The next problem is that half the jumpers were missing, and the rest didn't correspond to the 386DX-40; it won't power on and all the chips are cold. I've since jumpered it correctly for the processor and clockspeed (it has a programmable clock generator, not a regular DIL can), now when I power it on at least the processor and a few other chips get warm. Still no video output and no beep codes.

I've got a ISA POST diagnostic card arriving tomorrow, but I'm a little disappointed that despite rectifying all of the obvious problems I don't even get error codes from it 🙁

A hole inside 74LS245 may indicate that "magic smoke" (tm) flew away...
Too high voltage or bad polarity 🙁 I have bad feeling about this board.

zyga64 wrote on 2021-03-11, 10:49:

A hole inside 74LS245 may indicate that "magic smoke" (tm) flew away...
Too high voltage or bad polarity 🙁 I have bad feeling about this board.

That's a distinct possibility.

Good work so far! I love projects like this.
Just some questions that come to mind:

Did you doublecheck all damaged traces with a multimeter and did you find all (even slightly) damaged traces, even hidden ones?
Is there corrosion on the flipside?
Are there any deep scratches?
Are there any loosened pins at the larger chips?
Are all the cache chips o.k.?
Have you tried different RAM sizes / types?
Do any chips get warmer than usual or even hot?
HAve you tried some alternative BIOS chip?

Fortunately, the only damaged trace was from the battery +ve tab, all others that had corrosion have cleaned up and still have continuity:

There was one damaged trace when I removed the 74LS245N between ISA slot 2 and 3, it connected pin 2 of ISA slot 2 to pin 3 of the 74LS245 and then to pin 2 of ISA slot 3. I've bridged these using a (temporary) fix on the underside of the board. This was almost certainly caused by my removal of the cracked chip, and not present previously. Underside of the board is very clean with no scratches or damaged traces.

The jumpers were all over the place though - it has 128KB of SRAM fitted, but the jumpers didn't match it, and neither did the cpu type. There were no jumpers on the clock generator, which I guess could be one reason why the cpu didn't initially warm up (it does now, configured for 40MHz).

Tried with/without memory - but no beep codes either way. I don't (yet) have another 386DX bios rom, but will see if I could track one down and burn to an eprom.

With the clock generator running the Am386DX gets warm, as do the cache chips and several of the 74-series logic IC's next to the SIMM sockets, warm, but not uncomfortably so. The rest, including chipset, are cool to the touch after 5 minutes or so.

I identified your board http://www.win3x.org/uh19/motherboard/show/3600

Keep the good work !

Maybe you can use a diagnostic tool to see if the board does anything at all ?

POST card arrived:

+12v, +5v, frame are all lit. IOrdy flashing.

Clk, -12v all unlit. No codes.

At least here should be some oscillator / (rt)clock frequency present:

Do you have an oscilloscope or logic analyzer available?
Or (if you can find a datasheet, I couldn`t) you could measure the DC levels.

-12V missing -> maybe there´s still some trace interrupted between the PSU connector and the 1st (8bit) ISA slot. Or a shorted tantalum, if the PSU just shuts off the -12V rail in this case.

If the board has been exposed to reverse polarity any of the chips could have been damaged. The larger "chipset chips" seem to have some protection, they often survive. So I would focus on the birdseed first.

I guess if someone has plugged the power cables in the wrong way round previously, that could explain the missing -12v led and probably the missing CLK signal.

I'll see if I can trace where -12v stops and see if I can see the ISA clock signal anywhere (I've a cheap Hantek USB scope that should be just fast enough at 25MHz to do that).

Edit: not going to spend ages on this, as I've got several other "scrap" boards to clean and examine from the same site. If only one of them works I'll be happy.

Can you please make a good photo of the board for UH19 ? (The one you've uploaded has a watermark from the seller)

So it turns out that the POST card was broken:

Yep. I didn't spot those two 'cut' leds. They are both cut at the same height and angle, so can only assume that whatever assembly is done with these things in China was miscalibrated or something, and there's a whole load of the cards where the last two leds are chopped off.

