Ok so I tried re-flowing the solder on the pins of the PLCC chips as well as replacing the missing capacitor C22. Agreed that there was so obvious visual damage. In fairness, it wasn't delivered in a box but more a reinforced padded envelope with pieces of cardboard hugging the card. I was sort of cringing opening the package possibly expecting it broken in two pieces but it appeared to be completely in-tact and not bent.
I also did some cleaning with isopropyl alcohol. The card had some residue over most of it. Fairly certain it was nothing to worry about but didn't hurt to try.

Anyway, no luck. In fact, now it's permanently reporting controller not detected. Nothing I can do with the jumper switches changes it. In fact, I can now directly conflict the I/O with the cache or the video and it doesn't cause the boot to hang whereas before it would throw an error for controller configuration error. It never hangs and you can enter the SCSI utility but do nothing of course. Regardless or whether or not it's even compatible with this computer, I'm fairly certain there's a bad connection somewhere or it's otherwise damaged in some form so I'm going to have to return it.

I'll probably try a different SCSI card. Good deals don't come up often online for them these days and it's not really something I'd find locally in the classifieds! If I do test another card, I'll update this so for future reference.

Thanks!

One of the 10 PALs might have died of age and resoldering finalized the process?

suddenly going dead after reflowing the PLCC is suspicious...check for tiny bridges under magnification

kingcake wrote on 2024-03-18, 06:09:

suddenly going dead after reflowing the PLCC is suspicious...check for tiny bridges under magnification

The evil thing about J-type leads (PLCC, SOJ) is that solder bridges can "creep below" the chip and thus be invisible. Using a continuity tester to find connectivity between neighbouring contacts might be required to detect bridges. I can reference on my copy whether any neighbouring contacts are intentionally connected. Also, I can check the BIOS for the algorithm used to detect the card, so one could perform basic tests of some functionality. On the other hand, it seems the card is currently fully unresponsive on the VL side, as the conflicts are gone - this is worse than just missing 8 out of 32 bits during bus mastering.

If the root cause is a trace that has been torn apart during mechanical mishandling while shipping, the reflow could have made it worse, for example if (non-conductive) flux got into the cleavage.

I wish I took a photo of the packaging before I opened it. It would have been sufficient for something smaller like say a PCI NIC but it definitely wasn't enough for something as long and thin as a VLB card. Way too much opportunity to flex if mishandled by the postal system. Between the potential for mechanical damage or simply one of the PLCC chips was dying and the heat from the soldering iron finally finished it.
I've checked each of the chips and not one pin has continuity with the one beside it.
I was pretty conservative when it came to re-flowing them. I set the iron to 300C (presume it's lead (Pb) solder anyway at this age) and just touched each leg for a second to see the solder melt momentarily. I didn't add any fresh stuff to the joints. Just wiped the iron every few legs with fresh solder and cleaned it off before resuming. It's been a safe bet for me in the past.
It's a fairly uncommon card to see these days online so while it would be nice to fix it, I can't justify the potential for failure and sitting with an expensive paper weight when I would rather move on and use that money for something I know works! Not sure I'd trust it even if I found a broken joint somewhere.