I have several systems using IDE to CF adaptors (286, 386 DX/SX, 486 DX) and I have not experienced such issues at all. All of my systems use different IO controllers (the 486 is on the motherboard).

At face value I suspect that the problem might be with your IO controller. Can you share which IO controller you're using? And if you're able to, try a different one.

PS: Forgot to mention, I never used XT2IDE, as that isn't required on a 286. I only use 512MB CF cards for DOS systems.

I had similar issues many times and it is more often than not the cf card, not the adapter (either XT2IDE or standar IDEl to cf -card) or the drive geometry. In many cases, only 128 or 256 MB card work properly. This can be a trial and error Odyssee (Here an example : Dec 466d2 CF Card trouble)

Swapped characters sounds like something very unpredictable is happening, like electrical interference on a bus, bad memory used in a buffer, or a failing flash device.

Like hard drives, CF’s were made over a long period of time from a lot of vendors, so it’s hard to generalize too much, but I hear some CF cards can intermittently behave unpredictably if your isa bus is running slower than 8mhz.

Chaos can happen when a Multi I/O card plus a CF-IDE adapter connects pins with a different meaning on the CF card directly to your machine's ISA bus

Pins 27 and 28 are the most suspect culprits: Plucking certain IDE pins for better CF-IDE compatibility

i have several xt and 286 system. i always have more success with a xt-ide card that connect to a ide hdd drive than cf card. i tried a lot of hdd from 20gb to 80gb, they are works. i formatted them with 4 partition of 2gb each in dos 6.22 and i installed several games and all my xt and 286 with the hdd connected to the xt-ide card read the hdd and games perfectly.

Maybe the ISA bus frequency is too high for the CF cards? Or timings are too tight?
With a passive adapter, they're sitting directly on ISA bus, after all.
In the days of PCI IDE Controllers they ran on a different speed.

Edit: Is the PC RAM okay? CheckIt v3 has a memory test, it can be run for hours.

Thanks for the help so far. i just ordered a bunch of 512MB CF cards for testing and also a PicoMEM if it all fails. I will do some testing in the next days and report my results here. 😀

Was the system stable before the original HDD died? What RAM are you using? The Fox M is known to behave badly (lockups) when 3-chip SIMMs are used, you have to use 9-chip ones.

quicknick wrote on 2025-09-15, 00:14:

Was the system stable before the original HDD died? What RAM are you using? The Fox M is known to behave badly (lockups) when 3-chip SIMMs are used, you have to use 9-chip ones.

Thanks for the tip! I’m actually using SIMM modules with 3 chips. I’ll try testing with different memory sticks or just the onboard chips.
The system was stable before, or at least I didn’t notice any issues during operation.
In the meantime, I’ve installed and tested a PicoMEM card from another system.
It seems to run a bit better with it, but I’m still encountering errors.
For example, when I make changes using the editor and save the file, the system sometimes freezes with a
'Cannot load COMMAND.COM. System halted.' message.

Okay, so I swapped out the RAM modules and also played around with the timings. At first, the system seemed to run more stable,
but then it froze again during a test with 'Spear of Destiny'. I noticed that the PicoMEM’s status LED stayed lit continuously,
as if the system had crashed while accessing the card. So I moved the PicoMEM to the first 8-bit ISA slot
and since then I haven’t experienced any more crashes. Of course, I still need to run more thorough tests, as I’m not fully confident yet.

One thing remains strange though: just for fun, I installed Windows 2.11, which works and boots up fine.
However, when I exit Windows, I get a weird color mess on the screen (screenshot attached).

I’ve already tried different graphics cards, but the issue persists.
Currently, I’m using a Trident TVGA8900D-R with the COLORON fix installed.
Has anyone seen this issue before?

My 8900D non-R doesn't even work with my 12MHz 286 Suntac mobo. ALE signal issues. That mobo is very picky and VGA cards that do work must be in the slot closest to driver chips, or else I get random glitches on screen like bad pixels or wrong attributes (color/background) in text mode. Depends on the (S)VGA too, some slower ones, more period correct, seem to work better.

