Well, I got my hopes up. The original problem is fixed, the system POSTS correctly every time, but unfortunately it will not start an operating system. If I try to boot from a floppy I always get a read error - I tried 3 different drives and 3 known good boot disks (DOS 6, 6.2 and 5) with dos 5 on a 720k DD instead of a 1.44MB HD. If I try to boot from hard drive with an OS already installed it just locks up. I've also tried all these combinations with a XT-IDE ROM in there. It detects the hard drive but it can't boot on it and it behaves more or less the same way with the floppies.

At boot, it seems to read the floppy for about 30 seconds, then the system locks up, but the behavior varied a little with the floppy drive used. One of them got as far as it displaying "Starting MS-DOS" before locking up, but that's the furthest I ever got.

I also tried to slow down the memory as much as possible, with and without turbo on - it just takes longer but the same thing happened.

I know the I/O controller I am using is good, I stuck it in a 486 board shortly after and that booted off both my floppies and CF cards with and without XT-IDE - all the same ones I tried on the 386.

So i'm stuck again.

I'm trying to repair a 386DX board at the moment and I recently found a bad line 2cm 5cm away from the varta under an IC.

So there may still be bad lines - under keyboard controller, AT power, RAM...

Seeing how battery was able to eat resistor leg thru I would be pulling chips off all around this corner.
Floppy not working can be DMA lines, IDE maybe irq? or just lines going to ISA slots in general.

wow!

awesome detective work!
well done and another wonderful motherboard saved

All of the DMA and IRQ lines run to the UM82C206F from the first ISA slot. I removed the keyboard controller socket, one of the logic chips under it, and the first ISA slot to look for any kind of corrosion. I didn't find a smoking gun at all. There were a couple of traces with some surface corrosion and one via that looked a little iffy (the ones closest to the battery where the keyboard controller used to be) but all tested good with my multimeter. While I can't find a datasheet for the UM82C206F (only the UM82C206 which has 84 pins - the 206F has 100), I can verify that DRQ 2, DACK 2 and IRQ 6 (all required for the floppy to work) are continuous from every ISA slot to a pin on the 82C206F. With the ISA slot off it is easy to see all of the DMA and IRQ lines. I know there's a couple off-colour traces down by the 74LS32 there but those all test good as well. A few of the minor scratches visible on the traces was me getting my multimeter through the solder mask.

There are only two traces that go under the power connector, one from each inductor below the keyboard connector to R9 and R10. Both are short, easy to see and easy to measure - and they both test good so I didn't bother removing it.

At this point I don't think I will bother with reassembly unless I find some clear reason for the floppy drive not working properly and the system locking up. I have another 386 motherboard coming that I got a reasonably good deal on with what I hope is fairly minor battery damage. It has the same UMC chipset so I think I will tackle that one instead and put this one aside.

I'll gladly check any other lines if anyone has any ideas, but for now I think this one is beyond my ability to troubleshoot.