The damage certainly looks repairable - mainly because the motherboard designer was smart enough not to put thin inaccessible stuff around the battery.

First thing would be to clean up the corrosive mess still there. If you can still see blue-green, it's still eating away at your board. The opinions are divided on whether to use an acid (vinegar) to neutralize the alkali, or whether just to use distilled water. I'm in the former camp, I don't trust pure water to get it away, so I use cleaning vinegar, more specifically, I use toilet/kitchen paper soaked in the stuff and stick that on the affected parts for a few hours. Usually when I subsequently wash it away it takes all the mess with it.

Then it's a matter of determining the damage. You should measure with a multimeter (the kind with buzzer to show continuity are the best, but anything that can measure resistance is good enough) to be sure. Things look bad around the power connector, but those are BIG joins and probably only superficially affected. With a bit of luck only that one long, thin trace on the back is actually damaged to the point of needing repairs. If so, just get some thin wire and solder it to either end of the trace. Use sticky tape to keep it in place, both while soldering and later when in use. You're lucky nothing seems to have gotten into the ISA slots or under the SIMMs, or worse, under any big ICs.

Depending on how the rest of the solder joins look IRL it might be interesting to re-flow the big pins on the power and keyboard connectors, and possibly also the legs of those tantalum caps.

dionb wrote on 2021-07-27, 07:16:

The damage certainly looks repairable - mainly because the motherboard designer was smart enough not to put thin inaccessible st […]
Show full quote

The damage certainly looks repairable - mainly because the motherboard designer was smart enough not to put thin inaccessible stuff around the battery.

First thing would be to clean up the corrosive mess still there. If you can still see blue-green, it's still eating away at your board. The opinions are divided on whether to use an acid (vinegar) to neutralize the alkali, or whether just to use distilled water. I'm in the former camp, I don't trust pure water to get it away, so I use cleaning vinegar, more specifically, I use toilet/kitchen paper soaked in the stuff and stick that on the affected parts for a few hours. Usually when I subsequently wash it away it takes all the mess with it.

Then it's a matter of determining the damage. You should measure with a multimeter (the kind with buzzer to show continuity are the best, but anything that can measure resistance is good enough) to be sure. Things look bad around the power connector, but those are BIG joins and probably only superficially affected. With a bit of luck only that one long, thin trace on the back is actually damaged to the point of needing repairs. If so, just get some thin wire and solder it to either end of the trace. Use sticky tape to keep it in place, both while soldering and later when in use. You're lucky nothing seems to have gotten into the ISA slots or under the SIMMs, or worse, under any big ICs.

Depending on how the rest of the solder joins look IRL it might be interesting to re-flow the big pins on the power and keyboard connectors, and possibly also the legs of those tantalum caps.

Mate you are an absolute legend. Thank you so much.

First, I cleaned the board using vinegar, rinsed it in water and dried it thoroughly. Once dry, I used some isopropol alcohol to hopefully neutralise any remaining vinegar (not sure if there is any scientific basis to that but whatever).
I got the old multimeter out, set it to buzz mode and started testing. Only one circuit didn't have continuity. The one for the middle pin on the keyboard. I tried removing the AT plug to see if I could get under there and check the damage and confirm which pin the broken trace ran to, but that was rather difficult so I ended up leaving it and worked out a less stupid way to identify the correct pin.

I tried running a bypass wire underneath from the keyboard pin to the hole corresponding to where the circuit up top is supposed to finish. Unfortunately I couldn't get continuity between the bottom hole and top hole. I might be using the wrong sort of solder... and I had great difficulty getting it to stick IDK (my soldering skills are very limited - I get by for automotive stuff and don't usually start fires but that is about it) . I ended up passing a thickish strand of copper through the hole, running it along each trace and putting a bit of solder top and bottom. I got continuity from the middle pin.

Plugged it in, fired it up, got obligatory bios error due to missing battery - but the keyboard worked.
Next step, memtest.

Next step after that I need some further advice on. NiCad battery is 3.6v. Guessing a 3v CR2032 aint gonna cut it? What is the best way to power this bad boy? Any other advice or suggestions would be most welcome (please don't berate my lack of soldering skills too much 😜)

Thanks

It just isn't stable. Hangs / reboots after a few mins of memtest. Could it be because it has no battery?

Good news.... after some further cleaning with proper demineralised water, a good dry out and a small application of some silver paste stuff over some damaged areas, she is alive and kicking. Memtest stable. New fancy Lithium Ion battery soldered in place.

I went through my stash of 30pin RAM and found some 4MB 60ns SIMMS. Those were apparently rare is back in the day...at least I remember having trouble getting hold of them 20 odd years ago. Anyway, it booted up with 16MB RAM and is currently memtesting again this time around with as much BIOS memory optimisation as I could do. Sadly I don't appear to have 4x 1MB sticks of 60ns RAM so it is either stay with super fast 16MB or go with 20MB at 70ns and relax the timings. What to do... what to do... I know! BENCHMARK! hehe.

Assuming all good, I am going to spray the bits I soldered / pasted with some lacquer spray the electronics shop sold me, put it all in the case and try to find a VLB video card to replace the 512k Trident powering her at the moment. Add in a NIC and a soundblaster card then fire up some Doom and Quake!

