Are you sure the buffer 74F244 and the 74F125 are still o.k.?
I recommend always inserting DIL sockets after restoring the traces so you can try new integrated circuits if you suspect the old one to be broken.

Have you checked carefully under the cache chips if there´s further corrosion and tested all the traces in that area?

All in all this isn´t a bad case of battery leakage, but probably there are a few more traces damaged.

If you can, definitely desolder the non-socketed chips around that area and check for more corrosion. It tends to get under them quite badly. Continuity check on anything that has corrosion on it, and check that the first ISA slot has continuity to the last one.
I had a 286 board that had a sneaky broken trace under an ISA port once and it drove me crazy until I buzzed out each pin.

majestyk wrote on 2022-07-07, 17:00:

Are you sure the buffer 74F244 and the 74F125 are still o.k.? I recommend always inserting DIL sockets after restoring the tra […]
Show full quote

Are you sure the buffer 74F244 and the 74F125 are still o.k.?
I recommend always inserting DIL sockets after restoring the traces so you can try new integrated circuits if you suspect the old one to be broken.

Have you checked carefully under the cache chips if there´s further corrosion and tested all the traces in that area?

All in all this isn´t a bad case of battery leakage, but probably there are a few more traces damaged.

Yep, i have removed cache chips, and i checked traces beneath, all seems to be fine, and yes, i wasn't smart to solder dil sockets. So, good thing is to check ICs?
Thanks

DNSDies wrote on 2022-07-07, 19:34:

If you can, definitely desolder the non-socketed chips around that area and check for more corrosion. It tends to get under them quite badly. Continuity check on anything that has corrosion on it, and check that the first ISA slot has continuity to the last one.
I had a 286 board that had a sneaky broken trace under an ISA port once and it drove me crazy until I buzzed out each pin.

I will try to check continuity between ISA ports! Thanks
edit: All the traces between isa ports have continuity!

vias tend to get broken in a way you cant tell without measuring away from it, measuring directly often temporarily restores connection
I would also ty with half the ram mounted, and without cache installed

rasz_pl wrote on 2022-07-07, 21:36:

vias tend to get broken in a way you cant tell without measuring away from it, measuring directly often temporarily restores connection
I would also ty with half the ram mounted, and without cache installed

I tried with 1, 2, 4, 8 rams, but it doesn't POSTs everytime. With 8 rams inserted, it POSTs everytime. I didn't try without cache, so, that's next step!
Thanks

I wonder what might be causing the "broken picture", it looks very similar to a problem that I was unable to solve with a PCChips M209 (not caused by leaky battery).

386DX needs 4x 30pin simms at the minimum (unless chipset does magical things, but this one doesnt), try with 4 in either bank, and mark simms so you can try different combinations to exclude possibility of one of them being bad, not that bad ram is highly suspected here.

Im trying to figure out a scenario where that picture would be produced, and its difficult. ISA communication not working would not execute VGA bios and would not set proper text mode (clocks). Maybe partially corrupted BIOS contents? or one of address lines stuck making just one part of bios fail to execute properly?

manual and bios file https://www.ultimateretro.net/en/motherboards/4426
chipset http://www.dosdays.co.uk/media/opti/82C495.pdf
another picture of this board and picture of post screen http://www.dosdays.co.uk/topics/Manufacturers/opti.php

quicknick wrote on 2022-07-07, 23:00:

I wonder what might be causing the "broken picture", it looks very similar to a problem that I was unable to solve with a PCChips M209 (not caused by leaky battery).

Yes, the pictures are same as mine. Bad news - You were unable to fix? 🙁

rasz_pl wrote on 2022-07-07, 23:14:

386DX needs 4x 30pin simms at the minimum (unless chipset does magical things, but this one doesnt), try with 4 in either bank, […]
Show full quote

386DX needs 4x 30pin simms at the minimum (unless chipset does magical things, but this one doesnt), try with 4 in either bank, and mark simms so you can try different combinations to exclude possibility of one of them being bad, not that bad ram is highly suspected here.

