If I were you I'd tried the following:
- Check the voltage of installed battery. Possible short circuit draws battery in no time (besides other negative effects like heating up some elements and areas on MB).
- Locate RTC chip and check if is cold. If it is noticeable warmer than other parts of MB it draws power from the battery while it shouldn't.
- Check if the power from the battery is delivered to RTC chip. Pinout of chips are easily (almost) can be found in datasheets.
- Locate capacitors near RTC chip and check if they are short-circuited (they can after many years).
- Look at motherboard carefully: you may notice missing or damaged elements or their leftovers (don't be confused by unsoldered footprints those are left empty by design).

I'll have a look at it over the weekend though I'm not exactly sure where to look for the RTC chip. Most of the markings are very washed out on the chips so I can only really identify the chipset, I/O and memory chips, most other chips appear to be various logic gates (I can read SN7406N just barely on one of them). The manual barely goes into any detail unfortunately, I'll try and see if there's anything what could be a RTC chip when I take the motherboard out.

[Edit] Oh yeah, I have tested the battery that has been installed in the motherboard for about a week and it still read 3.2V on the multimeter. There are also some diodes installed around the battery connectors which I assume make sure the motherboard does not try to recharge it.

Since the SiS 497 has an integrated real time clock, the motherboard is probably using that instead of a separate chip. So you may want to check the 496/497 datasheet (there's info about the RTC in page 73):

Try to follow the traces to see how it connects to the battery. If the clock works while the machine is on, it means that at least it's not completely dead and the issue is probably related to how it receives power in the OFF state (for example, broken traces or tiny SMD resistors that may have gotten knocked off the board). A picture of the area may help.

Regarding the floppy drive issues, have you tried swapping out the RAM? I've had similar problems with late 486 boards not allowing the floppy drive to work with EDO ram installed. But it works with FPM.

lazycrypt wrote on 2021-02-16, 14:40:

The motherboard has a 4-pin header for the BIOS battery (1&4 for external power, 2&3 to use the internal battery or 3&4 to clear the CMOS). I tried hooking up a CR2032 to the external power pins but the computer just became really unstable randomly hanging with black screens. The manual doesn't say what the voltage of the external battery should be but I didn't want to push it past 3V.

External batteries for PC's are 95% 4.5v, the other 5% are 6v. Have never had an issue using 4.5v as an external except a 286 which did not keep time unless at 6v (which when I finally found the manual did state 6v ext)....
Try 4.5v, 3 AAA or 3 AA being the typical....do not use rechargeable or lithium but real old fashioned standard batteries.

TheMobRules wrote on 2021-02-17, 01:04:

Since the SiS 497 has an integrated real time clock, the motherboard is probably using that instead of a separate chip. So you may want to check the 496/497 datasheet (there's info about the RTC in page 73):

SiS 496-497.pdf

Try to follow the traces to see how it connects to the battery. If the clock works while the machine is on, it means that at least it's not completely dead and the issue is probably related to how it receives power in the OFF state (for example, broken traces or tiny SMD resistors that may have gotten knocked off the board). A picture of the area may help.

Thanks, I'll check pin 37 (VDDRTC) on the 497 chip to see what's the voltage on it when the PC is off next time I get the chance.

Horun wrote on 2021-02-17, 02:27:

lazycrypt wrote on 2021-02-16, 14:40:

The motherboard has a 4-pin header for the BIOS battery (1&4 for external power, 2&3 to use the internal battery or 3&4 to clear the CMOS). I tried hooking up a CR2032 to the external power pins but the computer just became really unstable randomly hanging with black screens. The manual doesn't say what the voltage of the external battery should be but I didn't want to push it past 3V.

External batteries for PC's are 95% 4.5v, the other 5% are 6v. Have never had an issue using 4.5v as an external except a 286 which did not keep time unless at 6v (which when I finally found the manual did state 6v ext)....
Try 4.5v, 3 AAA or 3 AA being the typical....do not use rechargeable or lithium but real old fashioned standard batteries.

I'll get a battery holder and try putting 4.5V on the external battery connector and see what voltage SiS 497 is getting and if that fixes the problem.

Thermalwrong wrote on 2021-02-17, 01:30:

Regarding the floppy drive issues, have you tried swapping out the RAM? I've had similar problems with late 486 boards not allowing the floppy drive to work with EDO ram installed. But it works with FPM.

