VOGONS


First post, by Cosmic

User metadata
Rank Member
Rank
Member

I started a thread yesterday about installing a card in a never-before-used ISA slot in my first and only 486 system, thanks all who commented! Removing slot covers from a pristine AT case

It seemed to work at first, but after a power cycle, I couldn't get the system to POST anymore. I figured I would start a separate thread for debugging in case anyone wants to read along or has ideas. 😀

Specs

  • Mobo: DTK PKM-0037S, SiS 85C461, PGA 168 (1994) (RetroWeb Manual)
  • BIOS: AMIBIOS (C)1993 American Megatrends Inc., DTKPKM0037S.001.03.2 (I need to dump this to RetroWeb, as it may be slightly newer)
  • CPU: 486 DX2-66 (SX955) (1992) /w cooler
  • RAM: 32MB 30-pin Samsung FPM (1994)
  • Disk: Quantum Bigfoot 1.2GB (1996)
  • CF: Syba Dual CF adapter + various Cisco CF cards
  • CD-ROM: Teac CD-W58E (2000) (Maybe was replaced later)
  • IO Card: Relialogic CA9342 ISA (1993/1994)
  • Sound: Sound Blaster CT2290 (1993)
  • Video: Cirrus Logic CL-GD5426 VLB (1993)
  • Network: Intel 8/16 EtherExpress ISA (1993?) w/ XT-IDE ROM

Upgrades
The system has been rock solid since I first got this past summer. The Dallas RTC chip was replaced and holds time. I upgraded the RAM from 8MB to 32MB to experiment with different operating systems, and upgraded the CPU from a DX-50 to a DX2-66 and added a cooler. This made it a little bit faster, but mostly it brought the bus speed down from 50MHz to 33MHz so I could one day run 2 VLB cards (VGA and IO). Also added a CF adapter but I still use the Bigfoot too, it sounds nice.

Breaking something...
I popped out one of the slot covers and moved the Intel 8/16 card to the never used ISA slot (ISA Slot #4, just below VLB Slot #2). The system finished POST and booted all the way to the card's XT-IDE ROM, but then I realized I didn't insert my CF card, so I powered off to add it. Afterwards the system still powered up, but no VGA signal. Pulled all the cards and still nothing.

So the troubleshooting begins...

Board photos

6e7xc5I.jpg

ATF9kOo.jpg

Reply 1 of 4, by Cosmic

User metadata
Rank Member
Rank
Member

The first thing I did was pull the motherboard and put it on an anti-static bag, then test it with just a POST card installed. It powers up and the reset line works fine (lights up on reset switch, clears when I release it).

I used my phone's slow motion mode to try and get a better view of the codes.

Using this resource (https://www.bioscentral.com/postcodes/amibios.htm) under the "AMI BIOS Post Codes (After April 1990)" section, I partially decoded the codes.

  • 04 - Passed keyboard controller test with and without mouse
  • 05 - Chipset initialized...DMA and interrupt controller disabled
  • 06 - Video system disabled and the system timer checks OK
  • 0F - ??
  • 10 - Ready to start 64KB base memory test
  • 12 - 64KB base memory test OK
  • 13 - ??
  • A 12 second pause, about the same time as it normally takes to check 32MB...
  • 00 - Call the INT19 boot loader

Even if this isn't the right list of codes, most of them seem to list 0x00 as "Call interrupt 19 boot loader" - so maybe it is trying to boot? This is a reminder to myself to capture the POST codes when the system is working normally, so I have some point of reference - hah.

No Beeps
But then there's the issue of no POST beep. I can hear the speaker crackle a bit when it first gets power, so it seems connected fine, but no beeps.

Next I pulled all the SIMMs out - surely the board should make some noise with no RAM, right? Still no beeps and the POST codes remain the same, even the long pause between 0x13 and 0x00 which I thought was counting memory. I really thought I'd see a difference in POST codes with no RAM.

Short checks
This will probably take the longest. I'm kind of figuring it out as I go. I checked the six large yellow tantalum caps under each ISA slot and got about 180 ohms on each, so those seem OK.

I may have to test all the ISA slots, eh? 16-bit ISA is 98-pins x 7 slots = 686 continuity checks. 😁

Questions

  • Do these symptoms make sense? There's clearly activity and the CPU is doing something. Could a cracked or broken ISA trace as suggested by Trashbytes break the whole ISA bus?
  • Would you expect the system behavior to change with no RAM? I figured it would just hang and beep angrily with no RAM, but instead it continues on like nothing happened.
  • Should I try to acquire a plain ISA VGA card to test with? I have plenty of PCI and AGP cards, but no ISA and only 1 VLB card.

