Looking at your pic, do your top 4 SIMMs have their right-most memory chip missing? (so do these have 2 memory chips on the SIMM instead of the 3 memory chips your other 4 SIMMs have?)

If so, this might explain the parity errors you're getting.

I'd say, remove 4 of the 8 modules and see what happens 😀

Theses modules are from other dead computers so yeah maybe they're not healthy. However that parity error doesn't appear all the time so I don't really know. I should plug the battery back in (if it still works), I know that some boards don't like to not have any battery. The chips were never built there, they're not ripped from the PCB so this looks to be "normal" (even though this was always curious to see that). Maybe these chips are "less stable". Again this motherboard gets a little random for nothing even though it's in a perfect condition (when I got it it was just a little bit dusty and that was all. Probably one of the cleanest old mobo I've ever seen).

The graphic card it came with (an ISA trident card) makes the computer not to POST just like the VLB ones (no POST code on the POST card) I didn't try it in another PC so it may be just dead but I'm not sure. Maybe something in the configuration makes that the motherboard "hates" some of the boards.

Another odd thing : the VLB io card I tried does the same thing as the VLB video cards unless I push it in one way or another, then it starts to work. You may say "well there must be some grime in the VLB or ISA slots" but here is the problem : the connectors look perfect. I tried to clean them and it didn't change anything

Edit :

It seems that for some reasons, pressing the keyboard controller chip helps the board to POST. One thing I forgot to say is that the board won't start unless a keyboard is connected.

I already tried to reseat it but again, it didn't solve that. Maybe there is some trace that intermittently is conductive and some other times isn't, but I don't know why, because like I said, the board looks perfect. Maybe some component have been harmed by the PSU ?
One thing to note is that when I bought that PC, it just had it's io and graphic card. No hard disk, no CD or floppy drive, no RAM, no cover for the case ... So maybe that PC had problems back then and the previous owner harvested some parts and left the rest in his garage. The problem now is : what happened to it ?

I have a board that's perfectly well-behaved with a Cirrus Logic GD5428-based card, but an otherwise identical card based on the GD5429 will cause it to fail to POST - it's possible your board is picky in the same way. However, the issue with the keyboard controller chip, etc, makes me think maybe something is hinky.

I tried 2 VLB cards : a Western Digital one and a Cirrus logic GD5424. The second ISA card was a trident like I said and it does the same thing but if it's picky now, it wasn't before because the trident card came with the computer.

One thing interesting is that even though the battery isn't plugged in, the CMOS holds the informations for quite a long time (maybe few hours)

I may take out the hot air station and look if it does something. Maybe some chip likes heat ? Or maybe some solder joints are weak

Also try another CPU if you have one laying around.

I already did ^^ the original cpu was an SX25 but I quickly replaced it with a DX33

I may have found the area causing issues. Right next to the keyboard chip is another smaller chip (not the bios rom). Every time it fails to POST (except for the vlb stuff), pressing on it makes the board to POST. I'll resolder most of the things in that area hoping it will fix that first issue

I find socket corrosion, ICs popping out of their sockets, and bad solder joints on surface mount chips to be especially troublesome on these older boards. The bad solder joint problem has affected both of my Hercules branded VLB cards. I was able to isolate the approximate area of the bad joints and reflow the solder with a low wattage iron.

Other things to look out for are dirty contacts on the cards, and in the expansion connectors. Damaged ISA slots with pins shorting out, bad keyboard controllers, and bad RAM. If you check all of these things, you can recover most boards that don't have bad chipsets or extensive corrosion damaged caused by leaky batteries.

Okay so I reflowed the solder joints in the area of the keyboard controller. At first it didn't change anything but then I saw that the video card was plugged incorrectly. Pushed it back into the slot and it worked flawlessly. Maybe it was just plugged incorrectly before but it seems to be more reliable.

Still no VLB video unfortunately. I'm going to reflow all of the pins of the VLB but they don't look bad (around the keyboard controller area all solder joints seemed correct (I mean they didn't seem to have break) but some seemed to lack a bit of solder.

Anyways, I'm starting to get somewhere. The CMOS battery seems to be dead. It's one of these external battery. Where can I find them new ?

Edit :

Hm I may have an idea on what's going on there ... since the ISA trident card and the VLB cards doesn't work, maybe that's not something related to VLB but rather to some ISA functionality not working properly ? I'm going to check more solder joints and I'll see what will happen

Edit 2:
Uh oh, I tried to reflow a suface mounted chip that is on the left of the keyboard controller chip but unfortunately some of the pads moved ... I don't know if they touch each other but the board output a "03" POST code. Probably that at least one of them is touching another one. One thing to note though is that now, when I put the trident card, the board still outputs the code so maybe once I'll fix that, the problem would be fixed ?

I'll order some solder paste for my hot air station in order to desolder the chip and then resolder it properly.

Silly questions but-

You changed the jumper settings when you put a different CPU in, right?

80486DX-33 […]
Show full quote

80486DX-33

JP4: closed
JP5: open
JP6: open
JP7: open
JP14: pins 1 & 2, 3 & 4 closed
JP15: pins 1 & 2 closed

Also is this same issue occurring on both VLB slots?

The reason you're able to save CMOS data when unplugging the system on a dead battery is capacitors. The system draws so little power in an off state that capacitors will often provide enough charge to compensate for this for quite a while.

Yes problems happened on both VLB ports. The settings were correct too (and the problem happened when the CPU was an SX too).

I repaired the surface mount chip and it worked back but it still had this problem. I tried to reflow another chip and now it doesn't work at all 🙁

One thing to note is that I tried the Trident card that it came with in another computer and it worked perfectly so this board definitely had something funny going on. I'll see if it's really dead but I'm not sure I'll be able to revive it this time 😒

But this mobo definitely had a problem and I already have two 486s that support 5v only CPUs so this is not a big loss

Thanks for everybody's help !