Try running this debug session without any device drivers loaded.

wierd_w wrote on 2025-07-18, 17:48:

Try running this debug session without any device drivers loaded.

Everything removed from both autoexec and config...

I am seeing what looks like corruption, or somthing walking on that address.

I see what looks like a mangled copyright notice, but also what looks like a dos device driver handler designator...

I'm not sure what's going on here if this is a clean boot with no device drivers or memory managers.

wierd_w wrote on 2025-07-18, 18:02:

I am seeing what looks like corruption, or somthing walking on that address.

I see what looks like a mangled copyright notice, but also what looks like a dos device driver handler designator...

I'm not sure what's going on here if this is a clean boot with no device drivers or memory managers.

I have removed all the entries in autoexec.bat and config.sys. It is DOS 6.22, moterboard is CHIPS GW286 CT, with EMS driver installed, sound blaster CT2270,.....
I don't remember, but i maybe installed DOS with all the configurations of YU language...

EMS for C&T uses the chipset shadowing functionality.

It could in theory be the culprit.

Really, do the debug run with NOTHING but dos present.

wierd_w wrote on 2025-07-18, 18:51:

EMS for C&T uses the chipset shadowing functionality.

It could in theory be the culprit.

Really, do the debug run with NOTHING but dos present.

So i need to do clean install....

Hold F5 on boot.

Same result i think ...

If this is C&T AT based board, can you show what the 'advanced system setup' / 'chipset features setup' looks like?

It may be prudent to disable shadowing between B000 and CFFF.

There are 3 screens in BIOS setup....

Turn ems function off, and boot holding F5 down.

Just in the off chance its putting the pageframe in an odd spot.

wierd_w wrote on 2025-07-18, 20:00:

Turn ems function off, and boot holding F5 down.

Just in the off chance its putting the pageframe in an odd spot.

same again...

So the good news is: The EGA card seems to be alive. It responds to BIOS reads at C000:0 and it responds to I/O port reads at 3C2. The bad news is: The BIOS reads do not work reliably. I don't think any of the changes you made (mono/color jumper, booting with F5) cause the slightly different debug output, but instead there is a hardware problem. Just to make sure that there is no ROM conflict, please remove the EGA card from the system (without changing anything else) and then re-run the D C000:0 command. If you get anything else than FF only, there is a conflicting device at C000:0. I consider this unlikely, because if I understood you correctly, you could use a VGA card in the same system without issues, but nevertheless if you get FF only without the EGA card installed, we can be sure the issue is located on the EGA card.

The expected output if the EGA were working correctly is:

1C000:0000  55 AA 20 E9 FF 1D FF FF-DB 32 32 30 33 30 2D 43  U¬ Ú .  ¦22030-C
2C000:0010  31 32 30 31 20 20 20 20-07 07 FF FF FF FF 49 42  1201    ..    IB
3C000:0020  4D 20 43 4F 4D 50 41 54-49 42 49 4C 49 54 59 20  M COMPATIBILITY
4C000:0030  20 30 43 4F 50 59 52 49-47 48 54 20 50 41 52 41   0COPYRIGHT PARA
5C000:0040  44 49 53 45 20 53 59 53-54 45 4D 53 2C 20 49 4E  DISE SYSTEMS, IN
6C000:0050  43 2E 20 31 39 38 36 20-50 45 47 41 20 42 49 4F  C. 1986 PEGA BIO
7C000:0060  53 20 00 40 20 00 C0 00-00 00 00 00 00 00 00 00  S .@ .+.........
8C000:0070  00 00 00 00 00 00 00 00-00 00 00 00 00 00 00 D0  ...............ð

Your output looks obviously quite different. It's interesting: It seems that D3 (the data bit with the value 08) is often wrong, resulting in A3 (the address bit with the value 08) instead. The same is true for D5 that often returns A5 instead of the ROM data value. The D5/A5 issue causes the lines with 0020/0030/0060/0070 to only contain lowercase letters, whereas the lines with 0000/0010/0030/0040 only show uppercase letters. Looking at ROM location C000:0001, where the expected result is AA, but the actual result is 00, it seems like all the odd data bits are unreliable and forced to 00 at that location. That would be D1 and D7 as well. In fact, A7 is zero in the whole address area you show, and D7 is zero way too often (but not always).

This makes me wonder whether some address/data multiplexing is going on on that card, and that one failed, causing a conflict between ROM data and ROM address on the read output. On an 8088 system, I would suspect issues with the address latches on the mainboard, but this is a 80286 system, and the 80286 does not use multiplexed address/data. It would thus be interesting to know whether the PEGA2A has multiplexed address/data input, and the PALs on the card maybe control '244/'245 chips for this multiplexing. If they get bad control signals, issues like this can happen.

Could it be that this fault is result of inserting this ega card in 16bit isa slot? This mbo had 2 8bit slots, but ega card is too long to insert it in this slot, because to the right from this slot are rams, and the card couldn't fit inside?

