286s were "AT" era - hard drives were MFM with two "ribbon" cables (a wider one and a thinner one) from the (relatively dumb) hard drive controller to the actual drive.
IDE and "on board" conrollers didn't appear till later.
This is correct. Most 80286 BIOSes from the 80s do not know about IDE/EIDE or AT-Bus drives yet.
An 80286 BIOS of that vintage does act as if an WD1003 compatible fixed-disk controller is in the system.
Which in turn has no support for the IDENTIFY DRIVE command yet.
So while AT host controllers and IDE drives can communicate with an 286 BIOS,
it's because the IDE drive itself will talk the WD1003 language.
It emulates the old fixed-disk controller, basically.
However, this only works so far.
If the BIOS is too old/the IDE drive too new, then problems happen.
In such a case, it's an option to use a dedicated IDE BIOS and disable HDD support in 286 BIOS (HDD=none).
Edit: That's basically the idea of using XTIDE Universal BIOS (XUB), a network card and a simple IDE host adapter (aka "IDE controller").
The AT version of XUB will provide the HDD routines originally provided by the 286's System BIOS.
Then, the IDE port of, say, an ISA multi-i/o card, an ISA "IDE controller " (IDE host adapter) or an ISA IDE CD-ROM "controller" will be used.
Because, XUB can boot from all master channels. Not just primary master.
A cheap, tiny IDE CD-ROM controller card (secondary master) will do just fine.
A network card holds the EPROM prpgrammed with XUB.
If XUB was configured to be in auto-detect mode, then no settings must be changed.
Last edited by Jo22 on 2026-07-21, 16:21. Edited 1 time in total.
"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel
286s were "AT" era - hard drives were MFM with two "ribbon" cables (a wider one and a thinner one) from the (relatively dumb) hard drive controller to the actual drive.
IDE and "on board" conrollers didn't appear till later.
This is correct. Most 80286 BIOSes from the 80s do not know about IDE/EIDE or AT-Bus drives yet.
An 80286 BIOS of that vintage does act as if an WD1003 compatible fixed-disk controller is in the system.
Which in turn has no support for the IDENTIFY DRIVE command yet.
So while AT host controllers and IDE drives can communicate with an 286 BIOS,
it's because the IDE drive itself will talk the WD1003 language.
It emulates the old fixed-disk controller, basically.
However, this only works so far.
If the BIOS is too old/the IDE drive too new, then problems happen.
In such a case, it's an option to use a dedicated IDE BIOS and disable HDD support in 286 BIOS (HDD=none).
Which the card in question, *HAS*.
It has a 32kb option rom living at D800, which supports ATA-2 command set, abd LBA-28 addressing. It knows (per the manual) about multiword dma modes.
Last edited by wierd_w on 2026-07-21, 16:25. Edited 1 time in total.
Does this mean the computer is unusable now because of this or is it still usable
Can I still use the sd card adapter and sd card or are they unusable now as well
It's usable, with some things to note:
Floppy disk drive types will likely default to single drive, low density.
System date and time will default to midnight, some year in the 1980s.
It will complain, and need you to press F1 on every boot.
The ide option rom does autodetection. It should fill out the hard disk drive table, and supply a 'modern' (386/486 era) int13 handler that knows about 28bit lba, and echs on IDE drives.
It has a 32kb option rom living at D800, which supports ATA-2 LBA modes.
Alright, I don't doubt that.
But does it work in tandem with an 80286 CPU or 80286 BIOS?
I do have some IDE Enhancer cards here, for example, but I never tested them in a 286 PC.
So I don't know if they do work with 32-Bit addresses etc.
I mean, MS-DOS 7.0 didn't work on my 286 PC, either when I tried once.
It gave an error message about requiring a 386.
For some reason, some disk routines must have become 32-Bit by then.
Maybe due to LBA support and support for extended int13h routines?
Something else wouldn't make much sense for a disk operating system, I suppose.
"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel
unsure. The datecode for the optionrom suggests late 386 era. Impossible to say if it expects 32bit cpu addressing or not without dumping it, and looking inside.
The suggestion I gave for this card, was with the intention of pulling the socketted ROM off, and replacing it with AT flavor XTIDE XUB inside a 32kb eeprom.
However, OP has no means of flashing such a chip to put in the socket.
I was just wondering I removed the jumper off rom on my ide data controller card
So the sd card and ide adapter would work but do I need to put the jumper back to rom
On the ide data controller card if I want to use a hard drive that I bought
Does this mean the computer is unusable now because of this or is it still usable
Can I still use the sd card adapter and sd card or are they unusable now as well
The computer should still work with a dead Dallas chip battery. However, it won't save any BIOS settings which means you have to reset everything manually every single time you turn on the computer.
A permanent fix would involve desoldering the existing Dallas chip and replacing it with a modern Dallas clone, which are available from Ebay.
Insofar as your issues trying to get fixed storage fixed, I again recommend getting a Lo-tech XTIDE card and using that in conjunction with an appropriate Compact Flash card. I use genuine Sandisk or Cisco Industrial cards in the 64MB to 256MB sizes for systems of this era. As I said before, I've set up multiple Epson 286 systems and other 286 computers with these configurations and can confirm this works.
