Desoldered the keyboard connector, BIOS socket and SN74LS244N IC.

The solder joints on these components were highly oxidized so it ended up being a royal PITA to get them off. Normally this would take about 10 mins, but this took the better part of an hour.

Fortunately, I managed to get everything off without losing any pads.

The parts have varying degrees of damage. The socket was so corroded in that one corner that one pin came completely apart when removing it. Fortunately the SN74LS244N chip has all its pins, so I should be able to reuse the chip. I'll just sand off the corrosion and re-tin the legs.

I then did some probing for any bad traces. While there is some exposed copper, I didn't find any broken traces yet. So I went ahead and applied some solder mask to protect the exposed copper bits.

I also poked around the ISA slots to see if there were any issues. One of the pins in the second slot (8-bit slot) turned out to be completely loose and came right out. It was the most heavily corroded of all the slots, so I imagine I'll just end up replacing the whole slot.

The missing pin shouldn't prevent the board from working, unless there is a broken trace underneath.

Tomorrow I'll solder everything back in including a new socket and keyboard connector, then test out the board.

>So I went ahead and applied some solder mask to protect the exposed copper bits

Im afraid all the discolored blackened traces under solder mask need grinding down. The discolored tracks go down all the way to keyboard controller. They will keep corroding. Same for all the VIAs around the broken socket, if they arent broken now they will in time.

rasz_pl wrote on 2024-11-20, 07:02:

>So I went ahead and applied some solder mask to protect the exposed copper bits

Im afraid all the discolored blackened traces under solder mask need grinding down. The discolored tracks go down all the way to keyboard controller. They will keep corroding. Same for all the VIAs around the broken socket, if they arent broken now they will in time.

That's a good point. I did end up going ahead this morning and re-soldered components, at least to do some initial testing. I figure if the board has other faults, I don't want to spend too much time on grinding solder mask and re-coating traces if the board wasn't fixable for other reasons.

If I can confirm the board is working, I'll probably remove various components to clean up all the traces better. I'll likely need to remove the keyboard controller IC, plus I know I have to replace that one ISA slot.

(I'm also probably procrastinating doing any more desoldering given how annoying that is with the corroded solder. 😅 )

I went ahead and added a new keyboard port and BIOS socket, along with the existing SN74LS244N chip. I know I'll probably need to re-remove them at some point, but it should be a lot easier the second time around due to the fresh leaded solder.

This board came with a TI 486DLC-40 processor, so I'll be using that to test. It supports both the 386DX-33 and the 486 DLC-40.

We actually had a 486DLC-40 growing up (our second computer), so I am invested in getting this board working. I might use it to try to recreate that particular PC, assuming everything works.

Upon plugging in a POST card and testing, it did have some life!

POST codes were a little bit erratic. Mostly it stopped at 13:12. Some times it would land on other POST codes (e.g. 42, C1, FF, and others). Some times it would also look like it was resetting and cycling through the same codes over again.

At least this is a sign the processor is doing something.

As I was testing, I did run into an odd issue with the 12V power. I was moving the POST card between different slots. I noticed the 12V LED started to dim, and then eventually went out altogether.

I probed the 12V rail on the slots and they everything seemed to be connected, so I don't think it's a broken trace. I also checked the PSU and made sure it was outputting 12V, and it was.

Then I did some probing while powered up. If I measured the 12V rail from the top of the slot, it was reading ~12.5V. But if I measured from the solder joint on the underside of the board, it was only reading 1.5V. If I measured the 12V rail at various points down the line, it was resulting in the same 1.5V.

Some of the solder joints on the power connector were oxidized with the 12V one appearing to be the worst. I went ahead and re-flowed them. After re-connecting and retesting, it was now reporting 12V.

I'm not sure if the oxidized joint was somehow acting as a resistor due to a poor connection or if this was just coincidental. Unfortunately I hadn't measured the 12V rail until after the LED stopped lighting up, so I don't know if the voltage had been progressively dropping. I've never had an oxidized solder joint act this way, but I suppose if it had poor conductivity, that could explain things? Not entirely sure at this point.

Did a bunch of testing this evening trying to determine what was stopping this board from booting.

The POST codes continued to act inconsistently, sometimes putting out the usual 13:12, but a lot of the time spewing out random POST codes. I also wasn't sure if corrosion in the slots might be interfering with the card, so I kept moving things around from slot to slot. Nothing changed though.

I also tried a variety of memory sticks to see if that would result in any changes, but it didn't. I further tried removing the cache and trying the 32kB option instead of the full 128kB. Still no change.

