This is interesting. I decided to replace one cache chip at a time, starting with the very outer cache chip on the mobo - the one closest to the edge of the mobo. I removed the factory fitted 20ns chip, and replaced it with a 15ns chip. I reran cachechk, and now I don't get any errors. Problem solved, perhaps?

Edit: That didn't work. Windows 95 still halts due to a memory parity error detected problem. I will replace all of the cache chips on the mobo, then retry running Windows 95.

Edit 2: I changed all 4 cache chips, but that hasn't solved this problem. The next thing to do is to change the tag cache chip. If that fails to solve this problem, then the last chip I can replace is the "QS 8888" chip...

Are the SIMMs too slow maybe? And as the memtest86 authors say, the program not finding a RAM error doesn't mean its error-free.

h-a-l-9000 wrote:

Are the SIMMs too slow maybe? And as the memtest86 authors say, the program not finding a RAM error doesn't mean its error-free.

They are 70ns, and I think they are all OK. Very unfortunately, I have had an accident with this board. Before removing the tag + QS chip, I decided to remove all 8 SIMMs, to replace them with another set of 8 SIMMS. I removed 5 SIMMs, when my screwdriver slipped, and landed heavily on to the board. It hit one of the chipset chips, and smashed up its pins. I have now abandoned this board, and I am going to test another Contaq 386 mobo that I was having the same problem with.....

h-a-l-9000 wrote:

Are the SIMMs too slow maybe? And as the memtest86 authors say, the program not finding a RAM error doesn't mean its error-free.

You're right, yet again! 😁 I put the 8 SIMMs that Memtest said were OK in to a known working mobo, and Windows 95 shows the same BSOD: memory parity error. So, one or more of the SIMMs are defective in some way.

Solved. I started by removing 4 of the 8 SIMMs. I replaced these 4 with another set of known working 4 SIMMs. That works. I no longer get Windows 95 BSOD errors.

I would suggest that you take out one of the working 8 you have now, and replace it by one of the 4 that were causing the parity error. Do that for every one of the 4, so you can check which one is the faulty.

retro games 100 wrote:

I removed 5 SIMMs, when my screwdriver slipped, and landed heavily on to the board.

I don't use a screwdriver when removing simm memory. I'll just use my thumbs.

Hint:when handling a screwdriver with force when working on a motherboard, use the other hand to 'control' the screwdriver. While it's not fail safe, it has saved me a lot of costly slips in the past 😉

Good advice. Thanks a lot people.

Damn it. I've just trashed another board. Same problem. One tiny slip of the screwdriver, and it went straight in to the pins of a chipset chip. Now I'll have to prize out the 9 new cache chips on this board that I only put in a few hours ago. 😒

Edit: note to self - next time you use a screwdriver to remove SIMMs, remove the mobo from the test area, and stick it on a table. Then, when you use a screwdriver, you can push the blade parallel to the table, so if the blade slips, it does *not* go downwards, towards the mobo. Instead, it will harmlessly go sideways, and not hit anything. Moron!

retro games 100 wrote:

Damn it. I've just trashed another board. Same problem. One tiny slip of the screwdriver, and it went straight in to the pins of a chipset chip. Now I'll have to prize out the 9 new cache chips on this board that I only put in a few hours ago. 😒

Edit: note to self - next time you use a screwdriver to remove SIMMs, remove the mobo from the test area, and stick it on a table. Then, when you use a screwdriver, you can push the blade parallel to the table, so if the blade slips, it does *not* go downwards, towards the mobo. Instead, it will harmlessly go sideways, and not hit anything. Moron!

Ah man, my condolences. Those 2 boards can never be replaced. Otoh, this is one the reason I got 3 of those boards, so the question to RG100 is...got any more spares? 😜

And another thing...don't do it again! 😜 It's a shame to kill these old NOS boards 🙁

Don't use a screwdriver to remove simm modules, just push the 2 tabs with a finger one by one. After the 2nd one snaps loose, the simm will be completely detached (shaking the board will make the simm fly out!).

