R59 is 10 kohm (marking "103" means 10*10^3 ohms)
The other resistor I can't read but the same value as R59 is a good bet.
The capacitors are unmarked but a starter value to try is 100nF (0.1uF)

All are 0603 size.

stamasd wrote:

R59 is 10 kohm (marking "103" means 10*10^3 ohms) The other resistor I can't read but the same value as R59 is a good bet. The c […]
Show full quote

R59 is 10 kohm (marking "103" means 10*10^3 ohms)
The other resistor I can't read but the same value as R59 is a good bet.
The capacitors are unmarked but a starter value to try is 100nF (0.1uF)

All are 0603 size.

thank you for that, I will source some and attempt a repair

If I had to guess I would be inclined to think that the "shady looking" resistors and caps have nothing to do with the issue. I wouldn't really expect this kind of result with the shady componets location on the board in mind.

If cleaning the board thoroughly isn't enough I would try to remove the cache and jumper the board for 0KB cache and see if that solves the issue, I would also try to see if some other memory sticks perhaps work better. In my experience these kinds of intermittent memory / BIOS issues are most of the time memory or... BIOS related so reseating the BIOS chip would also be on my short list before I would start messing with replacing SMD components. Double checking jumper settings would be yet another thing to try.

EDIT. I missed that you already cleaned the board so scratch that but still try to remove any rust with WD40 or similar, reseat all sockeded chips, try without the cache and so on before you try the harder time consuming stuff.

I tried removing the cache already but the lowest jumper setting is for 64k no setting I can see for 0.

After I cleaned up the sticky liquid which wasn't really that much but around the resistors that I circled it no longer posts.

I also did take off the bios chip and reseated it with no results.

I tried 3 different 72pin ram chips and also a few 30 pin chips separately ofcourse. I would get memory parity or offboard parity errors. Few times the system also locked up in bios.

I also disabled external and internal cache in bios. No effect

Im low on ideas after that, ofcourse I want to try easy things before soldering small components.

I am thinking that perhaps when some of the sticky liquid was there it was acting as a conductor between one of the charred resistors or transistors so when I cleaned it off the connection is gone? Just a theory

I doubt that C20, C21, C32 or C33 has anything to do with the issue unless shorted and perhaps you can measure R49 and R59 in circuit?

Perhaps you can post an image showing the full board (it seems to have vanished as I could swear I saw one before) so we can see where the trace going through R49 ends up?

yeah you are right after cleaning it only r49 looks suspicious at best. I just had 1 more idea, tomorrow when my energy returns I might carefully run a bit of a heatgun at 360 degrees Celsius around the components where the drink was, carefully not to melt any plastic. I will also have a look where r49 goes.

In the photo it's jumpered to run at 50MHz. Did you try a lower speed? Maybe try a different video card too, especially if it is a VLB one. I went round and round with a vlb board one day, trying to get it to boot without freezing, and when I finally swapped video cards it was fine.

thanx for the tips but unfortunately this is how I ended up on 50mhz...
started with 33 on overdrive 75 = booted had parity errors
put in amd 133 on 33 = booted = booted had parity errors
tried amd 133 on 40 = booted had parity errors
so I though try a older cpu with less interval care, tried a few 486sx at 25 = didn't boot
tried 486dx50 at 50 = didn't boot
cleaned the board
tried overdrive/amd133 at 33 = didn't boot
tried 486dx50 at 50 = still not booting

also during my testing I have been using 1 vlb card and 1 isa they both showed video at one stage

so by now
- Used 2 different video cards 1 isa 1 vlb
- used 4 or 5 different cpu (only 2 worked, both with voltage regulators, coincidence?)
- used 3 different 72pin ram chips
- used 4 different 30pin ram chips
- disabled internal+external cache in bios
- removed cache physically (but no jumper setting for 0 cache)
- removed and reinstalled bios chip

Going to leave it for the rest of the day and not rush maybye some ideas will pop up, then ill try heatgun

So today I fired it up again with a overdrive 75 and vlb card.
Same memory mistmatch error
swapped ram with more modules and got it to work with 1 72pin 8mb stick

no idea why says 100mhz cpu when it's 75 but not important atm.
powered on and off a few times looks stable, went bios no crashes etc.
Install a hard disk controller and error is back. remove hard disk controller error still there. try more ram still stuck in the error.

pressing F1 gives me this...

So. can't enter bios.
When the system works it says total memory 8.xxx on post
when the system gives me memory mistmatch it says 7.xxx detected, almost looks like its not detecting the 640kb base memory when the problem happens.

it looks like when pressing the motherboard to install ram or isa card makes things better/worse

here is a few pictures when I had memory mistmatch error was able to enter bios but it would hang

Could be cracked/cold solder joints, or a broken via somewhere if you're really unlucky.
Place the board on a flat, hard surface, apply pressure with your thumb to the UMC chips, power the board while doing this - see if anything changes.
I would also run the board cache-less until getting to the bottom of the problem(s).

I tried this with both the umc chips, when pressing the small one with my thumb the system does not post at all, after taking my thumb off it posts on the next power on. I tried it also on the isa slot on top of the small umc chip and also the same behaviour.

I had a close look at the pins they look ok but like you said hopefully cold solder joints. Do you think running a heat gun around the umc chip connectors would be the best way to try remelt the solder?

I've never used a heat gun so I cannot tell, but one of my nicer 386 boards had this problem - erratic behaviour at POST, hanging with different code every time, sometimes almost got it to boot. Turned out it was due to cracked solder joints around the TACT83443 chip. Used this technique, not without worrying because it was the first time, but turned out perfect. It is essential to use plenty of flux and use a tip like the one in the video. Also take care not to apply too much pressure when dragging the tip across the pins, they need to stay aligned with their pads.