Im trying to figure out a scenario where that picture would be produced, and its difficult. ISA communication not working would not execute VGA bios and would not set proper text mode (clocks). Maybe partially corrupted BIOS contents? or one of address lines stuck making just one part of bios fail to execute properly?

manual and bios file https://www.ultimateretro.net/en/motherboards/4426
chipset http://www.dosdays.co.uk/media/opti/82C495.pdf
another picture of this board and picture of post screen http://www.dosdays.co.uk/topics/Manufacturers/opti.php

Only thing i could do more is to "warm up" board a little with hot air (opti and sis chip), and try to flash BIOS (my friend has flasher for those old chips), right? Odd thing is that when i remove VGA card it beeps and when i inserted VGA there is no one beep for successful POST, but the monitor turns on with this broken textures. I'm pretty sure that the broken traces are fixed, so i need to find which chip is faulty. I was hoping someone had the same problem and remembering what was the cause for this. This would be my "best" 386 PC if i can fix it. 😀
Thanks

I had a similar garbled screen with a broken periphery chip:

Ways to safely transplant / replace 160-pin QFP chips

The errors responded to heat / cold in this case.

rasz_pl wrote on 2022-07-07, 21:36:

vias tend to get broken in a way you cant tell without measuring away from it, measuring directly often temporarily restores connection
I would also ty with half the ram mounted, and without cache installed

Tried without cache chips - same!

majestyk wrote on 2022-07-08, 06:58:

I had a similar garbled screen with a broken periphery chip:

Ways to safely transplant / replace 160-pin QFP chips&

The errors responded to heat / cold in this case.

You think i could do some work with changing sis chip?

You could try applying coolant spray or hot air to each of the two chips before starting the mainboard and check if there´s any change.

majestyk wrote on 2022-07-08, 07:16:

You could try applying coolant spray or hot air to each of the two chips before starting the mainboard and check if there´s any change.

Tried with hot air - nothing. Don't have any coolant spray....

rasz_pl wrote on 2022-07-07, 23:14:

386DX needs 4x 30pin simms at the minimum (unless chipset does magical things, but this one doesnt), try with 4 in either bank, […]
Show full quote

386DX needs 4x 30pin simms at the minimum (unless chipset does magical things, but this one doesnt), try with 4 in either bank, and mark simms so you can try different combinations to exclude possibility of one of them being bad, not that bad ram is highly suspected here.

Im trying to figure out a scenario where that picture would be produced, and its difficult. ISA communication not working would not execute VGA bios and would not set proper text mode (clocks). Maybe partially corrupted BIOS contents? or one of address lines stuck making just one part of bios fail to execute properly?

manual and bios file https://www.ultimateretro.net/en/motherboards/4426
chipset http://www.dosdays.co.uk/media/opti/82C495.pdf
another picture of this board and picture of post screen http://www.dosdays.co.uk/topics/Manufacturers/opti.php

I have desolder BIOS chip, and put a DIL socket, so, when my friend will have time, i will go to him to flash the bios with the image on ultimateretro. That will be my next step.

is the bios a flash chip or OTP prom? I have a vague memory of cheaper manufacturers doing non reflashable bios chips at that time frame
anyway since you are doing bios we can experiment - we(I) can mod the bios to do custom things and observe the results. Do you have a logic probe/scope/logic analyzer? POST card?

please try VGA in 8bit slot. BIOS execution is 8 bit, VGA initialization will also be 8bit port IO operations (same with keyboard), but bios finally trying to put something in 0xB8000 will use 16bit memory transfers. Now Im thinking would broken 8-16bit data bus arbitration produce such picture? Color Text mode stores data as [char, attrib, char, attrib, char, ...]. What are we seeing here exactly? https://int10h.org/oldschool-pc-fonts/fontlis … nt?trident_8x16
is it █? 0xDB? it looks like
128x blank 0x00 or 0x20
128x █
99x blank
1x green
28x blank
256x █
128x blank
128x █
144x blank
128x █
128x blank
this ... those next lines are not full height?!?! vertically it doesnt line up, however you count it it doesnt add up to 25, so maybe its not initializing proper VGA timings after all?

