so its not the ISA clock 🙁

> Dead Pentium SBC does the same with the clock as the beeping one.

I thought after replacing RTC both Pentium SBCs were in a beeping stage

Ok, the POST card arrived yesterday and I think I have some progress.

1. The broken, non-bleeping Pentium SBC is now partly fixed. It's bleeping again and showing garbage on the screen, but it seems to boot-loop. I will investigate that further later. The issue? I just happened to notice two pins at the edge on an IC near the RTC module were bent and shorting out. The IC is a CHIPS F82C735 and the shorting pins were 49 & 50, "D1 & D2". I must've bent these two pins together when removing the original RTC module and not noticed. Ooops!

2. The beeping Pentium is still doing the same thing and the POST card shows FF 2A while going through the boot sequence, then 85 FF when it's finished. I have the manual for the POST card and it's not clear what the digits mean; there is a decode table but it only refers to two digits, and the board has 4 of course. I can only think that it's sets of two digits for each brand of motherboard, for example left side for AWARD brand, and right side for AMI? If that's correct, then my codes are:
2A - "Initializing the different bus system, static and output device if present"
FF - "Flash ROM programming was succesful. Next, restarting the system BIOS"

Of course I could have read the codes wrong and they mean something different, but if not them it looks to be normal. This ties in with rasz_pl's theory that it's a graphics card issue.

Unfortunately I either have a loose ISA, PCI or power supply connector as well. The connection to the backplane is intermittent somewhere but I am going to check everything properly later on.

I'm now waiting for a PCI graphics card to arrive and will report back once it does...

Someone asked what the ICs were on the Scorpion graphics card- there are two CHIPS F82C452A

rasz_pl wrote on 2023-11-07, 01:50:

so its not the ISA clock 🙁

> Dead Pentium SBC does the same with the clock as the beeping one.

At this stage I am thinking you were right and it's the graphics card, but only issue is that graphics works fine on the 486 SBC and not the Pentiums.

1. One Pentium is beeping as it always did, with garbage on the screen. POST card says 85 FF which looks to me like it's completed a boot sequence, and you get two "beep-beeps" at the end.
2. Second, repaired Pentium now does the same, but takes a LOT longer to boot. POST card says CD, CB for maybe 20-30 seconds and then it'll do the normal beeping and booting thing.

I have a PCI graphics card on the way and am thinking about getting an ISA one as well, just because I'm not 100% sure the sockets on the backplane are good. I've cleaned them with contact cleaner and a toothbrush, for now it seems to have done the trick.

rasz_pl wrote on 2023-11-07, 01:50:

I thought after replacing RTC both Pentium SBCs were in a beeping stage

Sorry, no they weren't. I know my thread is quite confusing to read. Both Pentiums are now (roughly) doing the same thing, beeping and garbage on the screen.

Venture wrote on 2023-11-08, 13:12:

POST card shows FF 2A while going through the boot sequence, then 85 FF when it's finished.

https://mrbios.com/techsupport/award/postcodes.htm scroll to AMI New BIOS; 06/06/92-08/08/93
FF is not a valid POST code, and those numbers make no sense for standard AMI bios 🙁 ISA bus corruption would explain weird codes, and would explain broken Video, wouldnt explain how video card is able to be initialized by its video bios being read thru same ISA bus (maybe VGA chip sets up generous wait-states and that somehow prevents corruption?).

Venture wrote on 2023-11-08, 15:23:

only issue is that graphics works fine on the 486 SBC and not the Pentiums.

how thick is the 486 SBC edge connector and how thick are Pentiums? Hard gold contacts all look fine?

Venture wrote on 2023-11-08, 15:23:

POST card says CD, CB for maybe 20-30 seconds and then it'll do the normal beeping and booting thing.

Once again not valid POST codes. Do you have an eprom reader to dump 486, Pentium and vision card bioses?
Either those are special bioses for a special embedded version of AMI (if one existed at all) or there is corruption on the ISA bus.