Without ega card

2 more pics, back side, and i forgot to tell - first time i had turned this card on, those 2 highest tantalum exploded, but that is pretty normal on such old hardware, right? They couldn't make any significant damage?

Old tantalum caps shorting and exploding when power is applied is "normal". In the sense that back in the day when these were made the quality control on tantalum caps was way worse than what we have now. Some of them simply aged badly, cracked, etc, and that is the result. I've had SMD tantalum blow up on me, and not even on the first power-on after years but after several tries. So long there isn't any copper damage or big black spot of residue left on the PCB you should be fine.

As for the problem - it is interesting how the data seems to be almost what's in the EPROM but glitched. Bits 3 and 5 especially but some bytes are completly wrong. What else do you have in that PC? Any other cards with EPROMs, like XTIDE or something of the sort?

Also, I spy a 20MHz 286 CPU there? But your BIOS settings say that ISA bus divider is only /4 which would make it 20*2/4 = 10MHz. That's a big overclock for an old card like this. The EPROM and PALs there might simply be too slow. Try /5 divider in BIOS, or even /6 if you have such option. It should be 8MHz with /5 but if the EPROM degraded it'd be very slow now, so this is just to test.

EDIT: I mean the PROCCLK/4 setting. See if there is /5 at least for that CPU speed.
EDIT2: BTW the F5 key to bypass startup files, that's Win9x specific, isn't it? For DOS 6.x you need to press and hold left SHIFT key to bypass the files. If you are using XTIDE then preferably make a clean, bootable floopy with minimal files - just what you need to run the tests, like DEBUG.EXE, no CONFIG.SYS and only PROMPT $p$g in the AUTOEXEC.BAT. Then disconnect the HDD, remove XTIDE and boot from floppy only. This will reduce the number of unknown factors that might be affecting the system.

Hi,
there is normal old HDD on IDE ISA controller (HMC741), no XT-IDE card inside. There is a Sound Blaster CT2770, and those 2 graphics cards.
Yes, it is Harris 20MHz CPU, but there is only /2 and /4 ISA bus divider 🙁 Than i probably need slower 286 for this "experiment", right?
Here is the picture with Shift pressed when booting....

butjer1010 wrote on Yesterday, 11:01:

there is normal old HDD on IDE ISA controller (HMC741), no XT-IDE card inside. There is a Sound Blaster CT2770, and those 2 graphics cards.
Yes, it is Harris 20MHz CPU, but there is only /2 and /4 ISA bus divider 🙁 Than i probably need slower 286 for this "experiment", right?
Here is the picture with Shift pressed when booting....

Hm. Then I don't understand what that divider setting does. Perhaps it is applied to 20MHz, not the 40MHz input clock. Then 20/4=5MHz (kinda slow for ISA) and 20/2=10MHz (overclocked again). Seems odd, maybe a different system would be better for testing this card.

That being said, the left SHIFT must be pressed and held all the time until DOS prompt appears. I don't think it worked properly since on the screen I still see That D instead of \, so MODE and/or DISPLAY.SYS still got loaded. Or better yet forget the keys - make a copy of AUTOEXEC and CONFIG, and then just delete everything inside. Make them empty. This way for sure nothing will get loaded. Unless this is already a change in the font ROM on your MDA/Hercules card?

Somehow the debug dump got even worse now. But lets try again with empty config/autoexec.

Deunan wrote on Yesterday, 12:06:

Hm. Then I don't understand what that divider setting does. Perhaps it is applied to 20MHz, not the 40MHz input clock. Then 20/4 […]
Show full quote

butjer1010 wrote on Yesterday, 11:01:

there is normal old HDD on IDE ISA controller (HMC741), no XT-IDE card inside. There is a Sound Blaster CT2770, and those 2 graphics cards.
Yes, it is Harris 20MHz CPU, but there is only /2 and /4 ISA bus divider 🙁 Than i probably need slower 286 for this "experiment", right?
Here is the picture with Shift pressed when booting....

Hm. Then I don't understand what that divider setting does. Perhaps it is applied to 20MHz, not the 40MHz input clock. Then 20/4=5MHz (kinda slow for ISA) and 20/2=10MHz (overclocked again). Seems odd, maybe a different system would be better for testing this card.

That being said, the left SHIFT must be pressed and held all the time until DOS prompt appears. I don't think it worked properly since on the screen I still see That D instead of \, so MODE and/or DISPLAY.SYS still got loaded. Or better yet forget the keys - make a copy of AUTOEXEC and CONFIG, and then just delete everything inside. Make them empty. This way for sure nothing will get loaded. Unless this is already a change in the font ROM on your MDA/Hercules card?

Somehow the debug dump got even worse now. But lets try again with empty config/autoexec.

I did removed every entry in both config/autoexec, but that didn't changed anything.
I removed Hercules/MDA card and returned VGA card.... The problem with Đ was in the card (see pic with VGA). Probably the card was from some Croatian computer, and the fonts are Croatian in this ROM.
I have made a dump from this Hercules ROM, don't know if it's good, i couldn't find B57604, so i tried dumping it choosing 27C256?!?