I started working on this board again and I can now report that the board at least starts POSTing. The post card shows that the board stops at the error codes 40 7b. Unfortunately the screen keeps being black/no signal. The board beeps (if I count correctly) one times long and nine times short. The board has an AMI bios. The short manual of the post card describes the error as "check EPROM". I checked the EPROM and it reads fine in my EPROM programmer. I found an dump of the BIOS at https://www.ultimateretro.net/de/motherboards/7092. This dump differs one some points from the one of my board. I programmed the other BIOS onto a 27512, but the error beeps are the same.

maybe there's a problem with the ROM chip itself (it can happen to old ROMs, as they get old, they get "weak" which is fine for a programmer, but not for a computer). Or maybe there's a broken address line ?

In my experience, the risk of battery leakage actually damaging a board is overrated. It looks pretty bad, but how often does it really eat through protective varnish and all the copper?
Boards choose not to boot for all kinds of reasons but some of mine do work just fine after serious battery malfunction in storage.

Then again, capacitors are not supposed to blow up. It's generally not considedered civilised behaviour for electronic components and very much frowned upon.

To be blunt, if it's not about tinkering per se or the challenge for you and replacing that cap was a serious hurdle so far - consider cutting your losses and trying another piece?
Because, that there in the picture is a 486 board, alright.
Now, those with PCI bus, ZIF socket, Voltage regulators, PS2 style RAM sockets and mouse connector are truly rare and expensive by now. 50 bucks upwards in auctions.
But, that one has neighter of those. You could do better in any case.

Decent VLB boards are not that hard to come by. for a system up to 80 MHz. I have two PAT48PG ones, that's a rather unknown brand, won't pull too much interest.
(the downside being the rather arcane bios or documentation environment...)
Just look out for dusty boards with SIS 471 or Opti 895 chipset, you'll be fine.

Yes, VLB boards are a little cheaper to get these days. But when it comes to VLB addon cards you are in trouble. VLB I/O cards and graphics cards are not that easy to aquire.

feltel wrote on 2022-02-27, 18:42:

VLB I/O cards and graphics cards are not that easy to aquire.

Really? OK, half a dozen each lying around here since a decade, so I haven't checked in quite a while.
Consider, you won't actually need a VLB controller, ISA will do fine.
VGA, well, I keep seeing them on the bay all the time and 2 MB of memory are standard for those, while quite rare in ISA.
Retro being retro, but 2 MB for Windows is kinda nice.

The main reason for not using VLB is supposed to be the pain of getting them to run stable, but it's the classic 486.

Cuttoon wrote on 2022-02-27, 16:58:

In my experience, the risk of battery leakage actually damaging a board is overrated. It looks pretty bad, but how often does it really eat through protective varnish and all the copper?

I my experience, yes they do 🙁
After a while, it gets really ugly.

Tantalum caps blowing up is not "normal", but expected, because they tend to fail short, so when they get fed with a few amps of power, they explode. Replacing them should be good enough.

I understand your point about buying another one, which is definitely an option, but this kind of thing isn't produced anymore, so the more you trash, the less there will be out there for others. I don't think this is broken enough to be considered as trash yet, since it seems to work for the most part, with an (unknown yet) inconvenient issue

Let me bring up this topic again. The board landed on my todo stack and slowly worked up on top again. I've since replaced the blew up tantal with an 10uf/35V one. No change. The board does contrary to my former believe one long beep and eight short beeps. This should be a video related problem. The beep code is the same without an VGA card at all. The VGA card works in another system. Its an Trident TVGA 8900C. Nontheless I re-seated all socketed chips. Just for testing I replaced the keyboard controller chip and tested it in another machine. It works. The voltages are correct, 5V, 12V, -12V. I disabled the cache memory. No change. Could it be possibe that due to that is a 50 MHz bus clock system that some VGA cards wont work? I know that could be an issue on VLB and PCI systems, but on ISA only too? The bus clock is at 7,xx MHz, I checked on Pin B20 on the ISA slot. There is a jumper CLKOUT=CLKIN/CLKOUT=CLKIN/2 which should set the machine to 25 Mhz bus clock. Even with that set no change. What else can I check?

How did you disable cache memory when you cannot enter BIOS?

There is a resistor network block, where you can set between 64k and 256k cache. I'm under the impression that removing the resistor network at all disables the cache.

We can´t be sure this is sufficient to turn off the cache chips. To be on the safe side I would remove all cache /TAG chips. Defective cache chips can indeed prevent a system from starting.

Have you checked the single thin trsace that seems to end somewhere under the PSU connector for continuity? It wouldn´t hurt to remove the PSU connector to rule out any damage there.

There also seem to be a lot of bent pins (at resistor networks etc.) on the soldering side. I have had several cases where bent pins touched each other causing shorts. If you haven´t done so already you should check everything very carefully.

Thanks alot. I removed all the cache chips and the resistor network. Unfortunately the error persists. While I replaced the blown tantalum I removed the power connector. The trace across the battery place is fine and connects through an resistor and an diode to an 74LS00 chip right under the BIOS chip.

The bent pins on the solder side look factory. No pins touching each other as far as I checked them.

feltel wrote on 2023-05-28, 08:49:

Thanks alot. I removed all the cache chips and the resistor network. Unfortunately the error persists. While I replaced the blown tantalum I removed the power connector. The trace across the battery place is fine and connects through an resistor and an diode to an 74LS00 chip right under the BIOS chip.

The bent pins on the solder side look factory. No pins touching each other as far as I checked them.

How much RAM did you use for the test?

4 one megabyte sticks. The RAM is known good one.

It´s a pity since the mainboard is in very good condition and the battery has done very little damage.
Time to think about some loose solderings at / or even a defective SiS 85C460 chip.

Unfortunately I dont have another ISA VGA card to rule out any possible incompatibilities. I'll start toning out traces around the ISA bus at first.

feltel wrote on 2023-05-29, 08:57:

Unfortunately I dont have another ISA VGA card to rule out any possible incompatibilities. I'll start toning out traces around the ISA bus at first.

So when the BIOS beep codes are pointing to Video RAM defective, you should definitely grap another ISA video card. Better get two (different chips).

Yes, I'm on the hunt for it. The card I use is working fine in another board.

Also, it's worth checking if the board would work with good battery. Some motherboards require a battery to operate.

Success. I started checking the underside of the board for broken traces. i found a single one, which was affected by the battery leakage. Just this single one. It connects from the 74LS245 right before the first ISA slot to a via and then on the top side to the ISA slots. I temporary bodged it and the board came to life. 😀

I should replace the three passives (one resistor, two diodes) which were affected too. Strange that the parts look okay from the top side. The board must have sit in a drop of battery juice on this spot.