode (those days the supply of the backup memory came from the 5V via a diode and via a resistor from the battery; with supply the diode feeds the power and the resistor provided battery charge, without supply the resistor passed current from battery to the memory/RTC/...) and use Lithium cell instead. You will need to replace it every 10 years anyway (the NiCds tend to fail way sooner, the Eneloops may get to the 10Y but then will fail as well), plus as a collection item, it will be operated way less often than the system was originally designed for so the battery won't have much chances to get the charge it had consumed when off back into the battery.

Some system board have pins to connect replaceable battery pack.  You can find it in the motherboard manual if one exist.

NiCd is still used for emergency lighting fixtures, but probably not practical for this purpose 

NiCd is still used for emergency lighting fixtures, but probably not practical for this purpose 

I'm not sure if the true NiCd aren't just the NOS batteries...

Or more likely they are NiMH, but variant optimized for long term overcharging (so Ah capacity "just" comparable with NiCd's, but construction with higher margin for overcharge water recovery, to minimize the pressure so degradation during overcharging). I would not be surprised someone would market these as "NiCd" today...

But indeed, these are not usable, as batteries for emergency use don't need low selfdischarge characteristics, so I would see it as very likely these will have way higher selfdischarge leakage (as they are either being charged or feed the light by few watts till full discharge, so no benefit if they would last fully charged 10 years vs just a day, in both cases the light will discharge them within few hours after power gets interrupted anyway), too much for the CMOS backup (even for normal use, not talking about a collection item).


And for the connector: Some makers used a connector, but mostly only with the primary Li cells. Never seen one with the rechargeable batteries. But it is indeed not a bad idea to use one with the NiCd-accu -> Li-primary conversion mod. Because in most cases the original battery shape won't allow any easy fit of suitable battery holder. Otherwise when there is enough space, using CR2032 holder is the most practical in my eyes. Mainly because the CR2032 are the most common Li cells, so will be likely available fresh, not stored somewhere for 5 years before the sale (so with half of their lifetime already consumed)...

Edit: Now I get you meant an extension battery. With these I would first check how that interfaces with the soldered-on battery. Probably will need the bad battery to be removed, to prevent it from discharging the one connected via the connector. And really check, if the extension battery is supposed to be rechargeable (NiCd/NiMh) or a primary cell (Li). Eventual "charging" of the Li cell will kill it rather quickly...

I don't know, but the battery in our emergency light fitting specifically says it is NiCd. I don't know its age though, but it was a new fitting when we bought it a couple of years ago.

Thomas (Beta 5) would be able to tell you more on this subject as he collects emergency light fixtures, we only have this one for actual use above our front door.

Neither of those 2 PC's has a connector for an external battery pack. (I have seen such things on old computers before though)

I still see AAA NiCd's on eBay (whether they are actual new NiCd(they are from China-sellers afterall), or leftover old-stock, or "fakes" - NiMh labeled as NiCd, I'd have no way of knowing.)
Someone even had a NiMh replacement for the solder-in motherboard NiCd. And NiMh AAA could easily be obtained basically anywhere.

But isn't there a risk of NiMh going quite literally going 'boom' if overcharged?
One of the 2 computers it likely wouldn't be a problem, as that one only gets ran on occasional weekend days (if anything the 'risk' would probably be the batteries not getting charged enough.)... But the other one gets ran 24/7 from sometime in November til sometime in January. (so lets stretch it & say all of Nov/Dec/Jan since its possible it could be on that long... 3 months of continuous charging on batteries that shouldn't be overcharged) ((this 2nd PC is used to run some old software/controllers I made years ago - now used for the Christmas tree's lights inside. (outside lights are running off something quite a bit more modern .LOL. )))


I don't want to get into attempting to replace any resistors/diodes on the board, even if I could figure out which ones went to the charging circuit, it'd be a real pain to do .lol.

--
I have a few fluorescent fixtures with emergency ballasts..all state they have NiCd batteries inside. In the past I've opened ones that went bad to see what it'd take to replace the "non-replaceable battery" ... basically they were just a battery-pack made with C or D cells. Certainly could be replaced if you could find the parts at a reasonable cost & were up to a minimal bit of work

But isn't there a risk of NiMh going quite literally going 'boom' if overcharged?

Not at all, of course if the current is not excessive. There is always extra area of one electrode to serve as place where the H2 and O2 recombine back to the water.

Because the original cell size in the computers were some "button" cells, the AAA is for sure way bigger than that, so if the original cells were supposed to handle the permanent overcharging designed in there, the AAA size will handle the same overcharging current even way easier.

 
In fact overcharging happens and even needs to happen every time the batteries are charged:
First this is the only mechanism that takes care of balancing cell charge on a multicell battery.
And second: Virtually any charge control scheme used with these cells uses signs of the overcharging as triggers to terminate the charging.
And latest, the cells do not charge uniformly in their complete volume, so while some part may be overcharging, other is not charged yet. Overcharging it for some time is then the only way to ensure the whole cell gets fully charged.

So certain overcharge charge and certain (low) permanent overcharge current the cells can handle without issues, only the resulting pressure and temperatures aren't getting too high (the "too high" depends on if it is for just few minutes when reaching full charge at the moment the charging is cut off, or are getting lower if the overcharging is supposed to be permanent).

The misconception of "NiMH's very limited overcharge handling capability" came from the fact the NiMH's tend to have 5..6x the Ah capacity than NiCd's of the same size, while people are used to relate all charging currents to the Ah capacity of the cell. but in reality mainly the permanent overcharging current is limited by the heat it generates, so e.g. for an AA size it is limited to about 50mA.
That used to be 0.1C for 500mAh NiCd, but it appears as "only" 0.02C for a 2500mAh NiMH. That led to claims the NiMH survive only 1/5th of overcharge compare to NiCd and "therefore" are weaker.
Reality is, the AA size cylinder can handle only 50mA of overcharging, regardless if it is NiCd or NiMH, if the rated capacity is 450mAh (old NiCd), 900mAh (higjh capacity NiCd), 2500mAh (high capacity NiMh) or 600mAh of "solar light" NiMh (these use to be explicitely designed to handle the overcharging at elevated temperatures; but exhibit higher selfdischarge leakage so not suitable for the memory backup), the 50mA is just the limit for the AA size.

Thanks for the explanation Medved, might just try NiMh then.