Very likely an 82C206 (KS83C206 on that board) failure. 🙁

CMOS shutdown register read/write error is a common indicator of 82C206 problems, as it integrates the CMOS memory, reset logic, ISA IRQ/ DMA and a bunch of other X-bus peripherals.
If reflowing the pins on the 82C206 doesn't help, it'll probably be high time to get the hot air rework station and a replacement chip ready, preferably a working one taken from an otherwise junk board. Getting a working 82C206 from eBay is a big-time crapshoot. (Most of the ones I was able to get from eBay, turned out to be faulty! >_>)

The KS83C206 is simply Samsung's take on the 82C206, it is fully pin and function compatible with HMC, CHIPS, OPTi, SiS and UMC 82C206 chips. I'd personally try for an HMC or SiS '206, I see fewer dead chips of those two makes for some reason.

Thank you very much for info on that chip and replacement options. Good to know. So one might sey that those ICs are common source of problems? I assume the CMOS part is sensitive..😟 I think I am capable replacing that chip. But finding the new one would be an issue. So you think even buying a "lot" of new ones on e-bay is risky? What is killing them - static electricity? Again, thank you!

When you tested the traces, did you check only the top layer or the vias as well? It's usually vias that corrosion destroys first, but the fix on these mobos is usually simple - gently expose the ends by removing solder mask with something like fibreglass pen, then put a short wire through and solder on both ends. The inner planes of the PCB are typically for power distribution and only the outside ones carry signals on these 386SX mobos.

Deunan wrote on 2022-02-22, 19:34:

When you tested the traces, did you check only the top layer or the vias as well? It's usually vias that corrosion destroys first, but the fix on these mobos is usually simple - gently expose the ends by removing solder mask with something like fibreglass pen, then put a short wire through and solder on both ends. The inner planes of the PCB are typically for power distribution and only the outside ones carry signals on these 386SX mobos.

Thank you. I am aware of this. But I will re-check the small vias from both sides. It is possible I missed something. However that board is not corroded too much - but yes, you never know. The battery looked quite Ok but still I found some green-blue residue. The wierd thing is the residue was only on several places - like one tantalum cap. but not so much arround the battery itself. I have a plan to check the bios integrity as well. Don ´t know anything about the board history - bought for repair from a guy who lost the game as he is not in to board level repair.

If you have EPROM programmer and can read the BIOS chip, I might be able to tell you if the checksum is correct or not. As for '206 chips dying, I know people here reported such issues but I never had one of those going bad, most of the motherboards I fixed were down due to battery corrosion and deep scratches on the bottom that cut some traces (I usually pick out cheap mobos that already went through a garbage bin or two).
I assume you've tried different memory sticks? You need at least two in one bank, and some motherboards (though none as new as this one) don't like the 3-chip SIMMs. If you think you'll be fixing more of these or you really want this particular mobo running again, then consider buying a POST card. It really helps with issues like this one.

At least one of my 386SX boards (Opti 82c291 based) refuses to boot without all 4 simm slots populated. Can't remember what errors it throws, but it's not something memory-related, so this threw me off for quite a while when diagnosing it (dead KBC and heavy corrosion compounded the issues).

H.W.Necromancer wrote on 2022-02-22, 17:59:

Thank you very much for info on that chip and replacement options. Good to know. So one might sey that those ICs are common source of problems? I assume the CMOS part is sensitive..😟 I think I am capable replacing that chip. But finding the new one would be an issue. So you think even buying a "lot" of new ones on e-bay is risky? What is killing them - static electricity? Again, thank you!

Unfortunately, I've been noticing quite a few 286/386/486 boards with dying 82C206 chips, so I'd personally say that they have a bit of a failure rate these days.

I don't know why they are starting to die off all of a sudden, I suspect the chips were perhaps a little too integrated for their time, and seriously pushed the limits of their process tech, but that's just my best vaguely educated guess. It's also possible that, since they are part of the ISA bus, they are vulnerable to spikes/surges/etc. and other potentially damaging transients that may afflict the ISA bus. Given that the power supplies (especially the more cheaply-made units) in many older computers are not necessarily aging with dignity, that would seem at least remotely plausible.