I am using FPM RAM. Since the failures only happen on the first cold boot I am suspecting it might be related to the RTC issue, I'll have a better idea after testing traces and using a 4.5V external power supply for CMOS.

TheMobRules wrote on 2021-02-17, 01:04:

Since the SiS 497 has an integrated real time clock, the motherboard is probably using that instead of a separate chip. So you may want to check the 496/497 datasheet (there's info about the RTC in page 73):

SiS 496-497.pdf

Try to follow the traces to see how it connects to the battery. If the clock works while the machine is on, it means that at least it's not completely dead and the issue is probably related to how it receives power in the OFF state (for example, broken traces or tiny SMD resistors that may have gotten knocked off the board). A picture of the area may help.

According to this datasheet RTC is powered through Pin37 (see pages 93-94).
Voltage between Pin37 and ground should be 2.8±0.4V when MB is powered off.

So after messing around with the computer all day (mostly cable management, I hate ribbon cables...) here's what I found out.

Plugging 3AA batteries into the external battery header (4.5V, multimeter read 4.92V) the RTC now ticks on even when the computer is turned off so I'm going to take a guess there is something wrong with the internal battery holder, maybe the very original battery leaked and damaged a trace or it got damaged while being handled at some point. Either way I will keep the external batteries connected since I really don't want to deal with soldering unless absolutely required.

There is still however the cold boot issue where a FDC failure will be reported and if F1 is pressed to bypass it, the computer will not detect serial and parallel ports (which would explain why there's no mouse in WinBIOS on the first boot). As mentioned before, after rebooting the FDC, serial and parallel are detected without issue. I ran a memory test if that could be the issue but no errors were found.

And finally, I also got a Promise Ultra33 just to test some stuff out and I can confirm the IDE controller on this motherboard is buggy. The BIOS has LBA and 32-bit access options allowing it to detect drives up to 8.4GB however just using 512MB or 1GB CF cards , and 2GB and 6.5GB hard drives all reads and writes are incredibly unreliable, partition tables will be reported as corrupted and creating a bootable partition is impossible. Normally this can be worked around using OnTrack or EZ-Drive but I thought I might use the Ultra33 but unfortunately while it would detect the drives just fine I would get random "Divide by zero interrupt" errors so I'll stick to the disk manager software for now. Don't know if the error is caused by the card (the BIOS version on it is 1.20) or the motherboard but at this point I just can't be bothered with swapping all the cables around again.

lazycrypt wrote on 2021-02-20, 19:22:

So after messing around with the computer all day (mostly cable management, I hate ribbon cables...) here's what I found out. […]
Show full quote

So after messing around with the computer all day (mostly cable management, I hate ribbon cables...) here's what I found out.

Plugging 3AA batteries into the external battery header (4.5V, multimeter read 4.92V) the RTC now ticks on even when the computer is turned off so I'm going to take a guess there is something wrong with the internal battery holder, maybe the very original battery leaked and damaged a trace or it got damaged while being handled at some point. Either way I will keep the external batteries connected since I really don't want to deal with soldering unless absolutely required.

There is still however the cold boot issue where a FDC failure will be reported and if F1 is pressed to bypass it, the computer will not detect serial and parallel ports (which would explain why there's no mouse in WinBIOS on the first boot). As mentioned before, after rebooting the FDC, serial and parallel are detected without issue. I ran a memory test if that could be the issue but no errors were found.

And finally, I also got a Promise Ultra33 just to test some stuff out and I can confirm the IDE controller on this motherboard is buggy. The BIOS has LBA and 32-bit access options allowing it to detect drives up to 8.4GB however just using 512MB or 1GB CF cards , and 2GB and 6.5GB hard drives all reads and writes are incredibly unreliable, partition tables will be reported as corrupted and creating a bootable partition is impossible. Normally this can be worked around using OnTrack or EZ-Drive but I thought I might use the Ultra33 but unfortunately while it would detect the drives just fine I would get random "Divide by zero interrupt" errors so I'll stick to the disk manager software for now. Don't know if the error is caused by the card (the BIOS version on it is 1.20) or the motherboard but at this point I just can't be bothered with swapping all the cables around again.

Good job ! Odd about the FDC, serial and parallel. Is there an option for "fast boot" in the motherboard BIOS ? If so try it disabled. One of my older boards when Fast Boot is enabled has quirks, probably a buggy bios version or chipset stuff is not getting proper timing when enabled....
Not sure about the Promise controller unless it has a conflict with motherboard resources or motherboard bios...