Venture wrote on 2023-11-08, 15:23:

I'm not 100% sure the sockets on the backplane are good.

yep, I like this theory. POST card in PCI slot would be a good start. Can you post photo of whole backplane and 486 SBC?

Try POST card with 486 SBC.
Try POST card in PCI slot with Pentium SBC, make sure to plug it in correct orientation!
Record POST card codes, preferably at high framerate if they change fast.

rasz_pl wrote on 2023-11-09, 00:52:

https://mrbios.com/techsupport/award/postcodes.htm scroll to AMI New BIOS; 06/06/92-08/08/93 FF is not a valid POST code, and th […]
Show full quote

Venture wrote on 2023-11-08, 13:12:

POST card shows FF 2A while going through the boot sequence, then 85 FF when it's finished.

https://mrbios.com/techsupport/award/postcodes.htm scroll to AMI New BIOS; 06/06/92-08/08/93
FF is not a valid POST code, and those numbers make no sense for standard AMI bios 🙁 ISA bus corruption would explain weird codes, and would explain broken Video, wouldnt explain how video card is able to be initialized by its video bios being read thru same ISA bus (maybe VGA chip sets up generous wait-states and that somehow prevents corruption?).

how thick is the 486 SBC edge connector and how thick are Pentiums? Hard gold contacts all look fine?

All edge connectors same thickness & in good condition.

Venture wrote on 2023-11-08, 15:23:

POST card says CD, CB for maybe 20-30 seconds and then it'll do the normal beeping and booting thing.

rasz_pl wrote on 2023-11-09, 00:52:

Once again not valid POST codes. Do you have an eprom reader to dump 486, Pentium and vision card bioses?

Maybe, I have a programmer called Labtool 48UXP but it let out some magic smoke a few years back so not sure it is fully working. Also, I think they may charge to use the Software for it nowadays.

rasz_pl wrote on 2023-11-09, 00:52:

Either those are special bioses for a special embedded version of AMI (if one existed at all) or there is corruption on the ISA bus.

No, I was told by the engineer who worked on these machines that the motherboards are standard. He said I could replace it with any standard SBC, the only issue arising might be that it wouldn't like the old HDD with DOS and data on it.

Venture wrote on 2023-11-08, 15:23:

I'm not 100% sure the sockets on the backplane are good.

rasz_pl wrote on 2023-11-09, 00:52:

yep, I like this theory. POST card in PCI slot would be a good start. Can you post photo of whole backplane and 486 SBC?

POST card works the same in either ISA or PCI slot, same output.
After cleaning connectors with cleaner, boards seem reliable in the slots now.

rasz_pl wrote on 2023-11-09, 00:52:

Try POST card with 486 SBC.
Try POST card in PCI slot with Pentium SBC, make sure to plug it in correct orientation!
Record POST card codes, preferably at high framerate if they change fast.

Will do this ASAP and report back, probably tomorrow. Hopefully a graphics card will arrive by then as well.

Venture wrote on 2023-11-09, 20:51:

POST card works the same in either ISA or PCI slot, same output.

this is worrying, FF should not be there, if you got POST card with a button you can press it to scroll codes back
add POST card test without vision card just to see if codes are different, maybe video bios injects its own

Great news!

Both graphics cards arrived today and so I tried the ISA one first.

Immediately I got a proper image on the screen and was greeted with a BIOS message:

"CMOS battery low"
"CMOS Display type mismatch"

Put in the Necroware RTC module and "battery low" message went away.

Decided to change the video settings one-by-one and see if any made the weird Scorpion graphics card come back to life-

1. Adaptor Shadow Cacheable DISABLED - This was already disabled by default
2. System BIOS Shadow Cacheable DISABLED - This was already disabled by default
3. ISA Video ROM C000,32K ABSENT - This was ENABLED by default, so I set it to DISABLED

I plugged in the old graphics card and now it works!

Also, the "CMOS Display type mismatch" message disappeared using ISA graphics card.

