You could try cleaning the RAM sockets with some isopropyl or ketonatus alcohol and a soft tooth brush. Also check for any damaged pins in the RAM sockets, although I don't expect this to be the issue, since the system sometimes does work.

Since it doesn't work on the first attempt, but it does work on the second attempt, it kind of sounds like a capacitor problem. I'd try looking at the motherboard and see if you see any bulging or leaking capacitors. If it is SMD capacitors, look if there is junk underneath them.

My guess would also be capacitor and voltage related. Seems that warming components up causes more stability, which is a common failure mode of bad caps.

Agree is points to caps. Also besides the temperature thing the caps could be electrically leaking so after the first or second power on it sort of heals them so they function close to expected instead of not filtering/semi shorting/etc when first powered on.

I should note that even after being on for a long period of time, it doesn't boot reliably or anything. I wish I could tell if there was a bad cap somewhere, but visual inspection of the board is clean. I don't know where I'd start. I also noticed that it has a way higher chance of not booting when all three slots are populated.

Strahssis wrote on 2022-05-10, 20:31:

You could try cleaning the RAM sockets with some isopropyl or ketonatus alcohol and a soft tooth brush. Also check for any damaged pins in the RAM sockets, although I don't expect this to be the issue, since the system sometimes does work.

Since it doesn't work on the first attempt, but it does work on the second attempt, it kind of sounds like a capacitor problem. I'd try looking at the motherboard and see if you see any bulging or leaking capacitors. If it is SMD capacitors, look if there is junk underneath them.

Just tried cleaning the slots with compressed air and isopropyl alcohol, but no dice. I don't see any damaged pins in the sockets either. It is happening with all three sockets, so I'm guessing it's not an actual slot issue at this point.

Did you clean the DIMM's edge connector ? Use paper towel with ISO alc....and do not be afraid to scrub hard ;p

Horun wrote on 2022-05-10, 23:57:

Did you clean the DIMM's edge connector ? Use paper towel with ISO alc....and do not be afraid to scrub hard ;p

I did actually. Thanks for your suggestions though.

I took a closer look with better lighting after removing some components. Y'all were right on about this being the issue. I found some of these "Chhsi" branded caps that were part of the known bad caps from that era. I am a bad solderer, but will try to get these replaced, and report back.

rootinit wrote on 2022-05-11, 01:37:

I took a closer look with better lighting after removing some components. Y'all were right on about this being the issue. I found some of these "Chhsi" branded caps that were part of the known bad caps from that era. I am a bad solderer, but will try to get these replaced, and report back.

Just as a side note, Chhsi's web site did not change much between 2001 and 2004 (their corporate blurb mostly says they started in 1977, but at least one place mentions 1997, which I presume is possibly a typo) , but their contact info went from featuring an @chhsi.com.tw address to featuring an @yahoo.com.tw one . Infer what you will from that . 😉

https://web.archive.org/web/20001019005401/ht … w.chhsi.com.tw/

https://web.archive.org/web/20040727051541/ht … w.chhsi.com.tw/

Anyway, specs are here for their various capacitors series, if you need them :

https://web.archive.org/web/20010215170329/ht … ducts/index.htm
https://web.archive.org/web/20040806160321/ht … ducts/index.htm

EDIT : The WG series series were apparently arguably their "top of the line", which isn't saying much... Did WG stand for "Wasteful Garbage" ?

darry wrote on 2022-05-11, 03:39:

Just as a side note, Chhsi's web site did not change much between 2001 and 2004 (their corporate blurb mostly says they started […]
Show full quote

rootinit wrote on 2022-05-11, 01:37:

I took a closer look with better lighting after removing some components. Y'all were right on about this being the issue. I found some of these "Chhsi" branded caps that were part of the known bad caps from that era. I am a bad solderer, but will try to get these replaced, and report back.

Just as a side note, Chhsi's web site did not change much between 2001 and 2004 (their corporate blurb mostly says they started in 1977, but at least one place mentions 1997, which I presume is possibly a typo) , but their contact info went from featuring an @chhsi.com.tw address to featuring an @yahoo.com.tw one . Infer what you will from that . 😉

https://web.archive.org/web/20001019005401/ht … w.chhsi.com.tw/

https://web.archive.org/web/20040727051541/ht … w.chhsi.com.tw/

Anyway, specs are here for their various capacitors series, if you need them :

https://web.archive.org/web/20010215170329/ht … ducts/index.htm
https://web.archive.org/web/20040806160321/ht … ducts/index.htm

EDIT : The WG series series were apparently arguably their "top of the line", which isn't saying much... Did WG stand for "Wasteful Garbage" ?

Interesting stuff. Thank you. I've been poking around a bit more online, and see that these particular ones were failing as early as 2006. The other side of the caps had some specs for me to use, so I ended-up buying these Panasonics as replacements. I'm glad that all the other caps IBM used seem to be Rubycon and Nichicons.

Problem appears to be solved! New Panasonic caps are swapped-in, and I ran a few rounds of Memtest86 perfectly.