Hmmm no battery on the board, did you pull it ? and the Ext Bat pins are missing. just an observation...

Horun wrote on 2022-02-23, 00:38:

Hmmm no battery on the board, did you pull it ? and the Ext Bat pins are missing. just an observation...

Yes, the previous owner cut of the battery and tested with a new one and a bunch o memory sets and VGA. Always the same result. Than he fixed the old battery back (not very well as it had been already cut) and sent the board to me. I removed the battery and cleaned everything . Ext. battery pins are factory not installed - probably because it was a super-cheap design - consider how old 386SX already was when produced. The good quality GP battery is not as dangerous as Varta and it did not cause such a mess. The overal good condition of the board leads me to the assumtion there is an electronic failure../-: However I think the board is salvagable.

Eep386 wrote on 2022-02-22, 23:22:

H.W.Necromancer wrote on 2022-02-22, 17:59:

Thank you very much for info on that chip and replacement options. Good to know. So one might sey that those ICs are common source of problems? I assume the CMOS part is sensitive..😟 I think I am capable replacing that chip. But finding the new one would be an issue. So you think even buying a "lot" of new ones on e-bay is risky? What is killing them - static electricity? Again, thank you!

Unfortunately, I've been noticing quite a few 286/386/486 boards with dying 82C206 chips, so I'd personally say that they have a bit of a failure rate these days.

I don't know why they are starting to die off all of a sudden, I suspect the chips were perhaps a little too integrated for their time, and seriously pushed the limits of their process tech, but that's just my best vaguely educated guess. It's also possible that, since they are part of the ISA bus, they are vulnerable to spikes/surges/etc. and other potentially damaging transients that may afflict the ISA bus. Given that the power supplies (especially the more cheaply-made units) in many older computers are not necessarily aging with dignity, that would seem at least remotely plausible.

I would say you are close to the source of failures. I am now thinking the same way. I have been studying the datasheets for a while now and I think even a faulty or bad seated ISA card can kill it. And well the web information I have found says those integrated chips were not considered 100% reliable at their times - as you states - "integrated and pushing technology to the limits" - you are most probably right. And what hits my nose now - they have to keep the bios setting anc clock thus tey are still powerd from the battery... This board is so simple that there is not much to protect anything and the chance to kill the chip directly is probably higher. I have found an info these chips were often socketed in the past to make repair of expensive motherboards easier. However this board most likely was not so expensive in 1994 as his bulky older predecessors . But I have weakness for those cute compact mobos so will do my best (-:

quicknick wrote on 2022-02-22, 22:28:

At least one of my 386SX boards (Opti 82c291 based) refuses to boot without all 4 simm slots populated. Can't remember what errors it throws, but it's not something memory-related, so this threw me off for quite a while when diagnosing it (dead KBC and heavy corrosion compounded the issues).

Thank you. Tested with 2 and 4 and also no memory. chips 1Mb modules. No change. It seems the board is not even getting to the part of memory initialization.

Deunan wrote on 2022-02-22, 21:38:

If you have EPROM programmer and can read the BIOS chip, I might be able to tell you if the checksum is correct or not. As for '206 chips dying, I know people here reported such issues but I never had one of those going bad, most of the motherboards I fixed were down due to battery corrosion and deep scratches on the bottom that cut some traces (I usually pick out cheap mobos that already went through a garbage bin or two).
I assume you've tried different memory sticks? You need at least two in one bank, and some motherboards (though none as new as this one) don't like the 3-chip SIMMs. If you think you'll be fixing more of these or you really want this particular mobo running again, then consider buying a POST card. It really helps with issues like this one.