286 is pretty tightly coupled to the ISA bus, and the mobos seem to omit resistor packs that were added in 386 and later, and serve as pull-ups and termination for the ISA signals. I suspect that's why it matters on these mobos where you install your card, the faster ones need to be close to driver chips (VGA, ext. memory cards). Even some manuals state that the graphics card needs to be installed in specific slot(s).

That roughly matches the experiences I’ve had with the system over the past few days.
Maybe I should try rearranging the ISA cards again.

Currently, there are four ISA cards installed:
Slot 1 holds the PicoMEM, Slots 2 and 3 are COM/IDE/Floppy controller cards, and Slot 4 currently has the Trident graphics card.

uniQ wrote on 2025-09-15, 09:54:

That roughly matches the experiences I’ve had with the system over the past few days. Maybe I should try rearranging the ISA car […]
Show full quote

That roughly matches the experiences I’ve had with the system over the past few days.
Maybe I should try rearranging the ISA cards again.

Currently, there are four ISA cards installed:
Slot 1 holds the PicoMEM, Slots 2 and 3 are COM/IDE/Floppy controller cards, and Slot 4 currently has the Trident graphics card.

By my experience with similar problem in the past, try a real hdd or xt-ide card that connect a hdd and maybe your problem will be resolved. i had so much trouble with cf card in the past. so i prefer hdd, even if i have to use ontrack disc manager to get it working with old system like a 286, 386, 486. otherwise when still not working, i use a xt-ide card that connect a hdd, that always works.

i think your problem is the cf card that is not working correctly on this 286, but i can be wrong

i have a few builds that dont work correctly with cf card and do problem similar to yours.

I will test the PicoMEM card with a micro sd card for now. If this won't work either, I might use the XT-IDE card with a proper IDE HDD.
Before I rearranged the ISA cards and the graphics card in their slots, I was frequently getting screen displays like the one shown in the screenshot.

uniQ wrote on 2025-09-15, 12:58:

I will test the PicoMEM card with a micro sd card for now. If this won't work either, I might use the XT-IDE card with a proper IDE HDD.
Before I rearranged the ISA cards and the graphics card in their slots, I was frequently getting screen displays like the one shown in the screenshot.

i just looked at your motherboard online. Push gently on each 640k memory chips. maybe one is badly inserted. i had this problem with a 8088 that get lock-up but not screen display glitch. Still it worth a try.

Thanks I'll give it a try! 😀

That looks like pin 27 might be connected in an unexpected manner on your ide controller.

Your CF is probably asserting "IOCHRDY" on pin 27 when it has data ready to read. "IOCHRDY" wasn't a thing when your controller was created. It didn't really come about until 1994.

Normally on old controllers, that pin wouldn't be connected, but since it was undefined at the time, sometimes it goes someplace unexpected. It's possible that the pin is connected in a way that makes the CF think that "IOCHRDY" will work, and when it tries to tell the 286 to wait, the 286 doesn't wait and reads invalid or incomplete data lines and then the CF goes on to scramble the data lines for the VGA card later when its data arrives. Or some other oddball series of events.

It might not have been an issue before if your old hard drive was older than 1994, or was aware of the potential issue and didn't do "IOCHRDY" with your controller.

Disconnecting pin 27 on your CF adapter might fix the issue. You can cut the pin, or maybe just bend a pin over so it doesn't go into the socket if you testing.

Plucking certain IDE pins for better CF-IDE compatibility

https://en.wikipedia.org/wiki/Parallel_ATA

I’m currently no longer using any CF card adapters, but instead a PicoMEM with a microSD card.
The error shown in the screenshot also occurred with this setup.

I just spent a few hours with the system and had no more lockups. So in the end it might have been the RAM in combination with the arrangement of the ISA cards. 😀
The PicoMEM works great in the system. 😀

Thank you all for the help!