Some PCI video cards are keyed for universal PCI slots but require 3.3V. If you put them in an old 5V PCI motherboard they won't work.

Don't know if that's your issue, but the fact it works in newer mobos with 3.3V PCI slots makes me suspect it.

kingcake wrote on 2024-03-06, 00:04:

Some PCI video cards are keyed for universal PCI slots but require 3.3V. If you put them in an old 5V PCI motherboard they won't work.

Don't know if that's your issue, but the fact it works in newer mobos with 3.3V PCI slots makes me suspect it.

I thought something like that initially, but my ATI card is keyed for 5V, the same as the motherboard. The other PCI cards I've tried that do work with my G7S620 also have 5V keys.

fosterwj03 wrote on 2024-03-06, 00:50:

kingcake wrote on 2024-03-06, 00:04:

Some PCI video cards are keyed for universal PCI slots but require 3.3V. If you put them in an old 5V PCI motherboard they won't work.

Don't know if that's your issue, but the fact it works in newer mobos with 3.3V PCI slots makes me suspect it.

I thought something like that initially, but my ATI card is keyed for 5V, the same as the motherboard. The other PCI cards I've tried that do work with my G7S620 also have 5V keys.

To repeat, just because it has the key for 5V doesn't mean it will work on 5V. Some still need 3.3V. The Rage XL PCI cards you get from ebay are all like this.

Agree ! Some ATI PCI cards will not work proper on some older PCI boards, the Rage XL series is one.....ATI Rage XL 8MB on 486 PCI

Horun wrote on 2024-03-06, 01:08:

Agree ! Some ATI PCI cards will not work proper on some older PCI boards, the Rage XL series is one.....ATI Rage XL 8MB on 486 PCI

Yep, this was the exact issue I was referencing. It's usually pretty easy to bodge an AMS1117-3.3V LDO on the card itself to get it working. The AMS1117-3.3 is highly recommended because it doesn't require external protection diodes and is stable w/o external capacitors. Although, oddly, most of the ATI cards affected do have the 5V decoupling caps despite not having the regulator.

Is the 3.3V issue true for the Mach32? This was a chipset designed for multiple buses.

I'm looking for a Mach32 pin-out right now, but I'm not having much luck finding a datasheet. On the other hand, if I'm interpreting the edge connector properly, it doesn't look like the 3.3V contacts connect to anything.

fosterwj03 wrote on 2024-03-06, 01:21:

Is the 3.3V issue true for the Mach32? This was a chipset designed for multiple buses.

I'm looking for a Mach32 pin-out right now, but I'm not having much luck finding a datasheet. On the other hand, if I'm interpreting the edge connector properly, it doesn't look like the 3.3V contacts connect to anything.

That I don't know. I just know your exact issue does affect some ATI PCI cards. The power section will be external to the chipset.

This is about the time that I wish I had a PCI analyzer card that would show if the system is providing the correct voltages.

Unfortunately, I don't have any other ATI PCI cards of similar vintage to test. The next oldest I have is a Radeon 9000, which just worked in the board a moment ago.

possible reasons
- no 5V voltage while card hardwired for 5V
- no 3.3V voltage while card hardwired for 3.3V
- corrupted video bios
- only legacy video bios on UEFI platform with no/off legacy support
- pci incompatibility? maybe on some very rare bad designs? Certainly not when talking Intel chipset, but Mach32 was ATIs first PCI implementation so maybe?

Does it show in the bios PCI device initialization list? In windows device manager?

fosterwj03 wrote on 2024-03-06, 01:40:

This is about the time that I wish I had a PCI analyzer card that would show if the system is providing the correct voltages.

Unfortunately, I don't have any other ATI PCI cards of similar vintage to test. The next oldest I have is a Radeon 9000, which just worked in the board a moment ago.