Can you make a good photo of the crash? I think it might be fixable. That time technologies were rough and not so precise so a soldering iron can help.

RG100: destroyer of hardware 😜 🤣

I only bought one of those boards and its still chugging along fine although maybe I should buy a spare before they are all destroyed 🤣

Seriously though I'm glad that modern RAM goes in vertically and has plastic clips, because the angled SIMMs are a pain. I generally just use my fingernail though and don't have too much trouble.

Are you certain that your motherboards are toast though? It sounds like your original problem was just dodgy SIMMs, maybe your board is still OK? I slipped the screwdriver installing a HSF on a Socket A board and it looked like I actually cut some tracks on the PCB but the board worked fine for years.

Luckily, I bought a bunch of them, so yes I have some spares. In fact, I'm retesting them all today. The reason why I'm retesting them all is that some ebayer sent me a pile of SIMMs, and they worked OK in DOS, but as soon as you do anything in Windows 95, memory parity errors pop up. I've found a couple of bad sticks already, and I'm just to going to retest the other stuff today.

Unfortunately, the macro mode on my camera doesn't work, and so taking a photo is impossible. I did try and fix the problem with a pin and a magnifying glass, but it was very difficult to move the damaged pins. When I did move them, they uncontrollably bent the other way, making the situation worse.

I had a chat with weirdstuff yesterday, and he might have some more of those boards. He'll check on Monday. Apparently, he had to scrap some of them because of leaking batteries. However, mine are all fine. Tetrium, did you mention leaking batteries to him?

Edit: Actually, one of the batteries has a tiny amount of goo coming out of it. It hasn't touched the mobo yet, so any damage can be avoided by removing the battery. I must get a solder/desoldering iron!

retro games 100 wrote:

Tetrium, did you mention leaking batteries to him?

No, not to him. All of my boards had no sign of leakage, it was the 1st thing I checked when I got them. It could very well be that they are on the very verge of leaking though...I should really remove those batteries soon.

I think the seller deserves a BIG THUMBS UP for tossing out the bad boards instead of shipping them to customers! 😉

Tetrium wrote:

retro games 100 wrote:

Tetrium, did you mention leaking batteries to him?

No, not to him. All of my boards had no sign of leakage, it was the 1st thing I checked when I got them. It could very well be that they are on the very verge of leaking though...I should really remove those batteries soon.

I think the seller deserves a BIG THUMBS UP for tossing out the bad boards instead of shipping them to customers! 😉

I totally agree with you.

1st: Do not use a Screwdriver to remove the SIMMs ! Use your hands. And as always, don't hurry doing things. Take your time and be careful with the old stuff.
Just imagine this: You just bought it for new and paid a lot money for it, you wear your white cloth gloves...

2cnd: The old SIMMs have Parity, sometimes you see 9 Chips. The NMI is connected to the chipsets Parity Error detection. If the BIOS takes over and shows Parity Error it means that the NMI was triggered. If you didn't installed stuff as soundcard emulation TSRs that use NMI it is most probably a memory error. From this point of view it is easier to identify such things.

Please note that the following 2 paragraphs have been copied-and-pasted from the 386DX build thread, and so if you've read it, please ignore it. Thanks.

I wish my camera's macro mode was working. Because looking at some of the Contaq 386 mobos, the SIMM clips are styled differently, depending on what board I am looking at. The one I am working on at the moment has extremely convenient "outer" clips, whereby you simply press on them with your finger tip, and they go outwards, allowing the SIMM to "fall over", and out of the SIMM socket itself.

The other boards are evil, because they don't have this, only one very tight clip on either side, that requires a lot of force to move them. After you move them, you have to move the opposite clip a bit. Then you have to go back to the first clip, and move it a bit more. Etc etc. Until it finally lets go of its tight grasp on the SIMM, and it can be pushed over, and then removed from the socket.

You have to move both simultaneously

Message to either Donut King or Tetrium! Please look at your Contaq 386 mobo manual, on page 2-6. At the bottom. Hehe, what's the awesome advice they provide? 🤣