Thank you. Unfortunately I have no older 1Mb memory modules with more chips. Just one set of 4Mb. But all the 1 Mb are 3 chips design. However this board should make it with them. It is really one of the latest. I think. POST card is needed - waiting till it comes from China. Sadly not possible to get it here locally. I have tried more memory modules and the previous owner as well. But the beeps are the same even with no memory which shouldn ´t be like that. And yes, I am repairing more boards - buying equipment is not a problem. Same as you - trying to save boards from scrap. However even those are getting more and more expensive. I have just saved quite badly corroded 486 VL-BUS.

I'd say wait for the POST card to arrive before replacing the '206 chip. In the meantime re-test all the vias, including those right above C19 and C20. If in doubt unmask them and fill with solder, or use the wire method. There might also be a few of those under the NPU socket, as you can see the soldermask protected the traces pretty well but vias got hit hard. Chances are you'll need to remove that socket to make sure, and that BTW will require hot air station and some PCB pre-heating. Another method is to destroy the socket and solder in a replacement afterwards. But for that I'd wait for POST card results as well.

Deunan wrote on 2022-02-23, 11:25:

I'd say wait for the POST card to arrive before replacing the '206 chip. In the meantime re-test all the vias, including those right above C19 and C20. If in doubt unmask them and fill with solder, or use the wire method. There might also be a few of those under the NPU socket, as you can see the soldermask protected the traces pretty well but vias got hit hard. Chances are you'll need to remove that socket to make sure, and that BTW will require hot air station and some PCB pre-heating. Another method is to destroy the socket and solder in a replacement afterwards. But for that I'd wait for POST card results as well.

Magnet wire (or really thin wire unwrapped from a small transformer or choke out of one of those cheap firecracker PSUs) works good at bypassing and/or jumping vias that are damaged. If the wire is thin enough you can stick the wire through the via itself after cleaning the via of solder and corrosion.

Of course exhaust all other possibilities before taking the plunge into replacing the 82C206, but I still strongly suspect that you've got a dead chip on your hands. That said, also make sure the 2N3904 and 2N3906 transistors aren't shorted. They ought to be Q1 and Q2 on that board. You can test them both in and out of circuit with a multimeter - none of the transistor pins should be shorted in any case.

Deunan wrote on 2022-02-23, 11:25:

I'd say wait for the POST card to arrive before replacing the '206 chip. In the meantime re-test all the vias, including those right above C19 and C20. If in doubt unmask them and fill with solder, or use the wire method. There might also be a few of those under the NPU socket, as you can see the soldermask protected the traces pretty well but vias got hit hard. Chances are you'll need to remove that socket to make sure, and that BTW will require hot air station and some PCB pre-heating. Another method is to destroy the socket and solder in a replacement afterwards. But for that I'd wait for POST card results as well.

Thank you, I think I go this way. I have a decent hot air station, but in case of chip/socket removal a preheater must be improvised. But it is possible.

Eep386 wrote on 2022-02-23, 16:10:

Deunan wrote on 2022-02-23, 11:25:

I'd say wait for the POST card to arrive before replacing the '206 chip. In the meantime re-test all the vias, including those right above C19 and C20. If in doubt unmask them and fill with solder, or use the wire method. There might also be a few of those under the NPU socket, as you can see the soldermask protected the traces pretty well but vias got hit hard. Chances are you'll need to remove that socket to make sure, and that BTW will require hot air station and some PCB pre-heating. Another method is to destroy the socket and solder in a replacement afterwards. But for that I'd wait for POST card results as well.

Magnet wire (or really thin wire unwrapped from a small transformer or choke out of one of those cheap firecracker PSUs) works good at bypassing and/or jumping vias that are damaged. If the wire is thin enough you can stick the wire through the via itself after cleaning the via of solder and corrosion.

Of course exhaust all other possibilities before taking the plunge into replacing the 82C206, but I still strongly suspect that you've got a dead chip on your hands. That said, also make sure the 2N3904 and 2N3906 transistors aren't shorted. They ought to be Q1 and Q2 on that board. You can test them both in and out of circuit with a multimeter - none of the transistor pins should be shorted in any case.

Good point. I have a transistor tester if needed. the wire method is well known to me. I have some super thin copper wire.