dennisE wrote on 2025-07-13, 12:05:

How do I check the voltage at pin 24 of the HD146818? I understand that I connect pin 24 on one side of the tester, and the other side (battery or power supply)?

You need to measure between pin 24 and GND (ground plane). For GND you can pick pin 12 (opposite corner of 24) or perhaps one of the black wires in any spare PSU connectors. Using the pins directly will give you more accurate result but with decent PSU wires even using a plug far away from the chip the difference shouldn't be more than 0.1V anyway.

dennisE wrote on 2025-07-13, 12:05:

If I put in an external battery, for example, 3.6V, will it be enough to work?

It should be enough, but some older mobos required 6V external pack, and had more than 1 diode that caused voltage drop. These are rare though. The 3xAA pack will give you 4.5V which is a perfect value, high enough to ensure good chip power supply, low enough to not be drained if the mobo has just 1 diode (unlike 6V pack would be). Cheap, comes with wires, preferably look for one that already has the pin-compatible connectors.
Note you can't just connect a battery in place of the original rechargeable NiCD without modding the power supply path there. If you just do a 1:1 swap the mobo will force current through the new battery, trying to recharge it, and it will eventually cause the battery to fail - bloat and/or leak. That's another reason why I don't like doing CR2032 mods to these old mobos, it's not as simple as it looks.

Deunan wrote on 2025-07-13, 12:48:

If you just do a 1:1 swap the mobo will force current through the new battery, trying to recharge it, and it will eventually cause the battery to fail - bloat and/or leak.

And that's the reason why I claim, that his CR2032 battery "modification" is inappropriately done and I gave him a link and a picture to see how it should be done... but the last question about the chip voltage measuring explained everything.

I noticed the battery is a 3.6V LIR2032. It's rechargeable! What do you think? I checked the voltage at pin 24 and it's 3V with the PC off. If I turn it on, it goes to 0V! Is this normal?

Rechargeable is good, it'll probably work OK. There is no charge control or termination on the mobo, it'll just supply about 4.5V but through a resistor to make the current small. The question is if it's small enough for the coin battery, but I would give it a go. With 3V on the chip it should keep the settings, and time, just fine. But the 0V is not normal at all, who did the mod? Maybe they did remove some diodes and now there is no charging but also no power from PSU. The same resistor I mentioned will drop the voltage very low when the chip is accessed by CPU, it might be even connected to the ALS245 bus driver, which is TTL. That will draw a lot of power through the I/O pins, sagging the battery-only power rail.

I think before messing with this any further you need to know exactly what the battery mod does here.

Before mounting the motherboard in an AT case, it worked fine. It wasn't until after mounting it that the message appeared. The RTC is probably faulty.

I was wrong when measuring the RTC voltages! The voltages are correct:
When it is on, there is 4.8V between pin 12 and 24.
When it is off, about 3.2V.
So the RTC chip seems to be correct. The problem must be internal!

Other question, I have used the JP20 pins (power LED and Keylock) and the LED lights up by connecting pins 1 and 3. Will the Keylock be the two remaining pins 4 and 5? Best regards.
Best regards and Thanks very much

dennisE wrote on 2025-07-14, 04:33:

Before mounting the motherboard in an AT case, it worked fine. It wasn't until after mounting it that the message appeared. The RTC is probably faulty.

It could be, but usually the problem is the corrosion from the original NiCd that damaged the mobo. Did you clean it all up? Sometimes it requires both chemical and mechanical cleaning to even properly see the extent of the damage. Could be a weak trace or via that got completly broken when the mobo was flexed during installation in the case. A broken data line to the RTC chip will make it look like the chip is glitching for example.

dennisE wrote on 2025-07-14, 04:33:

Other question, I have used the JP20 pins (power LED and Keylock) and the LED lights up by connecting pins 1 and 3. Will the Keylock be the two remaining pins 4 and 5?

That's how it is usually wired. The last pin should be GND, you can check that with ohm meter vs the black wire in the PSU cable, should be very low resistance between the two. If so the very next pin is indeed KB LOCK input. You can just put a jumper there if your case doesn't have a key lock.

Will any disaster happen to the motherboard if I connect the key lock to the two pins if they are not those JP20?

Solved!: I checked with an ohmmeter on pin 5 and a black wire from the power supply, and the resistance was very low. I connected the Key Lock and successfully tested it. Thanks very much Deunan!

Regarding the RTC failure, I'm giving up on it. I can still use the PC. Thank you so much for your help. I appreciate it.

Other question, Can the RTC chip be replaced with a new one? Or is it too complicated?

That depends on your skills, can you desolder the old chip without causing any PCB damage? If so it's not complicated at all, except there are no "new" RTCs of that type. Best you can do is buy a NOS part, or even used/desoldered one, on ebay. Also you'd want 146818A if possible. Replacing what you already have with non-A variant from mid-'80 is not a great idea. I will work but will drain the battery even faster.

dennisE wrote on 2025-07-16, 11:41:

Other question, Can the RTC chip be replaced with a new one? Or is it too complicated?

Dont try if it is you first time.
A fault is very easy to be made.
Board can be damaged, and you will be sad.

You need to know how soldering works. Its not just keep the iron to the joint. It depends on more factors.
Also have to know that you dont make the via to hot. Otherwise it will wick it self from the board.

