VOGONS


First post, by tony359

User metadata
Rank Member
Rank
Member

Hello all,

I have acquired an old AMD 386DX 40Mhz motherboard as "not working". It had battery damage. The keyboard was not working. I found some broken traces, fixed them and it works.

Except when I run DOOM. DOOM starts, I can start a game, it works for 1-60 seconds and then one of the following happens
1. the keyboard stops working but the video keeps moving
2. the whole PC freezes.

I tried the keyboard under Windows 3.1 and it works no problems. All keys. Anything I try (not much to be honest) works. Except Doom.

I tried
1. disabling cache
2. removing cache
3. swapping RAM
4. several different settings for the RAM
5. Installed a heatsink on the CPU

What puzzles me is that it only happens in DOOM.

Picture attached. I am not entirely sure but the photocopy of the manual I was sent with the MB says M921. I see a potential capacitor missing.

Any ideas? I have other pretty similar boards and they work with the same RAM.
Could you also recommend a test software I can use to "soak test" a board? Something that stresses it a bit?

Attachments

  • IMG_6461.jpeg
    Filename
    IMG_6461.jpeg
    File size
    639.89 KiB
    Views
    1437 views
    File license
    Public domain
  • IMG_6463.jpeg
    Filename
    IMG_6463.jpeg
    File size
    638.22 KiB
    Views
    1437 views
    File license
    Public domain

My Youtube channel: https://www.youtube.com/@tony359

Reply 1 of 49, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Try it with typematic rate setting and Fast Gate A20 settings on and off.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 2 of 49, by tony359

User metadata
Rank Member
Rank
Member

Thanks BitWrangler.

I disabled Gate A20, enabled Typematic settings and tried with a slower rate but it still does not work.

I forgot to mention that quite often the keyboard stops working as if a key was stuck- that is, the player keeps turning left or walking straight etc. But at that point the keyboard stops responding - not even CTRL-ALT-CANC

The missing component should be identical to the one just below - what type of component is that? Looks like a non-polarised capacitor to me. - edit: all the others are 104 Z so I guess it's a 100nF decoupling capacitor.

attached the settings I have available and the missing component.

Attachments

  • IMG_6466.jpeg
    Filename
    IMG_6466.jpeg
    File size
    884.92 KiB
    Views
    1402 views
    File license
    Public domain
  • IMG_6467.jpeg
    Filename
    IMG_6467.jpeg
    File size
    324.9 KiB
    Views
    1402 views
    File license
    Public domain

My Youtube channel: https://www.youtube.com/@tony359

Reply 3 of 49, by tony359

User metadata
Rank Member
Rank
Member

Update.

I have replaced the coupling cap but that did not make a difference.
I am focussing on the keyboard BIOS? Controller? It's labelled DAC-PCKB42 181 33100. It's a 40 pin chip and its pins seem to be compatible with the intel 8042 I found on an interesting youtube video.
I have a clock and the voltages and I do see the keyboard activity on RD and WD when I use it.

Checking the TEST pins, they are normally 5V with some 0V pulses which follow my keystrokes. When the keyboard stops responding, the whole TEST pin goes low - but voltages are still there.

I would like to socket this chip and swap it with one from one of my other boards - they are called "JETI keyboard BIOS". Would that work? Are they compatible?

I am still not sure the keyboard is the problem here as sometimes DOOM crashes while loading with an "internal stack overflow - system halted". I think it only happened twice to be fair.

Thanks for your help!

My Youtube channel: https://www.youtube.com/@tony359

Reply 4 of 49, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

What I'd try next is seeing if the tag ram is a bit on the marginal side, that's the one alongside the ISA slot there. If it's identical to the rest of the cache chips (Usually is, but sometimes isn't) I'd number them all one to five and try each one out in the tag socket, see if one or two work better than the others.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 6 of 49, by tony359

User metadata
Rank Member
Rank
Member

bitwrangler

Interestingly enough I've heard of the tag ram for the first time tonight on a video. It's that fifth chip, right? It's different, I think you can see it from the picture above, horizontal. The speed is also different -15 instead of -20.

What is the TAG ram? is that part of the cache?

I've socketed the keyboard BIOS and no change (and also almost blew the board in doing so...). I've noticed that under the keyboard chip the PCB indeed says 8042! I have another keyboard BIOS but under the chip it says 8742. Are those compatible?

Anders
good ideas. I'm giving so many things for granted. The keyboard is the only one AT compatible I have I'm afraid. I can test the same keyboard on a similar system. And I can re-install Doom indeed, you never know.

Would you be able to recommend some good software for hardware test?

Thanks all for your help!

My Youtube channel: https://www.youtube.com/@tony359

Reply 8 of 49, by tony359

User metadata
Rank Member
Rank
Member

thanks, I shall give them a go and research a bit.

BTW - 8742 and 8042 seem to be pin compatible but I see the 8742 is a 12Mhz chip and I think the 8042 on the board is getting 8Mhz, if memory serves.

Thanks so far.

My Youtube channel: https://www.youtube.com/@tony359

Reply 9 of 49, by tony359

User metadata
Rank Member
Rank
Member

Ok, I've run all sort of tests.

Having re-assembled the board bending a cache chip did not help with the troubleshooting. Nor it helped having the CDROM plugged in the controller but unpowered - I had no idea that would cause the PC to fail to boot, but only sometimes.

So, I have installed CheckIt Pro and ran the comprehensive memory test for several hours. Nothing. All passed. Motherboard tests pass too.

