FWIW, using AAA NiCds would match the voltage of the original cell. It would be especially good if the external header supports charging.

You know I've honestly thought about disabling the recharging circuit on old motherboards all together, there can't be that many components to unsolder from the board right? I remember someone doing it on an old Neo-Geo arcade board and adding a CR2032 as a replacement.

I also thought if one were to disable the recharging circuit and or put a diode on the positive side of the battery terminal, would a Macintosh PRAM 1/2 AA Battery and holder be a good replacement? I haven't measured the holder just yet and to be honest I'm still leaning toward an external battery pack to keep it as far away from the board as possible.

I disable the "charging circuit" on all my 386/486 boards which have one of these damn barrell batteries and install a CR2032 coin cell holder. You sometimes must used a double-stacked CR2032 holder to hold two batteries in series. The process usually involves removing a diode which connects to the +5 V on the PSU connector. There was already another post on this process, so I won't get into it. You can trace out the circuit on your motherboard to determine which diode to remove. There are usually 3 or so diodes in that area and some go to the external battery connector.

If you don't want to hassle sourcing the appropriate diode to remove, or cut the lead from, you can do as has already been suggested and add a diode in series to your CR2032 coin cell to prevent current from charging the battery. Of course, this will add a voltage drop to the battery though (~0.7 V for a "standard" diode, or ~0.3 V for a schottky diode). If you are using a double stacked CR2032, I suppose the voltage drop won't matter much.

Anyway, I found it really interesting that you are using a UMC CPU - U5SX. I've never seen one of these soldered onto a motherboard like this in the QFP package. How do you like the CPU? Supposedly, it is clock-for-clock faster than an Intel 486SX.

gdjacobs wrote:

FWIW, using AAA NiCds would match the voltage of the original cell. It would be especially good if the external header supports charging.

Would using non rechargable alkaline AAA batteries cause issues if the ext battery header supports recharging?

Should I use rechargeable AAA just to be safe? Which rechargabke battery type would be most suitable to the task? More I read more I get confused..

feipoa wrote:

I disable the "charging circuit" on all my 386/486 boards which have one of these damn barrell batteries and install a CR2032 coin cell holder. You sometimes must used a double-stacked CR2032 holder to hold two batteries in series. The process usually involves removing a diode which connects to the +5 V on the PSU connector. There was already another post on this process, so I won't get into it. You can trace out the circuit on your motherboard to determine which diode to remove. There are usually 3 or so diodes in that area and some go to the external battery connector.

If you don't want to hassle sourcing the appropriate diode to remove, or cut the lead from, you can do as has already been suggested and add a diode in series to your CR2032 coin cell to prevent current from charging the battery. Of course, this will add a voltage drop to the battery though (~0.7 V for a "standard" diode, or ~0.3 V for a schottky diode). If you are using a double stacked CR2032, I suppose the voltage drop won't matter much.

Anyway, I found it really interesting that you are using a UMC CPU - U5SX. I've never seen one of these soldered onto a motherboard like this in the QFP package. How do you like the CPU? Supposedly, it is clock-for-clock faster than an Intel 486SX.

Could you link the post feipoa, I searched and there was a ton of results for battery replacement go figure.

I actually did a little bit of testing with my board and turns out the CMOS Jumper is what determines if the 5 volt change gets to the + battery terminal the downside is, the battery terminal is not connected to anything without the jumper. I found 4 diodes on my board and I'm honestly curious which one or if all of them are the recharge circuit. Pin 3 on the "dischange" CMOS jumper is connected to the positive terminal of the battery. Pin 2 is where the 5 volts from the recharge circuit is coming from.

In theory from the looks of it pulling diode number 2 and attaching the + battery terminal to the trace leading toward the RTC chip would bypass the circuit all together and if a CMOS reset is needed just pull the battery, would that be correct? Or do I need some of those diodes, capacitors and resistor in circuit?

Three NiCd AAA rechargeables would be completely compatible with an identical discharge curve. NiMH should work as well so long as the CMOS isn't very voltage sensitive (it shouldn't be). They might not charge correctly depending on the charging voltage, but the chipset as a whole should be 5V tolerant. Both are off the shelf items, although NiMH batteries are more popular and more available due to higher energy density and closer characteristic matches to alkalines. Energizer, Panasonic, Duracell, etc. all have their own lines with many house brands available as well.

