It's interesting that the error message shows "expected" and "actual" both being FFh. This means that probably the value did not read as 0FFh at first, but when the location got re-read, it had the correct value. Getting memory errors at those addresses doesn't make any sense. CGA and MDA cards are not supposed to respond to those memory addresses. It looks like something goes wrong on the system bus with Hercules and CGA. I see on your photo that you are using one of the more modern HGC clone cards, that possibly also supports CGA. In case you use this card for both Hercules and CGA: Are you 100% positive that this card is good?

If you have issues with multiple cards, likely there is some fault on the mainboard. CGA and MDA cards add a lot of waitstates if you try to access video RAM at the wrong point in time, VGA cards (especially modern ones) are much quicker. If your board has a broken wait-state line ("IOCHRDY"), this might make CGA and MDA cards fail.

mkarcher wrote on 2022-10-28, 23:07:

It's interesting that the error message shows "expected" and "actual" both being FFh. This means that probably the value did not read as 0FFh at first, but when the location got re-read, it had the correct value. Getting memory errors at those addresses doesn't make any sense. CGA and MDA cards are not supposed to respond to those memory addresses. It looks like something goes wrong on the system bus with Hercules and CGA. I see on your photo that you are using one of the more modern HGC clone cards, that possibly also supports CGA. In case you use this card for both Hercules and CGA: Are you 100% positive that this card is good?

If you have issues with multiple cards, likely there is some fault on the mainboard. CGA and MDA cards add a lot of waitstates if you try to access video RAM at the wrong point in time, VGA cards (especially modern ones) are much quicker. If your board has a broken wait-state line ("IOCHRDY"), this might make CGA and MDA cards fail.

Thank you for your interest.

I also have two original IBM CGA cards.
It's brown and green color long PCB.
These two cga cards have the same problem.

So is the motherboard's basic design wrong?

Or is it a Bios problem?

shimart wrote on 2022-10-29, 07:41:

So is the motherboard's basic design wrong?

I trust Sergey to have designed that board correctly. I'm more inclined to think about a fault in your specific build of that board, like a broken trace or a bad solder joint. Specifically, I suggested that the issue might be rooted in the connection of the signal typically called "IOCHRDY". It's called "IO_READY" on the ISA slots in Sergey's schematics (It's ISA pin A10, or pin 41 according to the numbering scheme used in the schematic) and "READY" on the Wildcard module (pin 40). I expect that a bad contact on this signal is a possibly cause for the symptom you are observing.

shimart wrote on 2022-10-29, 07:41:

Or is it a Bios problem?

As long as you use a BIOS that is meant to be used with the Wildcard 88 module, I'd expect this is a hardware problem, not a software problem.