Have you replaced or modded the RTC (the black ODIN chip)? At this point, the battery is probably dead, and many motherboards that use those types of integrated RTCs are known to behave unpredictably when the battery is dead or close to dead as they are unable to properly persist the BIOS settings, even if you keep the power on.

TheMobRules wrote on 2026-03-30, 21:32:

Have you replaced or modded the RTC (the black ODIN chip)? At this point, the battery is probably dead, and many motherboards that use those types of integrated RTCs are known to behave unpredictably when the battery is dead or close to dead as they are unable to properly persist the BIOS settings, even if you keep the power on.

Thanks for the reply. It doesn't have the integrated battery/rtc, but a separate battery. Not sure if it's original or modded by a previous owner though, but I think it looks factory. The battery reads 3.1V.

mwiik wrote on 2026-03-30, 19:39:

I have this Socket 5 board: https://theretroweb.com/motherboards/s/epox-p … onix-ep-p54c-sp And this ISA I/O controller: https […]
Show full quote

I have this Socket 5 board: https://theretroweb.com/motherboards/s/epox-p … onix-ep-p54c-sp
And this ISA I/O controller: https://theretroweb.com/expansioncards/s/long … cs-6624g-rev-w5

I got the floppy working on a Pentium 3 board. The exact same combination of jumper settings, ribbon cable, floppy drive and floppy disk fails on the Socket 5 board. The drive is detected, no errors during POST, but after POST it sounds like three short bursts of a "rattling" noise, like if it's trying to read the disk but fails (bootable DOS disk inserted). Then it goes on to try to boot from IDE, which is the next issue:

BIOS detects CF card, but fails to boot from it (fresh DOS 6.22 install which is working on the P3 system on the integrated controller). No error messages, just a blank screen after POST.

On the P3 system, the BIOS seems to detect the IDE controller (conflict if I don't disable the onboard controller), but fails to detect drives.

I've tried:
All sorts of jumper settings.
Deoxit on the ISA slot and edge connector.
Scoured the BIOS for relevant settings.
Different CF cards and even a known working 2GB IDE hard drive.

Is this an ISA timing issue? Is there something wrong with the board (ISA data lines or something)? Or is this card just not compatible with the board?

The problem is that both the Socket 5 board and the contoller card are new to me, so none of them are known working except the floppy part of the controller. I don't have any other controller cards to test with yet. I'm comfortable with a soldering iron if that's needed, but I just don't know where to begin looking. I wouldn't mind buying another controller if it's a compatibility issue, but I don't want to waste my money if it's not.

If anyone has any insights, I'm all ears!

Have you tried with floppy only, without a HDD or CF?

2GB is possibly too big - you could be hitting a BIOS limit (this is a 1994 board, could be limited to 528MB lacking translation) or the ISA controller might not support LBA.

dionb wrote on 2026-03-31, 13:38:

Have you tried with floppy only, without a HDD or CF?

2GB is possibly too big - you could be hitting a BIOS limit (this is a 1994 board, could be limited to 528MB lacking translation) or the ISA controller might not support LBA.

Thank you for the reply. No difference with floppy only unfortunately. Even disabling the IDE part with the jumper setting doesn't help. The CF card I've tried with is 1GB, which is the smallest I have available at the moment. Both the CF card and the HDD are detected with the correct size in the BIOS though.

I noticed one thing now in the BIOS. The clock ticks very slowly. One second every 20 seconds on my stopwatch. Also, the year setting resets to 2096 on reboot, but the other date settings stick just fine. Could this be a clue? I don't have access to an oscilloscope to check frequencies of the crystals unfortunately. But it seems like the RTC is faulty somehow, but could that explain the floppy failure?

That OEC12C885 is the RTC. If the battery is outputting 3.1V, measure the input pin (20) on the 12C885. If that's getting 3.1V vs pin 16 (GND), there could be an issue with the CMOS, if it's getting less, there's a bad contact somewhere between the battery and the RTC.

The 2096 thing (usually it's 2094) is a Y2K bug in the BIOS.

dionb wrote on 2026-03-31, 16:04:

That OEC12C885 is the RTC. If the battery is outputting 3.1V, measure the input pin (20) on the 12C885. If that's getting 3.1V vs pin 16 (GND), there could be an issue with the CMOS, if it's getting less, there's a bad contact somewhere between the battery and the RTC.