Checking the Keyboard controller, I see the clock signal going permanently LOW when the keyboard stops working. But the clock going into the chip is unchanged. 5V is supplied normally. So I am oriented to buy another keyboard control chip. The one from the other board does not seem to work. The PCB says 8042 so I suppose I can order an 8042 or compatible from the internet?

Edit: I found a 286 motherboard with a NEC 80C42, which I suppose it's what I am looking for. Both PCBs - the board's I am repairing and the 286's - say "8042" on the PCB. However, the NEC 80C42 does not work on the board I am repairing. The board refuses to boot up.

I am now a bit concerned in spending money on Ebay purchasing an 8042 to be honest. Given the 8042 is a programmable chip, maybe that board needs some kind of special programming? Does anybody know anything more about that?

Attached the current chip, which does not say much about it's origin...
But the label on top said "AMI KEYBOARD BIOS" which I am concerned may mean that that needs to match the BIOS firmware...

Attachments

My Youtube channel: https://www.youtube.com/@tony359

Reply 10 of 49, by tony359

User metadata
Rank Member
Rank
Member

more updated! 😀

I thought that if the Keyboard BIOS is programmed to work with that BIOS only, I could try to swap both BIOS and Keyboard BIOS from another 386/DX board. I swapped the BIOS with the Kyeboard BIOS from yet another board (the one paired to thet 386/DX BIOS is still soldered) and no boot. Then I tried the other 386/DX BIOS with the "faulty" keyboard BIOS, it booted up but it can only see 640KB of RAM instead of the 8MB I have installed.

So it seems that I am in that "rare" situation where the Keyboard BIOS are somehow programmed to match the board's BIOS.

Now, my Keyboard BIOS is still working. Is there a way to extract the programming and flash it to another one? I have a MiniPro but I do not see the 8042 on the list of programmable chips...

My Youtube channel: https://www.youtube.com/@tony359

Reply 11 of 49, by maxtherabbit

User metadata
Rank l33t
Rank
l33t

There aren't *that* many different KBCs that are mutually incompatible. If you can find any other AMI marked KBC it will probably work fine for you.

Reply 12 of 49, by Horun

User metadata
Rank l33t++
Rank
l33t++

When you boot the computer with older AMI bios the last letter in the bottom bios string is the Keyboard controller version.
example: 40-00AG-001247-00101111-060692-SIS3486- F. The -F is the "KBC rom version". Try to match the version as not all are interchangeable but many are.
Seeing 1986 AMI on the label makes it an early version and matches the 1986 on the BIOS tag.

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. https://archive.org/details/@horun

Reply 13 of 49, by tony359

User metadata
Rank Member
Rank
Member

I will give it a go, thanks. The other KBC I tried is indeed an AMI KBC. I'll pay attention to the version.

What about the VIA "generic" KBC which are mentioned on the internet? The VT82C42.

Thanks

My Youtube channel: https://www.youtube.com/@tony359

Reply 14 of 49, by tony359

User metadata
Rank Member
Rank
Member
Horun wrote on 2021-09-25, 20:10:

When you boot the computer with older AMI bios the last letter in the bottom bios string is the Keyboard controller version.
example: 40-00AG-001247-00101111-060692-SIS3486- F. The -F is the "KBC rom version". Try to match the version as not all are interchangeable but many are.

The last string is "8". Does that make any sense?

My Youtube channel: https://www.youtube.com/@tony359

Reply 15 of 49, by tony359

User metadata
Rank Member
Rank
Member

I've received the VT82C42. Unfortunately it works on other motherboards but not on this one.

I then realised I did not do a more simple test: I put the suspected faulty KBC on another motherboard and everything works fine. No issues. So the chip is good.

I then started scoping the chip on the working motherboard with the same chip on the faulty one. The main thing that catches my attention is that on the "good" board I have 7.15Mhz on XTAL 1 and XTAL 2. On the "bad" board I only have a clock on XTAL 1, nothing on XTAL 2. Is that relevant?

My Youtube channel: https://www.youtube.com/@tony359

Reply 16 of 49, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
tony359 wrote on 2021-10-26, 19:07:

I then started scoping the chip on the working motherboard with the same chip on the faulty one. The main thing that catches my attention is that on the "good" board I have 7.15Mhz on XTAL 1 and XTAL 2. On the "bad" board I only have a clock on XTAL 1, nothing on XTAL 2. Is that relevant?

Have you got some pictures showing the oscillator circuit on a good board and the bad board? I think that the keyboard controller can have a couple of possible clocking methods. One is have a 2 pin crystal and a couple of low value capacitors, with the crystal going between XTAL1 and XTAL2. The other is to have a oscillator chip (probably 4 pin) that can drive a clock input directly, which (from the datasheet) drives XTAL2, with XTAL1 grounded.

You're seeing nothing on XTAL2, which doesn't match the datasheet. So it's possible that the bad board has a duff crystal or a failed capacitor. Or I'm misreading the datasheet.

Reply 17 of 49, by tony359

User metadata
Rank Member
Rank
Member

I shall capture some screenshots and update here. In the meantime, would you be able to share the datasheet with me? For the life of me I cannot find it anymore... 😀

Thanks!

My Youtube channel: https://www.youtube.com/@tony359

Reply 19 of 49, by tony359

User metadata
Rank Member
Rank
Member

I didn't think of searching for the VT82C42, I was looking for the 8042! The datasheet is very clear and explains a lot. Yes, either case I should have the clock on XT2 so I'll investigate on that.
Thanks a lot, I'll follow up with my findings!

My Youtube channel: https://www.youtube.com/@tony359