• Please review our updated Terms and Rules here

Help resurrecting SER 386-ADIII motherboard - Battery leak damage

maferv

Experienced Member
Joined
Jan 5, 2017
Messages
62
Location
Argentina
Hello people,

I got this motherboard not long ago and at first sight I could notice it was heavily damaged by a battery leak.
Thankfully it booted, but the keyboard wasn't working.

ser-386ADIII-front.jpg

So I cleaned the leak which left some traces exposed.

At least one trace was damaged, one that goes from the keyboard DIN-5 female connector to a... is that a resistor? What is it? What are those two?
ser-386ADIII-front2.jpg

I soldered a wire to fulfill the duty of the damaged trace, and now the keyboard works!

ser-386ADIII-back2.jpg

But there's more.

The motherboard is very unstable.

Sometimes it doesn't even post
Sometimes the display doesn't work (I have tested all different ISA 16-bit sockets and many different video cards)
Sometimes the display works but then it freezes along with all the system
Sometimes if I finally boot to MS-DOS, I may see this message:

Invalid device request reading device CON
Abort, Retry, Ignore, Fail?


And then all hope is lost. It's frozen.

Only a few times it has worked OK but only for a few minutes.

I wonder, maybe these two black... resistors? Have been damaged and are causing this problem (I hope). I included pictures so you can please advice me what to do.

I looked at my dead motherboards and found this one that has these similar, again, resistors? (please tell me what they are) And I was wondering, could I replace them with the ones on this motherboard? Are they the same? Do they have a fixed orientation?

non-working-slot1-mb.jpg

Do you think this sounds like a good idea? What have been your experiences dealing with this kind of damage? What sorts of tests and diagnostics do you recommend me to do?

I tried with different PSUs, I/O cards, video cards, etc.

Thank you for sharing your knowledge, experiences and helping.

More info about the motherboard:

SER / SERITECH 386-ADIII or 386-AD3

Chipset
ALI M1429 A1/9403 TS6/B17707 and M1431 A2/9349MA020

BIOS String
American Megatrends Inc., 40-0212-001133-00101111-040493-ali1429-f

CPU
Am386 DX-40

Cache
128 KB

http://th99.classic-computing.de/src/m/S-T/32373.htm
 
Maybe you still have to look for damaged traces. Cleaning it with Lemon juice and after that with Isopropyl Alcohol works for always. Sometimes you have to do it twice! Those black things looks like just ferrite filters
 
Now I got a new error message:

8042 GATE-A20 ERROR
SYSTEM HALTED

The real problem with damaged traces is that on a multi-layer board damage can occur on the internal layers as well.

I honestly hope you're wrong. It would break my heart.

Maybe you still have to look for damaged traces. Cleaning it with Lemon juice and after that with Isopropyl Alcohol works for always. Sometimes you have to do it twice!

That's precisely what I used, lemon juice and isopropyl alcohol. A long time. I don't think it would help much to keep rubbing lemon juice but I don't know, maybe you're right. I looked for more broken/damaged traces but I couldn't find any other related to the keyboard. All checked out OK with the multimeter (although I didn't follow all the traces of the motherboard, as I'm not mad... yet).

Those black things looks like just ferrite filters

Hey thanks for enlightening me. Would it be OK if I replace the ferrite filters with the ones from the dead motherboard I mentioned (picture attached in my previous post)? I guess they may be related to the problem, although I'm not 100% sure of course. As you may have guessed already, my knowledge in electronics is very limited.

Would you recommend to replace any other components of the motherboard? The only thing that seems to be failing is the keyboard.

We're close but it's still far.

Any advices would be very welcome. Party on.
 
The real problem with damaged traces is that on a multi-layer board damage can occur on the internal layers as well.

I honestly hope you're wrong.
Liquids can disperse through capillary action so there's no reason to assume that liquid battery leakage hasn't crept through the tiny nooks and crannies in the area where the leakage has occurred. I've worked on many boards that have suffered leakage and even after having repaired all the surface problems with the traces, etc., there were still problems with the boards.
 
The two things I see are the proximity of the BIOS and what looks to be external cache, but it's in a ridiculously distant location from the CPU itself.
Are you positive you have cleaned the chips and sockets well? I have a battery damaged board where sometimes if you breathe on the BIOS EPROM wrong it won't POST.
 
I have some news.

I replaced the ferrite filters and a capacitor that wasn't looking all that great and I noticed some improvement, but still it kept freezing all the time.