I think its definitely this chip, I did the solder reflow like in the video, as long I had enough flux it stuck exact to the right pads. Had to do it 2 or 3 times before the system actually booted!
However it is still shaky, I tried doing horizontal and vertical and keep the iron tip for 2-3 seconds on each pad to make sure it flows well but in the end it always comes back to the same errors. Especially after changing cpu or adding new hardware which flexes the board.
I guess keep trying right? Maybye ill try out the air gun, soldering iron feels like a very classy and proper way to do it tho. I need a break for today however.

Oh yeah I also reflowed the pins on the cpu connector from the back of the board but that doesn't seem to have any effect.

I was running the 75mhz cpu at 33mhz so set it back to 25

System ran stable and got the furthest, but when I changed cpus back to parity errors

im not very experienced at soldering, does any1 see any major problems or cold joints with what I done?

Soldering on the pins looks good enough. I am willing to believe some SMD resistor or capacitor is cracked or damaged. I see several on the photos with darkened edges which is sign of potential corrosion damage, on any of my machines I always replace those resistors even if they measure good and usually no longer have problems afterwards.

I am thinking it's a bigger problem like w cracked via somwhere or yes one of those smd resistor or capacitor.

I reflowed the big umc chip and manually reflowed each pin on cpu socket with soldering iron. Also I reflowed the solder on most the small resistors and capacitors around the umc chips and cpu but its still random behaviour.

Going to give it a rest, I will try reflow a few of the through hole chips around the cpu/chipset but low chance of a fix. then ill leave it for a few weeks until I get inspiration to work on this again.

thanks everybody for input

Just some inspiration for when you feel like working on this board again.
On a board of mine that I was never able to resurrect (Socket 3, EXP8449) I noticed that the contacts inside the socket have lost their 'springiness' - perhaps the issue was compounded by me inserting countless CPUs in there trying to get it to POST... I solved that by dismantling the socket (the upper cover should come loose, exposing all the contacts) and then gently tilted each one with the tip of a very fine screwdriver as to make better contact when the lever is pushed.
It was in vain of course, but at least I knew the CPU was making great contact with the board 😁

thanx for the suggestion I took the top of the socket off and retightened all the pins which looked fine, unfortunately no fix 🙁

I also found a few same value smd resistors on a broken video card so I transplanted a few onto the ones in my first post where the suspected coke was.

The bios no longer crashes and I am now stuck in a repeatable pattern.

1. Insert intel overdrive cpu
2. boot system
3. 8192mb detected for 1 stick of 72pin 8mb.memory
4. bios values not set.
5. Enter bios save values.
6. reboot system with saved values now memory shows 7488 exactly 704kb less
7. Get memory mistmatch error press F1 to proceed (enter bios)
8. after F1 always mistmatch offboard error, when try enter bios or boot after bios values saved on first boot.

I tried many different memory sticks, and some of them did work when I got the board to boot at one stage (its not the ram or cpu)

so it looks like something is reporting the. memory wrong to the bios, but only after bios has values saved, maybye its the bios, maybye the error parity circuit, I don't even know where that is.

Apart ftom this probably changing every smd resistor and capacitor could fix this like mentioned before but I don't have all of them and all the capacitors don't have ratings on them.

I think this is where I give up for now

It has been a while but I have been revisiting this motherboard in the last few days with a fresh mind since I packed it up what looks to be about 4 months ago.

I actually thought I fixed it when I finally got a uv eraser and erased the bios and wrote it again from a backup I took using my 866 external reader/writer.

The board was posting fine with alot of different cpus no parity errors but it would randomly enable L2 cache on some boots and disable it on others, not happy with that

After more cpu and ram swapping parity errors again, so I left it alone.

Got some eeprom chips so I programmed one from the original backup I took and still parity errors, ok rules out the bios chip unless my original backup is corrupt.

I noticed that sometimes it shows conventional memory in the bios as 64kb other times 640kb.

This is my next angle of attack, what could be causing the bios to see different conventional memory sizes?

conventional memory is the first 640kb reserved from the ram? I have tried many ram chips both 32 and 72 pin (not all at once) so its not a problem with the ram chips and the ram slots.

I am now suspecting it could be the Chipset, could a corrupt/faulty chipset misrepresent and give faulty ram/conventional memory readings to the bios?

I would be very excited to try remove and install a new chipset if I can even find one if it is in the right direction.

My logic is the motherboard boots - > memory is detected - > memory is tested - > false size is reported to bios - > bios detects parity errors

now the missing link is what hardware part could give corrupt readings of the memory?

This is just an educated guess but I'd say your problem is one of the address or data lines that go to the SIMM sockets is bad. Not completly open but scratched, or a cracked via maybe, so it doesn't play well with high frequency signals.

BTW the damaged caps on the mobo should not be an issue, some are just glitch filters for keyboard and have very low value anyway (some 47pF or so), could be missing completly. The resistor is most likely pull-down for the POWER GOOD signal and this one is kinda important. But, depending on the PSU, mobo might work even if if it's completly open. And the issues you have don't really point to it being the source of problems.

Since you also have L2 cache issues, the actual problem is probably the UMC chip next to both SIMM slots and cache chips, inspect it, resolder. Don't need a heat gun, it can be done with flux using a decent soldering iron with flat tip. Slowly swipe over the pins. If you get a short on the pins, use more flux, if that doesn't help there's always solder wick. It's a much safer method since there no risk of metling any nearby plastic parts.