silvervest wrote on 2023-04-01, 09:43:

As the contact cleaner dries up though, it starts getting memory errors and eventually fails.

I have two ideas: Possibly it's not the contact cleaner that makes it work, but you actually have a temperature dependant fault, and spraying the stuff with contact cleaner cools it down. Broken traces that barely make contact are prone to temperature-dependent faults. Another thing to consider: You can clean the heck out of the modules, but that's not going to help if the contact pad has a hole at the point where the spring of the socket is going to touch. Such holes can develop from oxidation and scratching at the contact point. If that's the cause of your issue, you can try to cover up with conductive paint.

I'd take a look at the pins of the sockets themselves, see if they are soldered well, not grainy, dull or the hole half empty, touch up as required... if that didn't help, maybe I would retin all the pads on the SIMMs if they weren't gold. If they are gold, inspect with magnification.. if there's a pit through to copper, it could be growing verdigris oxidation off the bare copper which pushes up through the hole and lifts the contact from the edges of the hole... decide if you wanna get replacement SIMMs, or tin over all the gold contacts (might become a biannual thing)

Thanks for the thoughts, I'll have to have a good-er dive into it when my sanity restores.

I've tried (without hyperbole) 20-some pairs of SIMMs that have been fully tested elsewhere, and I can't find any good pattern to the failure. Some of them have gold contacts, some are tinned, and I've given all of them a run over with a melamine sponge to polish up the contacts...

I've checked the sockets under the microscope and noticed that some of the pins were indeed oxidised over, so scrubbed them up with my polishing pen, but still no dice. I did get it working for a while, and was able to run it through full Compaq diagnostics, and I thought everything was great, until I rebooted and it went back to the memory error. So frustrating!

Could it be power related?
Some bad electrolytics/bypasses that are bad enough to deliver required performance, but sometimes being "just ok enough" to support the ram which is cooled down and becoming less susceptible to the power noise?

silvervest wrote on 2023-04-02, 12:17:

I've checked the sockets under the microscope and noticed that some of the pins were indeed oxidised over, so scrubbed them up with my polishing pen

you cant polish a turd, you need to resolder dual simm socket, and if that doesnt work desolder it and look underneath for cracked traces

rasz_pl wrote on 2023-04-02, 13:51:

silvervest wrote on 2023-04-02, 12:17:

I've checked the sockets under the microscope and noticed that some of the pins were indeed oxidised over, so scrubbed them up with my polishing pen

you cant polish a turd, you need to resolder dual simm socket, and if that doesnt work desolder it and look underneath for cracked traces

This.👍 Use flux and some fresh solder to touch up all pins on the socket.
Good luck!

rasz_pl wrote on 2023-04-02, 13:51:

you cant polish a turd, you need to resolder dual simm socket, and if that doesnt work desolder it and look underneath for cracked traces

Oh for fu.... yeah... that was it. Reflowed all the pins and it's good now.

Like I said, I was missing the dancing bear. Thanks for the refocus!