VOGONS


Trying to resurrect an old 486 motherboard.

Topic actions

Reply 60 of 72, by LeeDerbyshire

User metadata
Rank Newbie
Rank
Newbie

Okay, thanks for all that. At the moment, I just have the very cheap capture-thing that all tinkerers have (I've only looked this closely 8-bit stuff before, and it has been sufficient for that). It claims to do 24Mhz, but, you know... On the other hand, I tried capturing the CPU clock with it, and it worked, although the pulse width looked a bit variable. But I can't say I'm too surprised by that - I'm amazed it did it at all.

Anyway, I'm expecting a Pico later today, so I can play with that.

The CPU does work, because I've tried it in another board I have, and I have a couple of other CPUs that don't work in this broken board.

I assume I'm running it at the correct voltage? The CPU that 'belongs' to the board is IBM26 486-V366GB, which I have set for 3.45V, because most searches tell me that's correct; but some say 5V. The manual isn't much use in this respect.

Reply 61 of 72, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++

Dont those jumper as Cyrix? https://theretroweb.com/motherboard/manual/pb … 600.pdf#page=11 pretty different form all other CPUs. Would be funny if that was stopping you from running now. EDIT: nope, you are already set correctly 🙁

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 62 of 72, by LeeDerbyshire

User metadata
Rank Newbie
Rank
Newbie

There are blips on all those pins, but that's just when I switch it on. After the reset has gone low, there is nothing. I couldn't think of a good trigger for this, so the times at the top don't mean much, but all those spikes are me switching it on.

The attachment Screenshot 2026-11-17 133335.jpg is no longer available

A2~A8 all remain low, but A9 has some activity (which could just be noise, knowing my luck). I did use a trigger for this: I triggered it by looking at the spikes, so this is well after the reset.

The attachment Screenshot 2026-11-17 133841.jpg is no longer available

Funny thing is, the BIOS doesn't have a vector in the place where I would expect, so maybe the address bus would look different to usual. Or maybe the BIOS is broken, as well.

The attachment Screenshot 2026-11-17 135248.jpg is no longer available

Reply 63 of 72, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++

trigger on falling edge of reset
zoom in, and take screenshot of whole window so settings are visible 😀
in theory BE0#-BE3# should stay up for long enough to be visible at 24MHz, are you capturing at 24mhz? did you switch mobo FSB to 25MHz? maybe FX2 is just too slow. pico will do 100MHz and 20channels, cant miss stuff with that.

Edit: but wait, ADS# and BEx# should be high all the time, if anything you would be missing fast blips going low. Are you connected to correct pins? 😮 😀

upload your whole bios file. You are missing 256 bytes at the end, but even corrupted bios file would get fetched and executed by CPU as is. bunch of zeroes decode to 'add' something, valid x86 opcodes.

You can try one from theretroweb https://theretroweb.com/motherboards/s/abit-ab-pb4-rev-1.1 afaik they are compatible across revisions.

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 64 of 72, by LeeDerbyshire

User metadata
Rank Newbie
Rank
Newbie

I tried to upload the BIOS file, but the editor will only allow me to attach images. If you are interested to see it, I hope you can get it here:
https://drive.google.com/file/d/1VGOnwwXQYaEQ … iew?usp=sharing
I did download a ROM from retro web, just to see what the last 16 bytes should look like, and yes it's clear that there is a whole chunk missing from mine. But so far I had not assumed that an older one would be compatible. I don't have any 27c1000's handy, but I have a couple of we29ee011's that might work. I don't want to overwrite my original yet. Even though I know there's something wrong with it, I might make things even worse.

Edit: It seems I couldn't overwrite the original anyway: it's an MX27C1000PC, and I don't believe it can be written to more than once. Which would also mean it can't be flashed by the user if an upgrade becomes available.

Reply 65 of 72, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++

MX27C1000 cant be reprogrammed, 29ee011 will work if you can flash it
Your dump is a bios version not on retroweb yet, would be great to somehow recover it
Sadly it has regular 00 filled 256byte long gaps every 4KB, every time address lines A8, 9 and 11 are active. So
b00-bff
f00-fff
1b00-1bff
1f00-1fff
2b00-2bff
2f00-2fff
and so on. I dont think this can happen with simply bad connection or blown address lines, as all other combinations seem to read correctly. Award bioses are compressed but I threw it into python and calculated CRCs of all 256byte blocks and no other combinations repeat (apart from empty space).
TLDR Either something wrong with your programmer or bios chip blown

ps: IMPORTANT for if/when you decide to poke at it with PICO - pico must be plugged over USB and powered _before_ probed PC is turned on. Its only 5V tolerant when powered 😀

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 66 of 72, by LeeDerbyshire

User metadata
Rank Newbie
Rank
Newbie

Yes, it's a shame this image is faulty, otherwise I would have submitted it for inclusion. I tried reading it again with the same results. The XGecu has not let me down before, and it seems to be pretty good at checking what's in it. I downloaded the BIOS version you provided a link to, and put it on a we29ee011. I put it in the PC (it still didn't work - that would be too easy I guess), but at least it didn't break it and I was able to verify it again. I noticed that the board has nearby jumpers to allow for a 12V flash BIOS. I couldn't help wondering if someone had mixed them up in the past and broken the EPROM.

Thanks for the warning about the Pico 😀 I'll try it later.