As for non rechargeables, the charging circuit might not push enough voltage (>4.5V) to create a charging current, but if it does, the resulting heat could cause the batteries to fail. I doubt it would happen as the circuit is likely impedance limited, but I wouldn't personally take the chance. If you opt for alkalines, please install a diode inline (you can safely shrink wrap it) if charging isn't disabled otherwise.

Also, good luck finding jumper info for this board. Google says it may be a model MB457 (manufacturer unknown). I didn't see anything in TH99 during my brief look.

zakurowrath wrote:

Could you link the post feipoa, I searched and there was a ton of results for battery replacement go figure.

There isn't a vogons thread that I know of which goes through a cookbook recipe on how to determine what to cut; they are mostly people just saying they cut this diode lead to cut the charging lead to the battery. Basically, you are disconnecting the +5V line from reaching the battery.

Normally, you want to trace out what the diodes are doing. Here is an example I did on a GA-5486AL motherboard. You'll notice how I desoldered one of the leads for diode D10. I also leave JP3 jumpered for "internal battery".

Usually it is easier to just use the external battery connector for the CR2032, but sometimes the manual calls for upwards of a 6 V battery, depending on how the circuit is configured. I find that most of the time, I can get away with a 3V CR2032 on the barrel battery solder points if I disconnect the charging lead. Most 386 boards I've tested seem to require two CR2032 batteries in series though.

Couldn't find NiCd rechargables available fast, so settled for NiMh rechargable AAAs, thanks.

elod wrote:

I would not consider soldering any NiCd on boards. Even coins can leak as we saw with the Roland SCs. If it's got an external connector pull a long wire and move it away from the board. Someone will thank you in the next 20 years 😀

TBH, if you are in the EU and your favourite electronics shop is still selling NiCd (not NiMH, those are fine) batteries, then they are in breach of EU regulations, since NiCd's have been banned for at least a decade...

How about phone battery packs like these? They look very suitable for fixing to the case and connecting to the external header:

Yup, just have to slap on a new plug.

LiIon should not be played around with in a circuit that potentially charges it. It needs a proper monitoring circuit.
Are NiMh 100% leakproof? I think not. No battery is.

Back in the day I've seen 3 solutions used for external batteries: lithium (not LiIon!), 3xAA and the classic lantern battery (3r12). That's what I would use and the cheapest one is the 3xAA battery holder. I would also remove any battery before placing mainboards into storage (so soldering is out of question).

Please disregard and do what you want.
(this is a reaction to a post that disappeared in the meantime)...

appiah4 wrote:

How about phone battery packs like these? They look very suitable for fixing to the case and connecting to the external header: […]
Show full quote

How about phone battery packs like these? They look very suitable for fixing to the case and connecting to the external header:

As good as that looks, I agree with everyone, a charging circuit met for another type of battery chemistry could cause catastrophic failure with a newer type of battery. I'd honestly just go the external battery pack with AA's or disable the recharging circuit all together and put in a double CR2032 holder.

feipoa wrote:

There isn't a vogons thread that I know of which goes through a cookbook recipe on how to determine what to cut; they are mostly […]
Show full quote

zakurowrath wrote:

Could you link the post feipoa, I searched and there was a ton of results for battery replacement go figure.

There isn't a vogons thread that I know of which goes through a cookbook recipe on how to determine what to cut; they are mostly people just saying they cut this diode lead to cut the charging lead to the battery. Basically, you are disconnecting the +5V line from reaching the battery.

Normally, you want to trace out what the diodes are doing. Here is an example I did on a GA-5486AL motherboard. You'll notice how I desoldered one of the leads for diode D10. I also leave JP3 jumpered for "internal battery".

Usually it is easier to just use the external battery connector for the CR2032, but sometimes the manual calls for upwards of a 6 V battery, depending on how the circuit is configured. I find that most of the time, I can get away with a 3V CR2032 on the barrel battery solder points if I disconnect the charging lead. Most 386 boards I've tested seem to require two CR2032 batteries in series though.