So, all along it was the damn RTC battery. The RTC battery must've run flat and the SBC defaulted to ISA Video ROM C000, 32k ENABLED. This causes garbage on the screen for the old Scorpion graphics cards, but with a normal video card you can get in to the BIOS and make changes.
It still doesn't explain why the 486 SBC worked with the old Scorpion card, but it does. The BIOS looks very similar between the two SBCs, 486 & Pentiums.

Even though one Pentium board is fixed, I still have some work to do though;

1. Fix the second Pentium SBC. After removing the short on the IC with the bent pins, it was working but booting really slowly. I went back to check that IC again and managed to break one leg off. Even with grinding away at the IC body to try and expose some metal, I couldn't solder a wire on. There were some "New, original" ICs available on Aliexpress for £2 each so I've ordered a couple. It's a gamble whether they'll work, but we'll see.

2. I want to replace old HDD with CF card. I have a couple of different CF card caddies and 2.5" to 3.5" IDE adapter.

Does anyone know if a CF card could work without mods?

Next step is to try and auto detect the CF card as a slave and see if I can get it to boot? Actually just done this, it doesn't.

Attached are some photos of the BIOS, video settings and slave CF HDD fail.

rasz_pl

You asked for a photo of the PC- please see attached. All the original cards are installed now and everything works. Note how they used a conventional linear power supply with huge transformer instead of SMPS, for reliability. The whole weighs maybe 25kg.

Also, a screenshot of the pick & place software, all DOS based. Not hugely interesting in its current state because it's running off the spare PC and no cameras are plugged in etc. etc. When I next do a job on the actual pick & place machine, I'll post a video if anyone is interested?

Also, just tried the PCI graphics card and that works too.

Is it safe to try two graphics card plugged in at the same time, for example Scorpion & ISA or Scorpion & PCI? You mentioned they might work together, so if this failure happens again in the future I can just plug in a spare graphics card & change the BIOS video settings without removing any other cards or disturbing anything.

Last job is to try and get the CF card working as a slave, then copy all the data to it and re-install it as a master.

Attached is a picture of the caddies I have, the one on the left has a master/ slave jumper and the one on the right has two CF slots, you flip it over to put a card in the slave position.
I have tried to auto detect the CF and and that works, but then I get an error message as in my previous post. Of course I have the option of manually selecting all the drive parameters in BIOS?

Currently, the 128MB CF card is detected as shown in the last attached photo. I did format the card on a modern Windows PC but can't remember which file system I used, NATS / FAT32 etc.

Thanks!

Venture wrote on 2023-11-10, 19:03:

Immediately I got a proper image on the screen and was greeted with a BIOS message:
"CMOS battery low"
"CMOS Display type mismatch"

Second one might be a clue to your troubles. What was the "Primary Display Monitor" set to by default with empty CMOS? Options are usually VGA/EGA, CGA 40x25, CGA 80x25, or Monochrome.

Venture wrote on 2023-11-10, 19:03:

Decided to change the video settings one-by-one and see if any made the weird Scorpion graphics card come back to life-
3. ISA Video ROM C000,32K ABSENT - This was ENABLED by default, so I set it to DISABLED
I plugged in the old graphics card and now it works!

ISA Video ROM C000,32K

When this option is set to Enabled, the video ROM code, that normally mapped into memory address space from C0000h -C7FFFh is copied (shadowed) from ROM to the system DRAM for faster execution. This will significantly improve the display performance of the system. The settings are Absent, No Shadow, or Shadow. Optimum default setting is Enabled.

This setting changes only cacheability of Video BIOS, it should not have any influence over video Bios initialization!?!? 😀

Venture wrote on 2023-11-10, 19:03:

Also, the "CMOS Display type mismatch" message disappeared using ISA graphics card.

is it still there with Industrial Vision card?

Venture wrote on 2023-11-10, 19:03:

The RTC battery must've run flat and the SBC defaulted to ISA Video ROM C000, 32k ENABLED. This causes garbage on the screen for the old Scorpion graphics cards, but with a normal video card you can get in to the BIOS and make changes.