The two things I see are the proximity of the BIOS and what looks to be external cache, but it's in a ridiculously distant location from the CPU itself.
Are you positive you have cleaned the chips and sockets well? I have a battery damaged board where sometimes if you breathe on the BIOS EPROM wrong it won't POST.

You are right about the external cache. I double checked and cleaned all the sockets and chips again, but still, no changes. So I tried disabling the external cache in the BIOS and... bingo!
Now the motherboard is working fine so far, but I miss the cache. And it was a generous one, 128 KB. I was interested in repairing this motherboard because of it mainly, since I only have another 386, and it has a smaller external cache, 32 KB and it is directly soldered to the board, no sockets.

I tried replacing the cache chips with others, but nothing changed. So I'm inclined to rule them out. Possibly there is some damage to the PCB, or internal traces.

Do you know of any way to troubleshoot this? Do you have any other ideas about what can be going wrong?
 
Do you know of any way to troubleshoot this?

If it truly is a multi-layer board and some of the internal traces have been compromised, then you'd need to find a true schematic to run trace-wire between the suspected break. If somehow the board is shorted internally (it can happen), then---sorry to say---you're wasting your time.
 
Something strange happened this time.

I turned the computer on and the cache worked fine. Just fine, for about an hour when I turned it off. I turned it on again, and it still was fine, but I turned it off and on again and... the same error as always.

it makes me think, if there was a damaged trace, why would it work fine once and don't work the next moment? Shouldn't it be consistently failing?

I wonder there must be some faulty connection or a faulty component (capacitor, or what else?), but I'm not sure what it is. What could it logically be? What would make sense?
 
...if there was a damaged trace, why would it work fine once and don't work the next moment? Shouldn't it be consistently failing?

Corrosion can cause strange things in a circuit board. I see it every day at work. Things will work just fine for hours and then "blam!", boards don't work anymore.... Same thing when you have components that have been "marginally-damaged". Unsolicited 2¢ opinion---find another board...you could waste years and still not get it going....
 
Corrosion can cause strange things in a circuit board. I see it every day at work. Things will work just fine for hours and then "blam!", boards don't work anymore.... Same thing when you have components that have been "marginally-damaged". Unsolicited 2¢ opinion---find another board...you could waste years and still not get it going....

You're probably right, but I have yet again good news.
The motherboard is still a little faulty at times: Blank screen or very dim screen at boot, COM port in an ISA I/O card not working (although maybe the card isn't all that well but ATM I don't have other cards to test it), and other minor things, but not very frequently.
Most of the times, after replacing some caps as I said before, it seems to be stable when I disable the cache.

And here's the good part:
I found a way to get the external cache working fine.

Usually enabling the external cache gets your computer frozen after a few minutes, depending on what you run. I usually test the computer using some programs or doom. Doom works for a few moments and then displays some screen glitches and that's it. But I decided to test a Win95 installation from another HD.
This is what I got:
Code:
HIMEM.SYS has detected unreliable XMS memory at address  0010d980

After reading this:
http://discussions.virtualdr.com/sh...-HIMEM-SYS-has-detected-unreliable-XMS-memory

I figured there could be a problem with the speed of the memory. So I accessed the BIOS and saw this configuration:

Code:
AUTO Config Function: Enabled (the options in this section are displayed but you can't change them unless you disable the auto config function)
-----AT Bus Clock Select: CLK2/10
-----DRAM Read Wait State: Normal
-----DRAM Write Wait State: Normal
-----SRAM Read Wait State: 0 WS
-----SRAM Write Wait State: 0 WS
-----Cycle Check Point: Fast

Slow Refresh: 60u
Ext. Cache WB/WT Feature: WB
Shadow Cacheable: Disabled
Polling Clock Select: CLK2/3
DMA Clock Select: ATCLK/2

After only changing the values of SRAM Read and Write Wait States from 0 to 1, the computer works fine.

I don't know very well how these options are supposed to work (I have an idea but it's not clear to me). Maybe you can tell me a more appropriate configuration?
The default BIOS settings do make sense? Is it logical that it wasn't working OK before with the previous configuration? Does it seem reasonable that just changing those two values from 0 to 1 has such a great impact and now the Motherboard works seemingly fine?

Please discuss.
 
Just like in a roller coaster, one day it was fantastic and now, today, I have to give you sad news.
We are all mourning the loss of this motherboard. She suddenly and without notice, decided to visit the electronic heaven.
It was a nice ride though, we'll all miss her.
xD

PS: I still keep some faint hopes to bring her back from the dead some day perhaps. It will be harder this time. No POST, no anything. No POST error code.
 
Back
Top