GA-5486AL_Battery_Replacement_Circuit.jpg

GA-5486AL_Battery_Replacement.jpg

Very nice, looks like I'll have to do that with my board, I'm no engineer but I have a basic understanding of electronic circuits, really just need a good multi-meter with an actual beeper so I can test continuity in the circuits and components, also a good soldering iron, I can always double check with people here and a few people I know if I do decide to modify the circuit.

zakurowrath wrote:

As good as that looks, I agree with everyone, a charging circuit met for another type of battery chemistry could cause catastrophic failure with a newer type of battery. I'd honestly just go the external battery pack with AA's or disable the recharging circuit all together and put in a double CR2032 holder.

This is especially the case for Li-Ion. NiMH *should* be okay, but for safeties sake is best located at the bottom of the case, away from anything disolvable.

I'd go with a 3 cell pack of Eneloop AAAs, and in fact, that's what I plan on doing with my 486 board I just got with this 3.6V barrel pack. NiCD and NiMH cells have similar voltage and charging profiles, so that shouldn't be an issue. Eneloops have very low internal resistance for NiMH cells, closer to that of NiCDs, and the AAs can even supply up to 10 amps sustained without too much voltage drop, though I wouldn't recommend it as an every day workload kind of a deal. They're also very low self discharge and will hold a charge for several years. The AAAs have a capacity of about 800mAh and the AAs have a capacity of around 2000mAh, so significantly higher than most NiCDs.

If you can, maybe check the charging voltage of the board. Since Eneloops hold a charge for so long, it's not the best to have them continuously trickle charged, as it can slowly degrade them over time. NiCDs can handle this a bit better, provided the current is low enough of course.

The Eneloops can vent off some hydrogen if extremely overworked and have a built in vent for this, though I've never had it happen to my cells that I know of. They shouldn't leak. In the 6+ years I've been using them, I've never had them leak, and I still have all my original cells and more, though I had heard a report from someone else that they had one that leaked. This was several years ago, and I'm not sure of the circumstance nor entire validity of the report. I used to be pretty active over at CandlePowerForums where Eneloops have been discussing in length and were (and likely still are) a favorite.

Charging to 1.48V per cell (4.44V for a three cell batter pack) is fine. Sustained trickle charge over that (given the NiCD cells used here in the circuit) might slowly damage the Eneloop cells over time, but still likely won't cause any other issues to your hardware. Full charge at rest they'll slowly level out to around 1.35V. Once they hit 1.2V they're starting to get low, and I would avoid discharging them down below 1V per cell to avoid the potential risks of reverse charging. This would likely take many years to happen for this kind of a use.

For sustained trickle charging while in use via the motherboard, NiCD cells might be the better option, but you have the additional risk of leaking with those cells. I certainly wouldn't use Alkaline cells myself (I use Eneloops now instead anyway), and I would try and avoid NiCD cells in general as well.

If you're able to disable the charging circuitry or bypass it, that's a good option too, unless you're specifically wanting to go with NiCDs.

Sorry if this post dragged on too long. Just giving some food for thought.

Cheers.

Eneloops are nice. But... with the trickle charge and a typical usage pattern of a retro PC I'm not sure AA/AAA will ever charge.

I'm not sure what you mean. They come pre-charged. I was just thinking the motherboard would more or less sustain them, if not slowly charge them, since they have such a low self discharge rate. It depends on the voltage the motherboard charges at, and I guess it also depends what the drain rate of the motherboard is. Given their capacity in comparison to these barrel packs, they should last a very long time either way.

Ok, sure, they might charge. But for about 5-10x times less money you can get some very good quality long life AA batteries that will last around 5-10 years.

IMHO, if you are going to solder in a new rechargeable NiCd or NiMH battery, it is best not to do it directly on the board as the old one but with a cable long enough to put it somewhere away form the board itself. That way when it inevitably leaks in a decade or two, it won't take the motherboard with it.

appiah4 wrote:

How about phone battery packs like these? They look very suitable for fixing to the case and connecting to the external header:

Yep, that's the way to go if you don't want to resolder another barrel-type battery. Those are correct nicd replacements, can be charged, and have the correct voltage. The only problem is that in some cases the charging voltage is not enough for them but this is a rare case and they still qualify as one of the best alternatives.