Also got it booting with the original hard drives. A 120MB and an 80MB. Heaps of games and stuff on there too. I think I will replace the drives with something closer to the 2GB mark (if I can find one) and use the smaller drives in a 386. Maybe use a bigger drive and dual boot Linux.

486 was my favourite era. The pinnacle of computing before it started opening up to simple people (hehehe). This is gonna be epic.

Ok dumb question time (yeah I know, my posts are full of those). I read in another post that you can use a CR2032 on these things despite being a different voltage, but you need to stop the board trying to charge it.
I used a Fanso ER14250BH 3.6v on the advice of the electronics shop. Do I need to do anything to stop the battery blowing up or getting damaged long term?

My research indicates that this is a Lithium Thionyl Chloride battery which is not the same as lithium Ion and likely not rechargeable. What should I do? Should I find a proper rechargeable type battery?

The battery was also slightly bigger - the board seemed to have two different spots for the negative terminal so I used the other one to make the battery fit nicer. It held the CMOS data overnight so all good there.

Thanks

You're right that it's not safe to put charge into it from what I can see, it doesn't appear to be rechargeable. Without charging, it should last for years, that's a pretty big lithium cell for a backup battery.

If you find the diode that feeds into the positive side of the battery's positive terminal (usually has a resistor in front of it), you can cut one leg of the diode or resistor to stop the charge from going into the battery.

That looks like D1 and R2 from the pictures you've uploaded 😀 Since they're both SMD components, probably the safest & easiest thing to do is to cut the trace going from R2 to the battery - kinda looks like the battery corrosion may have already done that for you.

Thermalwrong wrote on 2021-08-05, 12:48:

You're right that it's not safe to put charge into it from what I can see, it doesn't appear to be rechargeable. Without charging, it should last for years, that's a pretty big lithium cell for a backup battery.

If you find the diode that feeds into the positive side of the battery's positive terminal (usually has a resistor in front of it), you can cut one leg of the diode or resistor to stop the charge from going into the battery.

That looks like D1 and R2 from the pictures you've uploaded 😀 Since they're both SMD components, probably the safest & easiest thing to do is to cut the trace going from R2 to the battery - kinda looks like the battery corrosion may have already done that for you.

Thanks for the advice. I repaired that damage. How do I know which part (D1 or R2) charges the battery vs which part allows the battery charge to flow to the CMOS? I have taken a good look at the traces and can't seem to work it out. The last thing I want to do is defeat the purpose of having a battery in the first place.

It looks like I should have done my own research instead of trusting a random dude at the electronics shop. From what I can tell, an LR2032 seems to be the go? https://www.thebatterysupplier.com/products/l … lls-lr2032.html - 3.6v 50 mAh but the old ones are 60 mAh. Thoughts?

Cleaning is only one part of the process - as above said, you still have to assess and mitigate the damage caused.

The board from pic #1 has one of the traces going from the keyboard port completely green - good work on fixing that. From my also limited experience, standard Sn60Pb40 solder is just fine - main thing is to use flux. Bear in mind, too, that oxidized solder is VERY hard to solder to - it's better to remove it completely and start anew.

To answer your questions about the battery:

1) If you try to charge a non-rechargeable battery, especially in a circuit like on a motherboard where there is a steady inflow of current... bad things are going to happen.
2) I would not trust any barrel battery regardless of it's type. Biased and based - these caused me too much headaches in the past.
3) You indeed can desolder/cut a diode from the charging circuit and plug in a CR2032. Look up Necroware's YouTube video on this - very informative!
4) Diode is one thing, but you can also consider buying a battery caddy and plugging it to the external battery header (assuming the board has one, which it very likely has - first pic, top right). Back in the day people would run an AAA battery pack and do just fine, however I've seen people using CR2032 caddies without issue.

Ok thanks for confirming that I can't use it in the current configuration. Real glad I checked now.
I watched the video by Necroware. Very informative for sure but the board he was using was a bit less complex. I've also read that an LR2032 whilst being rechargable isn't designed for continuous current either so that sounds like a no go.

As much as I hate to say it, I think the best solution given my lack of sufficient skill, is to order a replacement that matches the original spec. Something like this: https://www.ebay.com.au/itm/331034570308 but I would rather order from a trusted manufacturer as I want something that will last at least as long as the original, hopefully longer. Any thoughts?

J20 could be an external battery header, but no way to know for sure. I don't suppose anyone could help identify what make / model this board is? A manual would be amazing.
https://www.dropbox.com/s/7stcop86kv2o1o9/202 … 003012.jpg?dl=0

That's most definitely a Soyo board - they always had those "Energy Star" stickers on them. Q.C sticker with "SY" sticker confirms that thesis as a SOYO board that I have also contains such sticker.

I'd be quick to say it's a SOYO SY-025G2 or similar 025 series board - but I'd definitely try to look for other markings, as well as under the sticker (if you're brave enough to ruin vintage stickers!) to confirm that 100%.
Here is a motherboard diagram for a SOYO board strikingly close in appearance to yours: https://www.stason.org/TULARC/pc/motherboards … -MAINBOARD.html

J20 is marked as an external battery header. Usually those headers are 4-pin, with one pin missing to aid in orientation (if my eyes aren't deceiving me yours does not have one pin), close to the battery charging circuit.