This is one of the outcomes I was hoping for 😀, but it still makes no sense, have you verified that changing 'ISA Video ROM C000, 32k' to ENABLED makes the graphic output corrupted again?

Venture wrote on 2023-11-10, 19:03:

Even with grinding away at the IC body to try and expose some metal, I couldn't solder a wire on.

its fiddly even with a microscope

Venture wrote on 2023-11-10, 19:03:

2. I want to replace old HDD with CF card. I have a couple of different CF card caddies and 2.5" to 3.5" IDE adapter.
Next step is to try and auto detect the CF card as a slave and see if I can get it to boot? Actually just done this, it doesn't.

You can try disabling IDE Block mode, but usually its due to CF cards IDE mode not being fully tested and IDE master/slave being very awkward https://www.os2museum.com/wp/the-dual-drive-ide-hell/ (one drive has to pretend to be both of them). You can try the other way around - set CF as master, IDE as slave, or image HDD and copy it to CF in another computer, or few other CF cards until you find one working. Removing file system from CF card might also help.

Venture wrote on 2023-11-10, 20:18:

Note how they used a conventional linear power supply with huge transformer instead of SMPS, for reliability. The whole weighs maybe 25kg.

is it a fully linear supply 😮 or an isolation transformer in front of SMPS?

Venture wrote on 2023-11-10, 20:18:

Is it safe to try two graphics card plugged in at the same time, for example Scorpion & ISA or Scorpion & PCI? You mentioned they might work together, so if this failure happens again in the future I can just plug in a spare graphics card & change the BIOS video settings without removing any other cards or disturbing anything.

No is my guess. ISA VGA cards are hardcoded to certain addresses (0xB8000 for text buffer etc) and there is no was of dynamically managing it. With two PCI VGA cards you can assign different hardware windows, but BIOSes werent equipped for it, Windows was able to handle couple cards at once but even then your only control over which card is primary is placement order in expansion slots.
CGA card maybe, but that still leaves the question of switching between them and CGA is kind of useless without CGA monitor anyway.
The question remains what is up with your industrial card, is it an ordinary VGA card with VGA bios at C0000 or maybe it has custom bios pretending to be something else. https://archive.org/details/checkit_30 with Vision card plugged will tell us what it identifies as, 'Base memory map' will show where Video memory and bios are mapped.

Its totally safe to plug both PCI and ISA VGA cards together, but no idea which one will be primary, most likely ISA with no option of switching to PCI.
I would start by playing with "ISA Video ROM C000,32K" option, Im not convinced it was the culprit. Also try different "Primary Display Monitor" settings trying to recreate fault condition.
My weak theory now is industrial Vision card bios not handling CGA or Monochrome modes and Pentium SBC having a jumper on it somewhere defaulting to one of those options with empty CMOS. Once booted and in BIOS Setup VGA/EGA is detected and saved.

Btw are still weird 'FF, CD, CB' POST codes showing up?

rasz_pl wrote on 2023-11-11, 02:49:

Second one might be a clue to your troubles. What was the "Primary Display Monitor" set to by default with empty CMOS? Options are usually VGA/EGA, CGA 40x25, CGA 80x25, or Monochrome.

It was always VGA/EGA, and I've never changed it.

rasz_pl wrote on 2023-11-11, 02:49:

ISA Video ROM C000,32K

When this option is set to Enabled, the video ROM code, that normally mapped into memory address space from C0000h -C7FFFh is copied (shadowed) from ROM to the system DRAM for faster execution. This will significantly improve the display performance of the system. The settings are Absent, No Shadow, or Shadow. Optimum default setting is Enabled.

This setting changes only cacheability of Video BIOS, it should not have any influence over video Bios initialization!?!? 😀

I notice that during boot, before you can get in to BIOS, you see a graphics card initialisation screen, and I'm seeing a "Scorpion BIOS" message, so it looks like Scorpion graphics card has it's own BIOS? Need to check other cards to see if they have a similar message?
Will attach photo of Scorpion BIOS screen.
Trident ISA Graphics card didn't have similar BIOS message, photo attached.

rasz_pl wrote on 2023-11-11, 02:49:

is it still there with Industrial Vision card?