The good news is that these are 360k disks, which should be the default that the system assumes if it's CMOS has not been setup,
meaning you should be able to use them
The "Reference" disk appears to be bootable (note the presence if IBMBIOS.COM, IBMDOS.COM and COMMAND.COM) and contains the SETUP.EXE program.
You should be able to boot this disk and run SETUP to get the CMOS setup properly configured (If you have not done so already, make sure the CMOS battery is good)
The "trick" will be getting these disks created (I can think of a couple ways)
- If you have a "modern" (ie: Winblows) with a 5.25" drive (you could "borrow it from the Emson if you have to), you can use DSKWRITE
- If the system has enough RAM to create a RamDisk and a serial port, you could "bootstrap" my DDLINK to it, then use that to move any other tools you might need (like RamDisk and editor to modify
CONFIG.SYS and AUTOEXEC.BAT), my XDISK.COM to write the image and the image itself
If you don't already have these disks, I can try and make them available somewhere as well as any tools you might need to create disks.
https://dunfield.themindfactory.com ; "Daves Old Computers" ; SW dev addict best known:
ImageDisk: rd/wr ANY floppy PChw can ; Micro-C: compiler for DOS+ManySmallCPU ; DDLINK: simple/small filecopy(w/o netSW)via Lan/Lpt/Com
(Dave conspicuously does not privide images for the disks. Giggle)
I almost want to ask to have the sticker peeled off the rom on this card so I can see which flavor eeprom should live on it, then burn and send an xtide chip to this guy.
The computer should still work with a dead Dallas chip battery. However, it won't save any BIOS settings which means you have to reset everything manually every single time you turn on the computer.
A permanent fix would involve desoldering the existing Dallas chip and replacing it with a modern Dallas clone, which are available from Ebay.
Yeah, I forgot that it's a Dallas chip. As I mentioned before, you CAN fix them:
It's best to completely remove the old cell - but I'm lazy, so I usually:
- grind to a nice battery cell sized circle above the existing battery and down the side to expose the other lead.
- disconnect that lead add a fine wire long enough to reach around to the top - strip & tin maybe 1/2 inch
- place the new cell upside down on top of the existing one
- cut a strip of thin plastic to fit over the top of the chip and put a small hole in it over the center of the new battery
- Place tinned wire through the hole and coil to make upper connection.
- Tape the "new top" down with "black tape" (which is springy enough to hold connection with a bit of pressure)
You *could* probably do this with the chip in place. but I'd desolder it and put a socket in for easy future repair (IIRC it's not as many pins as a usual 28pin IC)
I've had at least 1/2 dozen of these repairs (mostly to Dallas NVRs but a couple RTCs, some for quite a few years and I've never seen the original battery underneath leak (but do check it from time to time)
https://dunfield.themindfactory.com ; "Daves Old Computers" ; SW dev addict best known:
ImageDisk: rd/wr ANY floppy PChw can ; Micro-C: compiler for DOS+ManySmallCPU ; DDLINK: simple/small filecopy(w/o netSW)via Lan/Lpt/Com
(Dave conspicuously does not privide images for the disks. Giggle)
I just searched "Epson PC AX2 setup reference disk"
And found someone who said "For anyone landing here in the future, I have uploaded the 4 system disks to Internet Archive:"
with a link that went here: https://archive.org/details/epson-pc-ax-2-startup
(Which fwiw does have picture of the diskettes)
One of the download options says "ISO", but it's actually 4 diskette image files (.IMG)
I grabbed them, and used my FDI (FlopppyDiskImage) tool to list the content which is what I posted.
I was offering to package it up with the tools you might need, but you can get the disk images at the above link, and DSKWRITE.EXE and XDISK.COM (and other tools you might need) from DOSBOOT.ZIP on my site
Last edited by DaveDDS on 2026-07-21, 20:10. Edited 1 time in total.
https://dunfield.themindfactory.com ; "Daves Old Computers" ; SW dev addict best known:
ImageDisk: rd/wr ANY floppy PChw can ; Micro-C: compiler for DOS+ManySmallCPU ; DDLINK: simple/small filecopy(w/o netSW)via Lan/Lpt/Com
unsure. The datecode for the optionrom suggests late 386 era. Impossible to say if it expects 32bit cpu addressing or not without dumping it, and looking inside.
The suggestion I gave for this card, was with the intention of pulling the socketted ROM off, and replacing it with AT flavor XTIDE XUB inside a 32kb eeprom.
However, OP has no means of flashing such a chip to put in the socket.
"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel
unsure. The datecode for the optionrom suggests late 386 era. Impossible to say if it expects 32bit cpu addressing or not without dumping it, and looking inside.
The suggestion I gave for this card, was with the intention of pulling the socketted ROM off, and replacing it with AT flavor XTIDE XUB inside a 32kb eeprom.
However, OP has no means of flashing such a chip to put in the socket.