I then went back to probing the board, first checking traces and voltages. I didn't find anything initially.

Finally I decided to try probing the crystals. While the 14.3MHz crystal tested fine, the 32.8kHz did not. It produced a noise-type signal instead of a clean waveform.

At this point I suspected the crystal might be bad, so I started probing around the crystal to see if there was anything else. That's when I discovered a broken trace between the crystal and the BIOS chip. Despite my eariler testing, I somehow missed probing this one length of circuit and of course, it's the part with fault.

I then soldered a piece of wire to bridge the broken part.

After doing this and re-testing, the 32.8kHz crystal tested fine.

At this point, I plugged in a keyboard and video card and gave it another try... and success! The system booted!

I still have a lot more testing to do. I suspect the corroded ISA slots may prove to be less than reliable. And there still could be other faults on the board that have yet to manifest themselves. I'll know more once I spend some time testing the system.

edited to add:

I spent another 10 minutes or so testing it. Unfortunately, it did lock up a couple times. Once in Wolfenstein 3D and again in 3D Bench. CPU seems like it was getting pretty hot, so it might be thermal related. I did not have a heatsink or fan on it when testing. I'm going to let it cool down and then add a heatsink and fan before testing further.

nice.... Mr motherboard Poirot

nuno14272 wrote on 2024-11-21, 11:44:

nice.... Mr motherboard Poirot

Thank you, I appreciate the sentiment. But it might be a bit premature, as the board has returned to a non-working state. 😒

This evening proved to be quite the saga with this board.

As reported yesterday, the board was working and booting. I spent a couple hours testing it this evening. Mostly it seemed to work, but there were some notable issues.

First, I had some random lock-ups. Initially it seemed to happen every time I tried de-turboing the system and running 3D Bench. It would run fine full speed, but de-turbo it would freeze after a few seconds. But then weirdly after more rebooting and card juggling, it started working.

Second, I got these HIMEM.SYS errors when booting. The reported memory addresses were never consistent. Initially I thought this was due to the math co-processor (ULSI DX/DLC 4o). It seemed to be only happening when I had the math co-pro installed. However, it did eventually work with the math co-pro. I was able to benchmark the system with it installed and noted I got slightly higher 3D bench scores than without it.

On top of that, I did them remove it again and got the same HIMEM.SYS error without the math co-pro installed. Doing a hard reset and the error went away, but it was not entirely consistent in how this error occurred. (I also tried several different sets of 30-pin memory, but they all had the same issues.)

Third issue is I noticed that some times the POST card would not display the 3.3V. Now the fact it is displaying 3.3V is weird to me, since ISA slots don't have 3.3V. Maybe it's using that same LED to indicate -5V? Regardless, I noticed that on the first and third slot, it would be intermittent. I suspect this is due to the corrosion of those slots.

It was at this point I powered everything down, unplugged everything, and gave the ISA slots an additional scrub with some Deoxit.

I then powered the board back up and decided to test some voltages just to make sure they were reading normal. All voltages seemed to be reading okay.

After which I plugged everything back in and the board wouldn't start. The POST card was showing the dreaded "--:--".

After a few more attempts at removing and re-seating everything, I decided to swap the BIOS chip. I was using the original BIOS, but I had flashed a spare when I was originally trying to get the board going. The replacement chip did result in some POST codes being displayed, but not a full boot. It went through a bunch of random POST codes, got stuck reset loop once, once made it all the way to 41:36, but mostly got stuck at 13:11.

One weird symptom was that with RAM installed, it took forever to go through each POST code. But if I removed the RAM, it almost immediately would jump to 13:11.

I then decided to try a different processor, a 386DX-33. This one got all the way to 41:36 consistently, but still not a successful boot. I also got different beeping, sometimes a high-pitched continuous tone, some times a continuous beep, and some times the video card not found beep (with POST code 31).

I don't know exactly what is going on but I have a few theories:

1. Scrubbing the ISA slots triggered some additional damage, possibly a bad trace that is now broken.
2. I possibly inadvertently shorted something when probing the ISA slots. I thought I was being careful and didn't do anything silly like jamming my probe in between the pins. But maybe I slipped and didn't notice.
3. Some other component was already failing and this is just a symptom of that failure.

I am going to test all my 30-pin RAM and the cache chips in another system to see if they have any issues. Unfortunately I don't have another comparable board to test the processors. I usually take POST codes as a sign the processors aren't bad, but I can't rule out they don't have issues.