No, this only appeared once using the ISA graphics card. After changing the ISA Video ROM setting, message is gone completely, no matter which graphics card is plugged in.

rasz_pl wrote on 2023-11-11, 02:49:

This is one of the outcomes I was hoping for 😀, but it still makes no sense, have you verified that changing 'ISA Video ROM C000, 32k' to ENABLED makes the graphic output corrupted again?

OK, just done that. Changed ISA Video ROM setting back to Enabled, and saved BIOS settings. Then machine seemed to boot normally?

I powered down and re-booted. Now Scorpion graphics card is not working again. I will attach a video of POST card. https://vimeo.com/883701857?share=copy
Sorry video is sideways. Had to film with my phone and can't work out how to rotate. 😀

Venture wrote on 2023-11-10, 19:03:

2. I want to replace old HDD with CF card. I have a couple of different CF card caddies and 2.5" to 3.5" IDE adapter.
Next step is to try and auto detect the CF card as a slave and see if I can get it to boot? Actually just done this, it doesn't.

rasz_pl wrote on 2023-11-11, 02:49:

You can try disabling IDE Block mode, but usually its due to CF cards IDE mode not being fully tested and IDE master/slave being very awkward https://www.os2museum.com/wp/the-dual-drive-ide-hell/ (one drive has to pretend to be both of them). You can try the other way around - set CF as master, IDE as slave, or image HDD and copy it to CF in another computer, or few other CF cards until you find one working. Removing file system from CF card might also help.

I'm going away on a work trip for two weeks later today, so I'll have to tackle this CF card thing when I get back.
FYI, I did try reading the original HDD in a modern Win 11 machine using a caddy. The machine didn't even see the HDD. I tried to use a HDD cloning program called "HDD Raw copy tool" from www.hddguru.com. This time it saw the old HDD, but when I tried to copy to a file it came up with loads of errors (bad sectors/ unreadable etc. ) which I think are false because the HDD is working fine (for now!).

Venture wrote on 2023-11-10, 20:18:

Note how they used a conventional linear power supply with huge transformer instead of SMPS, for reliability. The whole weighs maybe 25kg.

rasz_pl wrote on 2023-11-11, 02:49:

is it a fully linear supply 😮 or an isolation transformer in front of SMPS?

No, fully linear. It has big banks of capacitors and 4-5 different linear regulator boards for +12, 5V, -5V etc etc.

Venture wrote on 2023-11-10, 20:18:

Is it safe to try two graphics card plugged in at the same time, for example Scorpion & ISA or Scorpion & PCI? You mentioned they might work together, so if this failure happens again in the future I can just plug in a spare graphics card & change the BIOS video settings without removing any other cards or disturbing anything.

rasz_pl wrote on 2023-11-11, 02:49:

No is my guess. ISA VGA cards are hardcoded to certain addresses (0xB8000 for text buffer etc) and there is no was of dynamically managing it. With two PCI VGA cards you can assign different hardware windows, but BIOSes werent equipped for it, Windows was able to handle couple cards at once but even then your only control over which card is primary is placement order in expansion slots.
CGA card maybe, but that still leaves the question of switching between them and CGA is kind of useless without CGA monitor anyway.
The question remains what is up with your industrial card, is it an ordinary VGA card with VGA bios at C0000 or maybe it has custom bios pretending to be something else. https://archive.org/details/checkit_30 with Vision card plugged will tell us what it identifies as, 'Base memory map' will show where Video memory and bios are mapped.