3.11V on the spot. What could the issue be with the CMOS?

jakethompson1 wrote on 2026-03-31, 16:54:

The 2096 thing (usually it's 2094) is a Y2K bug in the BIOS.

Thanks, 2094 it is, I misremembered. Then I guess that's nothing to worry about.

mwiik wrote on 2026-03-31, 14:55:

I noticed one thing now in the BIOS. The clock ticks very slowly. One second every 20 seconds on my stopwatch.

DOS does check if the RTC is ticking on startup, so that could explain the hanging...
Also check pin 21, it should (ideally) be pulled high or floating.

jmarsh wrote on 2026-03-31, 18:21:

DOS does check if the RTC is ticking on startup, so that could explain the hanging...

That's very interesting! Could the crystal oscillator be bad? I know I have a spare one somewhere since I bought two when I replaced it on another machine, but they are small and my mess is big...

jmarsh wrote on 2026-03-31, 18:21:

Also check pin 21, it should (ideally) be pulled high or floating.

I get 4.7V on pin 21.

Ok, progress made! I changed the crystal oscillator, which did not help at the time, but later the clock just started working properly again while troubleshooting other issues. So strike that one off the list.

Also, disabling the external cache actually made it boot to DOS from the CF card. But the floppy still doesn't work on the Socket 5 board.

So there are two issues remaining. Cache (which is easy to work around, will fix later), and floppy drive not working (works on another machine with the same controller).

Any ideas on what to test next regarding the floppy drive would be appreciated!

mwiik wrote on 2026-03-31, 18:29:

I get 4.7V on pin 21.

At first I thought jmarsh meant IDE pin 21, but that would make sense with pin 28.

CF card IDE converters love to pull that high, that will screw with whole computer. I would check if IDE pin 28 isnt somehow permanently wired to CF card/ground or just cut/isolate that pin. Re: "Fixed" 386sx motherboard works but not with 16-bit VGA card

rasz_pl wrote on 2026-04-30, 21:31:

At first I thought jmarsh meant IDE pin 21, but that would make sense with pin 28.

I meant pin 21 on the RTC, some boards have it grounded because they assume a DS12887 (without "A" suffix) will be used, which doesn't have pin 21.

Update:

I replaced the cache SRAM chips, and now it boots with cache enabled as well. The floppy started working when I put a jumper across the turbo pins. Go figure.

Thanks for all the tips and thoughts! This issue has been resolved.

Hmmm, I think there might be something a bit hinky with your clock and clockgen still. Possible issues will flare when more cards added.
Can't see anything obvious from the datasheet
https://theretroweb.com/chip/documentation/sc … b0905274153.pdf
But I'd give a good clean to the clock setting pins and make sure they have tight jumpers on them. Am thinking along the lines that 14.318 Mhz drive strength might be low still.

Nothing I can put my finger on yet, just a kind of "been to a few rodeos and I don't like the look in that bull's eye."

BitWrangler wrote on 2026-06-02, 14:02:

Hmmm, I think there might be something a bit hinky with your clock and clockgen still. Possible issues will flare when more card […]
Show full quote

Hmmm, I think there might be something a bit hinky with your clock and clockgen still. Possible issues will flare when more cards added.
Can't see anything obvious from the datasheet
https://theretroweb.com/chip/documentation/sc … b0905274153.pdf
But I'd give a good clean to the clock setting pins and make sure they have tight jumpers on them. Am thinking along the lines that 14.318 Mhz drive strength might be low still.

Nothing I can put my finger on yet, just a kind of "been to a few rodeos and I don't like the look in that bull's eye."

I ran the system board tests in Checkit Pro, and it failed the RTC tests. I have ordered a replacement RTC chip since replacing the crystal earlier obviously didn't help.

Do you think this could be a related issue? I agree that there seems to be something hinky, since I'd expect the floppy to work regardless of turbo status.

Yeah I'd wait to try that before anything drastic, it's flakiness in the CMOS, battery and RTC that can cause some really weird problems.