Thank you so much for your advice. I was planning on using a very low-power 14W soldering iron to avoid overheating. Then, to desolder, I'd use a manual desoldering iron or copper tape to hold the solder. What do you recommend?

14W is not enough for soldering big PCBs like these. Although with experience and these new soldering irons with high heat capacity tips, maybe.

It's not so much about power but temperature control. You'd want a soldering station where you can set the temperature. There's always a debate between soldering wick and various suckers - manual or more professional pump operated, I prefer sucker but I use wick too. Problem with wick is if you try to drag it too cold you risk ripping traces and vias when it sticks, and with only 14W and so much copper to take the heat away I would expect it to stick a lot.

In the end it's all about experience though, so perhaps find or get some junk mobo and try to learn on that first. Note, these 286 and many 386 mobos are big but still 2 layer, anything newer will be 4+ layers and way more difficult to solder with low power iron. In fact newer mobos require a preheater even with a good soldering iron with temperature control. As alternative consider cutting the current chip out and then desoldering the pins one by one, way easier that way but you'll ruin the existing chip so you'll have to get a replacement. But cutting the pins is a skill too, you need good, sharp cutter with tip small enough to reach each individual pin. Otherwise you'll end up ripping them out of the mobo, along with vias.

Hi, retro computer geeks!
I've tried again to find the problem with the memory failure the BIOS is showing me. I removed the memory chips, cleaned them with alcohol, and reassembled them. I accidentally mounted one chip upside down, and when I turned it on, memory errors appeared. I realized this and put it back in its correct position, and it worked. Could something have happened to the chip? I'm ruling out that the RTC failure is due to the memory.

To be honest these old DRAMs can die on their own, even without such adventures as being plugged in backwards. That being said I would not rush to replace a chip just because of that, if it was spotted quickly. There should be some memory testers for 286, I think Check-it could do it. So try running a good memory test (this might take some time since you want 3 full passes or so to be sure). It's always good to be sure that RAM is working properly.

I've tried Check-it and there were no problems. Memtest won't start. I don't know what other software to use.

dennisE wrote on 2025-07-19, 12:11:

I've tried Check-it and there were no problems. Memtest won't start. I don't know what other software to use.

Memtest is 386+ software. If Check-it says there is no problem then it is good enough. That chip, if it were to fail, would most likely die completly and not be just glitchy. So if it works to pass the tests I'd say you got away with it.

Deunan wrote on 2025-07-02, 19:23:

So you did try a dumber card and it still didn't work. Then I guess there might be some damage. Things to try: - Connect and use […]
Show full quote

wbahnassi wrote on 2025-07-02, 16:56:

[*]MDA card (8-bit) - No progress

So you did try a dumber card and it still didn't work. Then I guess there might be some damage. Things to try:
- Connect and use RESET button, in case the mobo doesn't init properly for some reason on power-on.
- Make sure the 146818 RTC chip is getting power from PSU, if there is any battery corriosion in that area it migt be unpowered and that will cause havoc.
- While testing the RTC see if you can draw some basic schematic of the connections to the battery header. Some of these mobos require a jumper for external battery for example. IIRC I once ecountered a mobo that needed it for the on-board battery as well, except on different pins. But missing jumper should not prevent the RTC chip being powered from PSU.

And finally there is usually ALS245 chip near the ISA slot and the RTC and KBC. It's a bi-directional driver for the lower 8 bits of the ISA bus, and it often dies (usually not completly, one or more bits get weak or stuck). This will prevent the mobo from being able to talk to any of the cards, and even detect the video memory. You might want to desolder it and put a socket there. If the chip is good it can be socketed, if not you can replace it with ALS or ACT series. Note it's sometimes very hard to figure out the bad data bits if the output is just weak, so consider the socket if you don't fancy in-depth logic probing of the ISA signals.

Ok, here I am again with some more effort to try to fix the motherboard (reminder of its entry on TRW).
I'm quite convinced the board is indeed unable to see ISA cards. I've tried many BIOSes (Phoenix, Ami, Award). Both Ami and Award beep their respective missing display card beep pattern. What I have tried since:

• Probed RTC to see if it's doing its basic things. Check. It has good 5V voltage input, and its input and output clocks are spot on.
• Probed the ALS245 chip. I noticed that between the A and B bus signals there is a considerable difference in the magnitude and shape of the signals. I was poking each matching pair, and they didn't look similar. This indicated to me that it's indeed possibly a dead chip, so I took it out and replaced it with a Motorolla SN74LS245N from another donor board. This unfortunately didn't solve the mobo issue, but probing this replacement chip, the signals are now very similar indeed between A and B busses, so I think this is progress even if it didn't yet solve the problem of the board.

You asked me if I can trace the circuit around the external battery connector. I never did this before, so here is my incomplete attempt at it.. I don't claim I fully understand it yet honestly, so take it with a big grain of salt. BAT J5 is the external battery connector. BAT1 is the original leaky battery that's not installed. The MC chip is the RTC chip. Tracing after R25 wasn't possible as it went into a via to something under a chip. Through this exercise I'm not sure I was able to identify if there's a jumper to control external battery vs. internal. Usually you have to jumper something (like the middle two pins of the external battery connector), but I'm not sure that's the case here.

At this point, I'm open to suggestions if there's anything further I could chase down.

Cheers!