If I may kindly suggest, please stay away from replacing the battery with another one. It will leak its guts sooner or later. If you are 100% certain you will be checking on it periodically in, let's say, 10 years or so which is the time these usually start going bad, then OK, go ahead. If I were you I'd grab a battery caddy for 2xAA (reference image: https://www.cosplaycraft.pl/wp-content/upload … Baterie-AA.jpeg) or 2xCR2032 (reference image: https://www.markland.pl/images/produkty_alleg … _7062281986.jpg).

Add a pin connector from an unused speaker/CD drive cable/whatever, and you're golden.

One of the things on my queue is testing out this thing here connected to an external battery header: https://th.bing.com/th/id/R.9b6c496c070 ... ImgRaw&r=0

It only holds one CR2032, but has a power off/reset CMOS switch built in. Some folks on a Polish retro FB group reported that it works quite well actually! And it's a great add-on to dead RTCs. I'm hoping to start researching it myself soon. It's in the same price range as a standard plastic caddy, and you have to solder in two wires to it.

Suggested reading:

http://pc-restorer.com/replacing-cmos-batteries-in-old-pcs/
https://vswitchzero.com/2018/02/20/the-486-re … oration-part-2/

Fantastic - thank you so, so much for such an informative post.

I will identify the correct polarity for the external battery as described then rig up a connector using the existing lithium battery (as I already have it so why waste it).
I'll try to do it properly so I don't end up in a situation where I can mistakenly reverse the polarity changing the battery in future. Should be good for a long, long time that way.

Thanks again.

Wire up a breadboard for the existing battery so that you can put some tape on the bottom side and safely put it somewhere in the case.

Check this: http://minuszerodegrees.net/battery/cmos/vint … d_batteries.htm

It's also an informative source. You should have little problem figuring out the correct polarities on the connector.

Thanks. Yeah I worked out the polarity no worries (the pin with continuity to a ground connection is obviously negative) and used an old CDROM cable, blocking the appropriate hole. I used the 3D printer to print a box to hold the battery so it can't touch anything else.

Question: How long might you reasonably expect the CMOS to hold its values without a battery hooked up. I turned it on with no battery and saved settings to CMOS. I want to wait for it to clear so I can hook up the battery and verify all good.
One would assume that without a battery, the CMOS would lose its saved data almost immediately but that assumption is clearly incorrect as half an hour later I plugged it in and it still had the saved settings.

Thanks

kevmif wrote on 2021-08-06, 09:54:

Thanks. Yeah I worked out the polarity no worries (the pin with continuity to a ground connection is obviously negative) and use […]
Show full quote

Thanks. Yeah I worked out the polarity no worries (the pin with continuity to a ground connection is obviously negative) and used an old CDROM cable, blocking the appropriate hole. I used the 3D printer to print a box to hold the battery so it can't touch anything else.

Question: How long might you reasonably expect the CMOS to hold its values without a battery hooked up. I turned it on with no battery and saved settings to CMOS. I want to wait for it to clear so I can hook up the battery and verify all good.
One would assume that without a battery, the CMOS would lose its saved data almost immediately but that assumption is clearly incorrect as half an hour later I plugged it in and it still had the saved settings.

Thanks

My experience is that, depending on the board, CMOS RAM looses contents after 3 minutes to 6 hours. Some boards have a capacitor on the battery power line that powers the RTC chip for some time even if the battery is disconnected. The main current consumer in the RTC chip is the ticking clock. If it were plain RAM, you could expect to keep the data even for days without a battery connected.

mkarcher wrote on 2021-08-06, 09:58:

My experience is that, depending on the board, CMOS RAM looses contents after 3 minutes to 6 hours.

Thanks. In this case it lasted some time between last post and now.
Battery plugged in. We'll see how it looks in the morning. Good thing I took measures to ensure the polarity couldn't be accidentally reversed as I am super inebriated and would have plugged it in the wrong way if not for my earlier efforts to ensure that was not physically possible.

Battery now working as expected. System seems stable. Time to move on to optimisation. Running norton sysinfo it shows CPU performance to be about 10% below par for a DX2-66. Disappointing as the CPU is a DX2ODPR66.

Are these Soyo boards with SIS chipsets known for being under performers or have any other notable characteristics over other boards from that time?

I've set memory setting to fastest and all timings to the lowest possible values. I tried various bus clock frequencies but it seemed to make no difference so I left it at the default of 7.15.
Board jumper settings appear correct as best as I can tell. I'm guessing I probably don't want to change the bus speed from 33mhz for this CPU?
Turbo is always on via jumper (removed the jumper and it dropped to about 1/4 of previous speed).

Turns out there is a bios setting called auto config which had to be turned off. I assume it was overriding my aggressive memory settings. Norton sysinfo now has it slightly exceeding the dx2-66 baseline.
Keen to know what people think about the bus clock frequency setting and if there is anything else I can look for.
Thanks