how does the keyboard act? do you get 3 leds flash? with the garbled screen what happens when you press capslock/numlock/scrolllock?

hmm, general question - would FDD initialize/seek if CMOS is empty?

ps: Im more and more tempted to do an advanced POST card with three RP2040 (cheaper than a bunch of 74ls374 latches or cpld) capturing all ISA traffic and tracing whats happening on boot. Would be able to account for every single byte read/write and IO operation by the system in early boot.

rasz_pl wrote on 2022-07-08, 11:37:

is the bios a flash chip or OTP prom? I have a vague memory of cheaper manufacturers doing non reflashable bios chips at that ti […]
Show full quote

is the bios a flash chip or OTP prom? I have a vague memory of cheaper manufacturers doing non reflashable bios chips at that time frame
anyway since you are doing bios we can experiment - we(I) can mod the bios to do custom things and observe the results. Do you have a logic probe/scope/logic analyzer? POST card?

please try VGA in 8bit slot. BIOS execution is 8 bit, VGA initialization will also be 8bit port IO operations (same with keyboard), but bios finally trying to put something in 0xB8000 will use 16bit memory transfers. Now Im thinking would broken 8-16bit data bus arbitration produce such picture? Color Text mode stores data as [char, attrib, char, attrib, char, ...]. What are we seeing here exactly? https://int10h.org/oldschool-pc-fonts/fontlis … nt?trident_8x16
is it █? 0xDB? it looks like
128x blank 0x00 or 0x20
128x █
99x blank
1x green
28x blank
256x █
128x blank
128x █
144x blank
128x █
128x blank
this ... those next lines are not full height?!?! vertically it doesnt line up, however you count it it doesnt add up to 25, so maybe its not initializing proper VGA timings after all?

how does the keyboard act? do you get 3 leds flash? with the garbled screen what happens when you press capslock/numlock/scrolllock?

hmm, general question - would FDD initialize/seek if CMOS is empty?

ps: Im more and more tempted to do an advanced POST card with three RP2040 (cheaper than a bunch of 74ls374 latches or cpld) capturing all ISA traffic and tracing whats happening on boot. Would be able to account for every single byte read/write and IO operation by the system in early boot.

ufff, i don't have have a logic probe/scope/logic analyzer or a POST card.
I tried put the vga in 8-bit slot, same result, same picture on monitor.
I get 3 leds flash on keyboard when i turn on the computer, but nothing happens when i press any of "locks" when garbled screen shows up.
I have connected Floppy with isa ide controller, but no sound or light from it.
I will find some compatible chip for BIOS to write on it, if this one is read only (as i think it is).

butjer1010 wrote on 2022-07-08, 13:53:

without VGA board also beeps, but 1 long and 4 short beeps

did you mean 1 long 2 short?

butjer1010 wrote on 2022-07-08, 13:53:

I tried put the vga in 8-bit slot, same result, same picture on monitor.

🙁 such a nice and convenient theory disproved

butjer1010 wrote on 2022-07-08, 13:53:

I get 3 leds flash on keyboard when i turn on the computer, but nothing happens when i press any of "locks" when garbled screen shows up.

means bios crashes/gets stuck

post codes and order of initialization: http://mrbios.com/techsupport/award/postcodes.htm [AMI New BIOS; 06/06/92-08/08/93]
must be hanging somewhere between 2C and 2D? hmm how is the 5V on ISA slot? lack of proper power would make VGA draw parasitic power thru data pins and act weird

What happens when you do a hardware reset using S1 reset jumper?

Off the wall idea - try taking off JP2 jumper switching to mono monitor, with any luck this will force text mode 7 with different base address and attribute meaning, and produce different picture

poke carefully at all the 82C495XLC chipset pins with a toothpick to check if there arent any loose

Im running out of ideas, deeper diagnosing would require some tools (any of: post card, eprom flasher, logic analyser)

PS diagram of reverse engineered board on this chipset https://alexandrugroza.ro/microelectronics/sy … c-schematic.png sadly modular and not pdf , but still great resource. You can see where all the 244/245 buffers required for proper ISA operation connect