Power up the board with it flipped over. Probe the +3.3V solder joints on one of the PCI slots. Make sure you pick one of the +3.3V rails, not the +3.3V I/O. If it has +3.3V then the issue is elsewhere.

rasz_pl wrote on 2024-03-06, 02:00:

possible reasons - no 5V voltage while card hardwired for 5V - no 3.3V voltage while card hardwired for 3.3V - corrupted video b […]
Show full quote

possible reasons
- no 5V voltage while card hardwired for 5V
- no 3.3V voltage while card hardwired for 3.3V
- corrupted video bios
- only legacy video bios on UEFI platform with no/off legacy support
- pci incompatibility? maybe on some very rare bad designs? Certainly not when talking Intel chipset, but Mach32 was ATIs first PCI implementation so maybe?

Does it show in the bios PCI device initialization list? In windows device manager?

OK, so I gave the Mach32 another try to see if the BIOS listed two graphics adapters in the device list, and the computer booted with the ATI card as the display devoice. Unfortunately, the BIOS showed corrupted characters.

I reseated the card, and the BIOS again booted with the Mach32 as the display device with good characters. I then tried to boot into OS/2 1.3 which I have configured for the Mach32. It booted to the GUI, but the colors were corrupted. I reseated the card multiple times and in different slots, but I got the same corrupted colors each time.

fosterwj03 wrote on 2024-03-06, 02:40:

rasz_pl wrote on 2024-03-06, 02:00:

possible reasons - no 5V voltage while card hardwired for 5V - no 3.3V voltage while card hardwired for 3.3V - corrupted video b […]
Show full quote

possible reasons
- no 5V voltage while card hardwired for 5V
- no 3.3V voltage while card hardwired for 3.3V
- corrupted video bios
- only legacy video bios on UEFI platform with no/off legacy support
- pci incompatibility? maybe on some very rare bad designs? Certainly not when talking Intel chipset, but Mach32 was ATIs first PCI implementation so maybe?

Does it show in the bios PCI device initialization list? In windows device manager?

OK, so I gave the Mach32 another try to see if the BIOS listed two graphics adapters in the device list, and the computer booted with the ATI card as the display devoice. Unfortunately, the BIOS showed corrupted characters.

I reseated the card, and the BIOS again booted with the Mach32 as the display device with good characters. I then tried to boot into OS/2 1.3 which I have configured for the Mach32. It booted to the GUI, but the colors were corrupted. I reseated the card multiple times and in different slots, but I got the same corrupted colors each time.

interesting, no corruption when you put this card in another mobo, right?

Correct, it works properly on my DFI G4V620.

fosterwj03 wrote on 2024-03-06, 02:40:

OK, so I gave the Mach32 another try to see if the BIOS listed two graphics adapters in the device list, and the computer booted with the ATI card as the display devoice. Unfortunately, the BIOS showed corrupted characters.

Hmm maybe the "two graphics" is the issue ?

I don't have much more time to troubleshoot tonight, but I might try cleaning the card's contacts tomorrow to see if that helps. They don't look bad, though.

fosterwj03 wrote on 2024-03-06, 02:40:

OK, so I gave the Mach32 another try to see if the BIOS listed two graphics adapters in the device list, and the computer booted with the ATI card as the display devoice. Unfortunately, the BIOS showed corrupted characters.

I reseated the card, and the BIOS again booted with the Mach32 as the display device with good characters. I then tried to boot into OS/2 1.3 which I have configured for the Mach32. It booted to the GUI, but the colors were corrupted. I reseated the card multiple times and in different slots, but I got the same corrupted colors each time.

do any other PCI cards work fine in that motherboard?

Try thoroughly cleaning the PCI slot (alcohol/toothbrush). I had an audio card that wouldn't work initially in the same PCI slot that another PCI card would work in.

I cleaned the card's contacts pretty throughly, and now the card initializes at boot reliably. Unfortunately, the color corruption still occurs in super VGA graphics modes (8-bit color mode specifically).

I ran some ATI diagnostics on the card, and it's finding memory errors. I'll need to run diagnostics on the card in another motherboard, but I don't remember having these issues on the board I normally use with this card.

I still wonder if I have some sort of PCI bus timing issue in the G7S620. It's weird that my other PCI cards don't have any problems on this board if that was the case.

Can you post some good pics of the front and back of the card?