Haven't gone through the whole thing yet, but datasheet for the W24159AK in the TAG RAM slot says it's 16k x 8, when the MicroHouse sheet for the board only lists 32k options for the TAG. Dunno if that'll matter.

[worth trying putting JP3 on JP4 to fix at 5V?]

Good eyes snufkin ! Yes would set the voltage to 5v instead of Auto.
And Yeah manual says same thing: TAG For 33MHz clock, the tag RAM 32Kx8 is used 20ns and 40MHz or 50MHz system, 15ns RAM is used (U2).
And it appears it does not matter what size cache is installed or else they would state so as in some other motherboard manuals....
Wondering if modding the BIOS to have Shadow disabled as default (or maybe it all ready is? per manual: "Shadow RAM is available as an option" ?) might help at least get the board to fully post but am probably wrong.

Thanks, I'll try to put a umc chip as the tag ram instead this evening (in about 5-6 hours). I'll see if that works. I don't have much hope though because it did that two years ago, and taking the cache off doesn't change anything (though maybe the board simply can't work without cache)

I tried it ... Unfortunately that did not work, still the same issue 🙁

Corroded / worn out contact(s) in DRAM sockets?
Loose solder joints at the chipset´s pins? (I had this VERY often recently)

Have you tried squirting contact cleaner into the CPU socket and DRAM slots?
I've ran into a few boards that acted extremely erratically, that a few spritzes of QD Electronic Cleaner magically made them behave.

Bother. I didn't spot anything else either. Does anything appear on a video card? C5 looks to be pretty far through POST so video should be up (I'm assuming POST stages are always step forward); if I'm look at the right list on that BIOS code website, it looks like basic RAM tests are done back in the 30s.

[edit:whilst I'm obsessing about it, there is a non-TAG cache config, with JP9:2-3 and JP12:12, using 4 64k chips, which gives at least one less point of failure]
[um, or not. That was from the MicroHouse sheet, but it isn't mentioned in the actual manual]
[I'm assuming you've cleared the CMOS, and it's a new battery? Can't see a specific jumper for it, but grounding JP35 pin 2 (battery source select output) should do it.]

Eep386 wrote on 2022-01-03, 18:08:

Have you tried squirting contact cleaner into the CPU socket and DRAM slots?
I've ran into a few boards that acted extremely erratically, that a few spritzes of QD Electronic Cleaner magically made them behave.

Yes, I've put contact cleaner everywhere several times. I also cleaned the pins with vinegar the best I could then cleaned them with alcohol and water.

snufkin wrote on 2022-01-03, 18:23:

Bother. I didn't spot anything else either. Does anything appear on a video card? C5 looks to be pretty far through POST so v […]
Show full quote

Bother. I didn't spot anything else either. Does anything appear on a video card? C5 looks to be pretty far through POST so video should be up (I'm assuming POST stages are always step forward); if I'm look at the right list on that BIOS code website, it looks like basic RAM tests are done back in the 30s.

[edit:whilst I'm obsessing about it, there is a non-TAG cache config, with JP9:2-3 and JP12:12, using 4 64k chips, which gives at least one less point of failure]
[um, or not. That was from the MicroHouse sheet, but it isn't mentioned in the actual manual]
[I'm assuming you've cleared the CMOS, and it's a new battery? Can't see a specific jumper for it, but grounding JP35 pin 2 (battery source select output) should do it.]

No, nothing shows up on the monitor using an ISA video card. I don't think there are many steps done before, I think I only see Cx codes. Nothing else.

On a side note, "I'm assuming POST stages are always step forward" gives me an idea. What if we recorded the normal activity for most AMI/Award bioses versions so we can know what's the next step to expect ? This could be very helpful for others !

Ok, I was assuming it was pretty far along as I thought shadowing was about speeding up performance once the system's up, which I wouldn't have thought would happen early on. Is it worth trying a different BIOS chip, there are 6 given on UR? Or a completely different BIOS from a board with the same chipset (HOT-433 has looks to have both AMI and Award), just to see if it gives a different fault?

Just read a bit more of that BIOS website you linked. The sequence for the Award Elite 4.51PG lists 08, which includes OEM chipset routines. Maybe it jumps from there to the C block since that's where OEM specific tests are, then jumps back to continue the rest, with video starting at 0D/0E.

Well I solved the problem 😀

The reason ? It's between the keyboard and the chair.

I checked the BIOS rom to see what I had put in the board just to make sure, and good thing I made that.

So I checked the POST string and this is what I saw : 09/18/97-i430TX-8679-2A59IO09C-00

i430TX ?! That can't be right.

Guess what ? My 2 years ago self got confused and have mistakend hippo 15 and rhino 15 🤣 Of course it can't work with a pentium bios ...

I'm glad I finally figured out this problem. The board seems to be capable of booting (I haven't tested that yet, but at least it's trying to boot !).
Now there might still be one last issue that will require more testing : the board complained about the battery being missing despite being present. I'll check that out to make sure it's not just a hiccup caused by the CMOS ram being erased (the battery didn't have a battery until recently).

D'ohhh!!!!

Been there, done that too. 🤣
Glad it wasn't a more major failure!

Ha. I'm sure with a bit more work they could have made a BIOS for the 430TX backward compatible with a UMC board. Slightly surprised you got any POST codes at all, although maybe that's just more indication that the Cx codes are chipset specific stuff.

Glad you figured it out, my next guess might have been a bad rom chip or something else unrelated 😀
At least you did not force flash a soldered ROM chip with wrong BIOS like someone did, I can neither confirm or deny that it was me 😁

Made some tests today. There was indeed a problem with the CMOS battery. Basically I needed to mess with the jumpers a little and also the arm that holds the battery was loose, so I tried my best to tighten it and now it works.

I've ran a couple of CPUs and benchmarks in the motherboard and it is rock solid 😁

So this thing is finally fixed 😀

The only minor issue it has now is that the socket's arm is broken in half, so it a bit hard to swap CPUs. But it's usable anyways. It's not worth the risk to remove the entire socket I think.

It's always so nice to see a 486 motherboard with a native 2032 battery holder instead of those leaky barrel batteries! Happy you fixed it!