Anyway, I of course repaired it before doing anything more:

With that done, and retested, the Forex board does seem to have the full range of signals (+5v, +12v, -12v, CLK) lit on the board and steady. IORDY is steadily flashing. Shorting the motherboard reset header does show on the card, too. There's still no codes shown however:

Image of the board:

That can't be confirmed so far because this board has been through lots of things but I'm starting to see a pattern here ...
I wonder if FOREX chipsets don't have a defect actually making them die quicker than chips from the same era. Has anyone a working forex 386DX chipset in 2021 ?

Also, here's the BIOS:

It's a M27C512 device. AMI BIOS, as expected.

The BIOS ID string appears to be in there:

00008000 28 41 41 4d 4d 49 49 42 42 49 49 4f 4f 53 53 29 |(AAMMIIBBIIOOSS)|
00008010 30 36 2f 30 36 2f 39 32 28 43 29 31 39 39 32 20 |06/06/92(C)1992 |
00008020 41 6d 65 72 69 63 61 6e 20 4d 65 67 61 74 72 65 |American Megatre|
00008030 6e 64 73 20 49 6e 63 2e 2c 20 41 6c 6c 20 52 69 |nds Inc., All Ri|
00008040 67 68 74 73 20 52 65 73 65 72 76 65 64 00 00 00 |ghts Reserved...|
00008050 28 43 29 31 39 39 32 20 41 6d 65 72 69 63 61 6e |(C)1992 American|
00008060 20 4d 65 67 61 74 72 65 6e 64 73 20 49 6e 63 2e | Megatrends Inc.|
00008070 2c 00 00 00 00 00 00 00 34 30 2d 30 31 31 37 2d |,.......40-0117-|
00008080 34 32 35 30 30 34 2d 30 30 31 30 31 31 31 31 2d |425004-00101111-|
00008090 30 36 30 36 39 32 2d 46 4f 52 58 00 00 00 00 00 |060692-FORX.....|
000080a0 32 33 4c 2d 30 2d 30 30 30 30 2d 30 30 2d 30 30 |23L-0-0000-00-00|
000080b0 2d 30 30 30 30 2d 30 30 2d 30 30 2d 30 30 30 00 |-0000-00-00-000.|
000080c0 00 00 00 00 00 00 00 00 32 34 48 2d 30 2d 30 30 |........24H-0-00|
000080d0 30 30 2d 30 30 2d 30 30 2d 30 30 30 30 2d 30 30 |00-00-00-0000-00|
000080e0 2d 30 30 2d 30 30 2d 35 00 00 00 00 00 00 00 00 |-00-00-5........|
000080f0 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 | |
00008100 41 4d 49 42 49 4f 53 20 28 43 29 31 39 39 32 20 |AMIBIOS (C)1992 |
00008110 41 6d 65 72 69 63 61 6e 20 4d 65 67 61 74 72 65 |American Megatre|
00008120 6e 64 73 20 49 6e 63 2e 2c 20 20 20 20 20 20 20 |nds Inc., |

So it appears the unique ID is: 40-0117-425004-00101111-060692-FORX

Deksor wrote on 2021-03-12, 21:39:

That can't be confirmed so far because this board has been through lots of things but I'm starting to see a pattern here ...
I wonder if FOREX chipsets don't have a defect actually making them die quicker than chips from the same era. Has anyone a working forex 386DX chipset in 2021 ?

I've seen a few posts with dead Forex boards for no obvious reasons, so the thought has crossed my mind.

Quick check with the scope and it reckons the 32KHz timer and the main 14.318MHz can are both outputting their respective clocks. The scope won't get anywhere near the main system clock though - it's just noise - so I can't be sure it's running at the jumpered 40MHz.

Great, everything's been added to the base 😀

Crossed my mind too about Forex having shorter lifespan than the rest.

Board looks pretty beaten down.
Hope you can awaken it.

Might be a good idea to decap one of these to see what's up in there ? (and perhaps find a pin compatible chipset that could be used to fix these boards ??)