Just downloaded the checkit program and realised I have no way of getting it on this machine! The pick & place machine has a USB/ Floppy emulator (unfortunately I can't easily remove that) but this spare machine has no peripherals apart from old HDD & Keyboard. Once I get CF card working, it may be useful to try that.

rasz_pl wrote on 2023-11-11, 02:49:

Its totally safe to plug both PCI and ISA VGA cards together, but no idea which one will be primary, most likely ISA with no option of switching to PCI.
I would start by playing with "ISA Video ROM C000,32K" option, Im not convinced it was the culprit. Also try different "Primary Display Monitor" settings trying to recreate fault condition.
My weak theory now is industrial Vision card bios not handling CGA or Monochrome modes and Pentium SBC having a jumper on it somewhere defaulting to one of those options with empty CMOS. Once booted and in BIOS Setup VGA/EGA is detected and saved.

rasz_pl wrote on 2023-11-11, 02:49:

Btw are still weird 'FF, CD, CB' POST codes showing up?

I have attached video above of POST card with Scorpion card failing to work. POST codes don't look the same as they did before, nor do you get garbage on the screen, just a flashing cursor this time.

Venture wrote on 2023-11-12, 12:53:

It was always VGA/EGA, and I've never changed it.

Hmm, where does "CMOS Display type mismatch" message come from then? Its possible BIOS itself fixes this setting as soon as you enter it. EDIT: Thats exactly what happens, I just tried to set Pentium/AMI computer to monochrome and 'color 40x25' and in both cases it complained on next reboot but in bios it was already automagically fixed to VGA.

Venture wrote on 2023-11-12, 12:53:

it looks like Scorpion graphics card has it's own BIOS?

Yes, every VGA card has to have its own bios, CGA/MDA cards work without one. Since your Scorpion Vision card is industrial there was slight possibility it might be doing something non standard like using 'Option ROM'.

Venture wrote on 2023-11-12, 12:53:

Trident ISA Graphics card didn't have similar BIOS message, photo attached.

you attached a photo of Trident BIOS message 😀

Venture wrote on 2023-11-12, 12:53:

OK, just done that. Changed ISA Video ROM setting back to Enabled, and saved BIOS settings. Then machine seemed to boot normally?
I powered down and re-booted. Now Scorpion graphics card is not working again. I will attach a video of POST card. https://vimeo.com/883701857

Bizarre. One possibility I can think of is Scorpion card memory mapping IO inside Video bios address space (C0000), or having multi banked Bios?
I just rechecked with another Pentium AMI bios and starting D2 D3 D4 POST codes are normal, weird that https://mrbios.com/techsupport/award/postcodes.htm doesnt include them under AMI.
FF should mean board is finished initializing and is starting to boot, that would suggest Scorpion Video bios executed fine, but somehow fails to fully initialize VGA hardware (two beeps = VGA ram check error)? but in that case it should stop at 2D and not FF 😒
I was guessing screen garbage again, but

Venture wrote on 2023-11-12, 12:53:

nor do you get garbage on the screen, just a flashing cursor this time.

so garbage on screen (and weird C1 Post codes?) were the result of glitchy backplane?
Blinking cursor means Video bios at least started initializing VGA card.
Cant hear ram test so it doesnt seem to continue booting! That one might be because with "CMOS Display type" set to VGA wont continue if Video bios fails, but with empty CMOS Bios usually defaults to safest option and will continue producing garbage.
Whatever the case 'ISA Video ROM C000,32K' is the key, now you can stash spare ISA VGA card inside the PC Case in case CMOS loses contents again.

Venture wrote:

Win 11 machine using a caddy. The machine didn't even see the HDD.

modern computer BIOS/Win11 might refuse to see HDDs not supporting LBA (<~520MB).

Venture wrote:

"HDD Raw copy tool" came up with loads of errors (bad sectors/ unreadable etc. )

No ideas about this one.

Venture wrote:

Just downloaded the checkit program and realised I have no way of getting it on this machine!

serial cable will work DDLINK: Easily move files between/To/From DOS systems can bootstrap itself over the cable Re: DDLINK: Easily move files between/To/From DOS systems

That is exactly why I like simple video cards that is compatible. Good thing you did buy two.

In Gona's compatibility list, the C&T C452 is somewhat compatible with DOS. Use one of either two ISA video cards that was purchased instead.

Cheers,