Keyboard update
I had a shower thought - what if I plug in the keyboard and see if I can navigate the BIOS blind? I have a near identical machine setup in 86box for this purpose. The keyboard is an IBM KB-8923 from 1998. All the lights light up on power-on, but then they go off and no response on Caps/Scroll/Numlock, so I think it won't work.

Post code update
Here are the POST codes from 86box running an almost identical chipset (SiS 85C460 vs my 85C461).

  • 04 - Passed keyboard controller test with and without mouse
  • 05 - Chipset initialized...DMA and interrupt controller disabled
  • 06 - Video system disabled and the system timer checks OK
  • 0F - ??
  • 10 - Ready to start 64KB base memory test
  • 12 - 64KB base memory test OK
  • 13 - ??
  • <-- Board seems to fail here
  • 21 - ROM type 27256 verified
  • 22 - The power-on message is displayed
  • 2B - ??
  • 2C - ??
  • 3B - Ready to test the lower 1MB of RAM
  • 40 - Cache memory now on...Ready to disable gate A20
  • <-- BIOS splash appears here
  • 4E - ??
  • 4F - ??
  • 50 - DMA page register test OK
  • 51 - Starting DMA controller 1 register test
  • <-- memory count occurs here

The number of missing descriptions for these codes makes me think I'm not looking at the right POST code list, but it's something.

BIOS chip update

Here's the text from the BIOS chip, will need to check under the label to get the exact chip. I'll dump it to verify it against the similar ROM on RetroWeb.

The hang around POST code 13 followed by the jump to 00 could maybe indicate the chip somehow went bad and is corrupt? Worth a shot checking I suppose. Also, removing the chip entirely causes the machine to do nothing, which is expected.

AMIBIOS
AMERICAN MEGATRENDS
486DX ISA BIOS (C) 1993
AB186594 (serial number as far as I know)
Last edited by Cosmic on 2023-12-14, 03:48. Edited 4 times in total.

Reply 2 of 4, by MikeSG

User metadata
Rank Member
Rank
Member

If all you did was change ISA cards, try reinserting them a few times and/or cleaning the contacts on each ISA bus (isopropyl alcohol + a toothbrush).

One sure fire way to test each slot is to use an ISA VGA card like you said. If it boots, that slot is good.

Some cards don't like a bus clock divider set too low as well (more than 8MHz), if that's set. Maybe a BIOS reset/clear would help.

Reply 3 of 4, by H3nrik V!

User metadata
Rank Oldbie
Rank
Oldbie

In your other thread, you state that you have a network card to have an XTIDE BIOS. Could that be playing tricks on you (have no idea how it works, just babbling from the top of my head here 🤣 )

Please use the "quote" option if asking questions to what I write - it will really up the chances of me noticing 😀

Reply 4 of 4, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie
Cosmic wrote on 2023-12-12, 01:50:
Questions […]
Show full quote

Questions

  • Do these symptoms make sense? There's clearly activity and the CPU is doing something. Could a cracked or broken ISA trace as suggested by Trashbytes break the whole ISA bus?
  • Would you expect the system behavior to change with no RAM? I figured it would just hang and beep angrily with no RAM, but instead it continues on like nothing happened.
  • Should I try to acquire a plain ISA VGA card to test with? I have plenty of PCI and AGP cards, but no ISA and only 1 VLB card.

POST card displaying different codes is a good sign, especially once the 64k test passes because then you know you have working ROM and RAM. One of the address or data lines might be weak (or even bad on ROM side, depending how the code is structured) but it's a remote possibilty at this point. RAM errors past the first 64k should be detected or if not the system would usually at least try bo boot much further.

No RAM on 486 is a lottery. Depends on the mobo and BIOS - sometimes there are beeps, sometimes the system appears dead. That can also happen with bad RAM sticks, somehow unsupported sticks (size or parity) or wrong banks populated. Stick to the configuration that you had when the system worked, for now assume it's not RAM if the first 64k test passes.

VLB cards are notoriously difficult to get them to work properly. Also, tend to die rather easily - for some reason (speed I guess) the VLB part is not as robust against ESD as the ISA part is. A pure ISA VGA is very much recommended - especially in your case where the system kinda boots up to a point but fails to display anyting at all. You can also reset CMOS settings in case it's something there preventing the BIOS from doing a full cold boot and thus not initializing all hardware properly.
You can try plugging the card into different VLB slot, reducing system clock to 25MHz via jumpers, and perhaps look into mobo manual to see if there are any VLB related jumpers (mastering, waitstates) to play with. Could be the system will not work properly with your current settings and so far you were just lucky it booted at all. Hopefully it's not a dead video card.

Oh and BTW the VLB slots are also rather prone to bad contact due to dirt and dust, some compressed air and cleaning with a soft brush could help. The slot can also be very picky about how deep the card is inserted and if there is any angle to it - if the screw holding the card down tends to pull it out of the slot it can cause such issues (as well as permanent damage in some cases).