As for the board, I feel like it's back to square one. I need to re-start diagnosing this board. Testing traces, shorts, voltages, logic circuits and so on.

This motherboard continues to confound me. I went to do some more testing and when trying to power it up, it now gives a different set of POST codes, ending in 03:01.

With the 386 DX-33 processor it beeps a few times while cycling through a few different codes, then makes a noise similar to a memory test, before landing on 03:01.

With the 486 DLC-40, I don't get any beeps, but it cycles through a few POST codes and then also lands on 03:01.

I did some further testing of the original BIOS chip. I tried reading it with my programmer and it read just fine. The checksum matched the BIOS checksum for the latest version from the RetroWeb.

I then tried erasing it and reprogramming. However, when trying in the motherboard, it still results in a "--:--". I don't know exactly what happened to the BIOS chip, but I find it odd that it works fine with the programmer but not the motherboard.

I did some re-testing of all the clock signals just to make sure they were still working. This includes the ~14.3 MHz and 32.8kHz crystals, plus the main CPU clock.

I checked the jumpers to test both 80MHz and 66MHz and both worked fine. Additional jumper settings provide 50MHz and 40MHz, although those are not specified in the manual.

Its time to take care of all the corroded vias

rasz_pl wrote on 2024-11-24, 01:09:

Its time to take care of all the corroded vias

That's probably a good idea. I think I've been procrastinating because of the desoldering involved. 😅

Yeah, there are a lot of bad vias seen on the photos. It is gonna be a pain to deal with all of them...Also all the darkened spots on the traces will need attention since those areas will grow bigger and will disconnect sooner or later.

Started desoldering everything to do a more thorough job of inspecting and repairing any broken traces/vias.

So far I've removed the first two ISA slots, the keyboard controller, BIOS socket, a couple capacitors, jumper JP5 (which is pointless on this board), and the SN74LS244N IC. Oxidation of the joints near the battery area did not make this the easiest, but more of the work is done. I still plan to remove the third ISA slot.

Had a few resistors fall out while removing the ISA slots, but I suppose that's the consequence of using hot air for something like this.

I have started removing more of solder mask on the corroded traces, but based on continuity testing, haven't found any other broken traces yet. There also doesn't appear much damage underneath the ISA slots. The corrosion seems to have been largely confined to the ISA slot pins themselves.

Did some more desoldering and I think I've now removed everything I need to remove.

Removed components include:

• ISA slots 1,2, 3
• BC10, BC19
• C27
• RP2
• R4,R5,C16
• R35,R26,R16,R23
• JP5, JP9
• keyboard connector
• keyboard IC
• BIOS socket
• octal buffer (SN74LS244N)

I scraped off a lot of the damaged solder mask and discovered a couple more broken traces under the resistor array (RP2). One appears to be for the power good line to the UM82C493F chip, while the other connects an address line from the 30pin memory to the octal buffer (SN74LS244N).

I tested a bunch of the vias that are underneath the main power connector and memory sockets. I'd rather not desolder those if I don't have to, so I wanted to make sure all those vias (among others) were all connected. After tracing things, they all appear to be fine. The vias with the blue circles all connect to the main UM82C491F chip.

Well done! Desoldering skills are top notch.

Nexxen wrote on 2024-11-26, 00:44:

Well done! Desoldering skills are top notch.

Thank you very much! 😀

Good job, so far. Those corroded traces need a very good cleaning, tinning and covering with new solder mask. I think you have to "sacrifice" some silkscreen markings (and more old solder mask area) in order to restore them correctly, otherwise all your work for removing the elements from the PCB will be pointless. The end result will not be pretty to look at, but after all, the board won't be going to beauty pageants, so getting it back in working order is way more important than its good looks 😀

analog_programmer wrote on 2024-11-26, 10:33:

Good job, so far. Those corroded traces need a very good cleaning, tinning and covering with new solder mask. I think you have to "sacrifice" some silkscreen markings (and more old solder mask area) in order to restore them correctly, otherwise all your work for removing the elements from the PCB will be pointless. The end result will not be pretty to look at, but after all, the board won't be going to beauty pageants, so getting it back in working order is way more important than its good looks 😀

I am working on the polishing and tinning process now. I'm taking my time with it, since ideally I want to remove as little good solder mask / silkscreening as possible. And I do plan to try some things with applying the solder mask, to try to get as good a result as possible. Even though some areas will be hidden under components, I do want the exposed spots to look half decent. 😉