Reply 67 of 72, by LeeDerbyshire

User metadata
Rank Newbie
Rank
Newbie

I don't seem to be able to get a long enough capture out of the Pico. With my cheap/lo-res capture, I can leave it going in Salae, and see a nice long reset pulse.

The attachment Screenshot 2026-11-19 163219.jpg is no longer available

And if you zoom in to the first part (just after 0.7s), it is very unsettled.

The attachment Screenshot 2026-11-19 163335.jpg is no longer available

I think that the Pico is just capturing the first part of this, and then stopping.

The attachment Screenshot 2026-11-19 163300.jpg is no longer available

I can't see any option to prolong the capture. I assume the data is stored on the Pico, and it only has so much memory.
All this means that I can't use the falling edge of reset to trigger it.

Reply 68 of 72, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++

yes pico has 200-500KB of ram for buffering, doesnt do streaming. More channels you enable shorter the capture buffer.
Until last week cheapest fast streaming LA (USB 3) was 16 channel $70 Sipeed SLogic16U3, after that its $150 and up to thousands.
10 days ago this dropped https://blog.adafruit.com/2026/11/12/weact-lo … analyzer-board/
Their V1 was a Chinese cloned FX2 chip https://github.com/WeActStudio/LogicAnalyzerV1 sold with devboard (same thins as your FX2 24MHz LA, but manufactured with cloned chip) for $2.
This one is so new https://github.com/WeActStudio/WeActStudio.LogicAnalyzerV2 nobody tried it yet, 3.3V max so not applicable without translator board here. Deffo looks interesting 😀

so anyway
Lets start with stating the obvious - reset shouldnt look like that 😀 either M1487 is somehow screwed (it generates reset internally, there should be no bouncing, especially not 1ms of it) or what you see isnt there and/or there is something wrong with measurement. Im leaning towards the latter.
Which pin was this? Did you sample on a manual reset button pin? because that looks like a button bounce :] you can zip and upload captures instead of images.

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 69 of 72, by LeeDerbyshire

User metadata
Rank Newbie
Rank
Newbie

Yes the instability has bothered me somewhat 😀 I have been probing it by soldering under the socket 3 at the reset pin (and sreset, which has been the same). It's certainly possible that the M1487 is screwed: after all I have replaced it, and I'm no expert - far from it.

But if I probe directly at pin 20 (CPURST) of the M1487, it looks much better apart from the spike where I switch it on. I can zoom in as much as I like, and it's fine.

The attachment Screenshot 2026-11-19 180942.jpg is no longer available

BTW: If I use the [+ Add files ] button here, it only allows me to upload images. Maybe I need a more 'elevated' contributor status?

Reply 70 of 72, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++

wait, so we are back to 'tracing where the signals go'? 😀
I already forgot if you confirmed if reset is connected directly between CPU socket and chipset pins, maybe there is borken chip between? no idea why there would be one, but you never know. Do you see same bouncy reset on socket when probed with FX2 logic analyser?

Something tells me you might be just probing wrong pin on the socket, and its my fault as I think I gave you C16 which is correct when counting on CPU or PGA168 (socket 0)
socket 3 has one more row of useless pins https://theretroweb.com/connectors/157 so reset moves to D17, also Re: Socket 3 pinout help

You should be able to attach .zip/7zip files

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 71 of 72, by LeeDerbyshire

User metadata
Rank Newbie
Rank
Newbie

Yes, I'll have to do a lot of continuity testing. CPURST goes first to a nearby resistor network (5 X 100 ohm - RJ3 on the jumper diagram), but that seems okay. There is 10K of resistance between CPURST and where I have been probing, so there is something in between. At the socket end, I am probing here (and underneath when the CPU is in). From here, I do get the bouncy reset. I am using the L/H pin of the reset switch for ground.

The attachment IMG_20261119_193542[2].jpg is no longer available

Reply 72 of 72, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++

thats correct spot 🙁
that 10K you are emsuring might be just common internal pullup and they are simply not connected at all

datasheet

a) L1WB or SMI CPU
In a L1WB or SMI-supported CPU systems, the CPU
INIT( or SRESET ) pin should be connected to the
M1487 CPURST pin (pin 20), and the CPU RESET pin
connected to the M1487 SYSRST pin (pin 63) shared
with the ISA slot RSTDRV signal. And, the M1487
SYSRSTJ pin (pin 81) has the same timing and
specification but with an inversed polarity with the
SYSRST pin of M1487. The M1489 RSTDRV pin (pin
131) should be connected to the M1487 SYSRST pin.
(b) L1WT or non-SMI CPU
In a L1WT or non-SMI-supported CPU systems, the
CPU RESET pin should be connected to the M1487
CPURST pin (pin 20). The M1487 SYSRST pin (pin 63)
should be connected to the ISA slot RSTDRV signal
and the M1489 RSTDRV pin (pin 131). And, the M1487
SYSRSTJ pin (pin 81) has the same timing and
specification but with an inversed polarity with the
SYSRST pin of M1487.

TLDR: it might be using pin 81 SYSRST instead, and that on your board goes into 74f04 you replaced.

buzz out in continuity mode between CPU socket RESET and all pins of 74f04 you replaced, and all the small resistors around it. Same thing pin 20 (CPURST) and 81 63 (SYSRST ) of the M1487 against f04.

EDIT: but you already did that!
"SYSRST at M1487 pin 63 has a resistance of about 150 ohm."
so now put LA on 63 (SYSRST) , maybe its also crazy
and for good measure also do on same capture 81 (SYSRSTJ) as it also goes under F04

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly