When I started the system for the first time, I could hear the floppy drive seek at startup. This was the sign that the BIOS was at least able to detect it and command it.
I prepared a set of MS-DOS 5.0 install floppy disks and gave it a try.
The system booted! Awesome!

The MS-DOS installer asked me where to install the OS, but no hard drive was available.
OK, that's fair—I didn't even configure it in the BIOS. Let's have a look.

After a reboot, I accessed the BIOS and looked at the supported disk drives. Unfortunately, mine was not detected.
The geometry was not written on the disk label, and I didn't even know if it was working at all.
I tried several other disks that I knew to be working, with no luck.
CompactFlash adapter? No luck either.
I quickly realized that hard drive support would be something painful to tackle. I don't keep a lot of old IDE drives, and most of them were from the mid-90s and not supported by the BIOS.

I started to look at the XT IDE project (https://xtideuniversalbios.org/).

XT IDE is well known here on Vogons. It's a piece of software that aims to bring support for "modern" hard drives to systems as old as XTs. It is generally loaded as an option ROM and allows disks unsupported by the motherboard BIOS to be recognized. Like a RAID card that makes a disk array appear as a single disk and allows booting from it, XT IDE detects and supports modern disks and allows an old system to boot from them.

Using XT IDE requires a way to load the option ROM during boot. Looking online, there are several ways to achieve that—from dedicated cards (you need to build or buy one) to network card boot ROMs. Having several ISA network cards, I wanted to do it that way.

The network card itself is not enough. Of course, you need an E(E)PROM to put in the socket and a way to program this E(E)PROM. Fortunately, I have a box full of old E(E)PROMs.
With the help of (this topic), I knew that I needed a 16 KB ROM to store the "full" binary of XT IDE (there are versions requiring less ROM space, but with fewer features). I found several EPROMs of this type, but none of them were EEPROMs. So my ROM programmer would not be able to erase them. I needed to use a UV eraser.

I ordered a UV eraser from Amazon and a XGecu T48 programmer. In the meantime, I configured the ISA network card.

The card I used is a D-Link DE-220PT. It is a network card without jumpers that requires configuration from a DOS-based utility. I installed the card in another system and used the driver found on TheRetroWeb (https://theretroweb.com/expansioncards/s/d-link-de-220pt).

The settings are saved on the card in a dedicated EEPROM.

I received the EPROM UV eraser and the programmer a few days later. After erasing the EPROMs, the T48 was able to burn the XT IDE image. But I was unaware that the XT IDE image needed to be configured first. The XT IDE option ROM could not start. Exactly the same thing happened to the YouTuber Bits und Bolts in this video: https://www.youtube.com/watch?v=lnb6zamDI6A

This is very important: you MUST configure the XT IDE ROM with the utility, writing the settings in the image file, and ONLY AFTER this step, burn the ROM.
I used the 'atl' (AT Large?) image and left almost everything with the default values, except the number of IDE controllers (set to one instead of two).

I wrote the image on a new EPROM and it was able to detect my CompactFlash card.
Awesome!

In the BIOS, I disabled both disks and left them to be handled by XT IDE.
The next step would be to try to install MS-DOS 5.0 from the floppy drive to the disk.

After inserting the MS-DOS 5.0 floppy disk #1 in the floppy drive, I pushed the reset button. No drives visible in the POST, as expected since I left the HDD configuration empty in the motherboard BIOS. Shortly after, I saw the XT IDE message, detecting my CompactFlash card. I hit F2 for the XT IDE boot menu and I selected the floppy drive. The MS-DOS 5.0 installer started… but I was unable to select "hard disk installation". I was stuck instead with "install to other floppy".

I had an intuition and pressed reset again to access the motherboard BIOS. I added a random disk configuration in the motherboard and retried the operation. Reset again—disk not detected by the motherboard. No problem. F1, continue. On the XT IDE option ROM screen, I pressed F2 for the boot menu, booted from the disk, and back to the installer. The "hard disk installation" option was available. MS-DOS 5.0 was installing correctly and was able to boot to the hard drive (CompactFlash).

After the installation, I went back to the motherboard BIOS, cleared the hard drive configuration, pressed reset again, and had no problem booting to the DOS system. Apparently, the issue was only with the MS-DOS disk detection routine.

But pressing reset all the time was just a workaround to avoid turning off the system entirely. I avoided that because the BIOS settings would vanish as soon as the power would be turned off because of the Dallas chip battery being dead. Even by pressing the reset button, the motherboard still complained that something was wrong with a beep, and I had to press F1 to continue every time. Fixing this would be my next step for this restoration.

Dallas chips are well known among retro PC users for always being dead and needing replacement. Fortunately, it's well documented and easy to do.
I followed this documentation: https://8bitclub.com/2025/07/28/modifying-a-d … cr2032-battery/
I've read that some also cut the connections to the old battery to avoid the new CR2032 being drained by the old empty battery. I chose not to cut the connections. Maybe I'm wrong. If the battery is drained too quickly, I will try that, but I just soldered wires on the internal connectors after exposing them with a metal file.

I applied hot glue on the connector solder points to protect them and under the coin cell battery holder (scavenged from a dead motherboard).

I placed the Dallas back in the CPU board inside the system and powered it on. I configured everything properly in the motherboard BIOS, saved, then powered it off. I waited a few seconds and powered it back on. And everything worked properly. The date was maintained (funny to see 2026 on this system), no error beep, no F1 to press—the system booted to DOS without any human interaction. Great!

Until now, I mainly used the system outside of its case. Now that it's working, the next step will be to install it back in the case, but not before a thorough cleaning.

I used soap, a brush, a sponge, and hot water to clean it. After letting it dry for a while, I used a "magic eraser" for small scratches and to make it a little bit brighter. I also cleaned the cables, the boards, the metal parts, and the PSU (with a soft brush and alcohol when required). The case looked really great after this cleaning.

Manipulating the case gave me an idea. The system came with only one floppy drive, but there are two floppy drive slots. One will be empty. I could use some black tape to seal it and make it look like it had two drives. Or… I could integrate the CompactFlash card reader in the empty floppy slot. I really wanted to explore that idea. I knew what to do next.

The CompactFlash to IDE reader I have is one that can be mounted on an expansion slot to the back of the case. See picture.

What I found interesting was that I had two screw holes to attach it to the bracket. But these holes could also be used to mount the card reader flat on a surface. I removed the bracket and mounted the card reader on a piece of wood of the correct size: the width of a floppy drive. With proper measurements, I was able to align the card reader correctly with the floppy drive hole on the case.

I added black tape to the side to hide that the card reader does not take the full width of the hole. I also filled the ejection button hole with masking tape to make it look like a real button (almost). And the result looks OK to me.

There is one little drawback: the activity LED on the card reader is not visible and the activity LED on the case is not connected. That's something for later.

So at that point, the list of modifications was:

• New power supply
• XT IDE on a network card
• Modded Dallas chip
• Cleaned case
• Integrated CF reader in a floppy drive bay
• MS-DOS 5.0 installed

The system is usable and I used it a little while for tests.
I installed SimCity and The Secret Of Monkey Island to try it.

Two things were quite frustrating when using the system:

• No mouse
• PC Speaker sound

No mouse? Well, not exactly. I tried a serial mouse on the motherboard serial port and it worked (I'm using the CuteMouse driver https://cutemouse.sourceforge.net/), but the motherboard also provides a "mouse/joystick" connector which takes the physical form of a serial port. There is also a configuration option in the motherboard BIOS—"mouse/joystick"—with valid options being "Mouse," "Joystick," and "Disabled."
No luck with my serial mouse with CuteMouse on this port. The mouse worked properly on the regular serial port, but since this was the only available port, this was quite limiting for the system. This would have prevented the use of anything on the serial port along with a mouse.

I suspected this port to be a variant of a BusMouse port: https://en.wikipedia.org/wiki/Bus_mouse
Once again, the service manual came to the rescue with the schematics. It was obvious that the port was not connected to the same controller as the serial port. And the pin names of this mouse port matched the ones found on a BusMouse device: XA, XB, YA, YB, BR, BH, BL.

Using my regular mouse is not possible and there was another issue: this mouse connector was not soldered to the motherboard but attached to the old PSU with rivets, which I had replaced with a new one. Of course, there was the option to move it to a bracket, but I found a better option: replacing the "Winchester" disk card with a more integrated "Super IO" card, sporting IDE like the Winchester card but serial and parallel as well (optionally floppy and joystick ports). Fortunately, I had plenty of these around and I installed one in the system.

For the settings, I enabled COM2, LPT2, and IDE, while disabling COM3, the floppy controller, and the joystick port. Everything is configured through jumpers on the card, with the reference printed on the back of the PCB.

The "Super IO" card was detected by XT IDE without any issue and booted to DOS. From the system, I was able to use the mouse on both serial ports: the one from the motherboard and the one on the "Super IO". I added black tape to cover the hole in the case for the BusMouse port. Adding it back is still possible—the connector headers are on the motherboard—but I think that having two serial ports is more flexible than a BusMouse port.
And for the joystick port, well, the next step is to add a sound card, so there will be one soon.

To be clear, no, having a sound card on a 286 is not very period correct. But the same goes for a network card (with XT IDE, but the network card is still usable), or a CompactFlash reader. A Super IO is period correct, but not the one I installed. I'm definitely in the process of making this 286 flexible and usable like a modern system and not performing a strict restoration. But nothing irreversible (except the PSU, which was dead anyway).

I thought about using a homemade Covox Speech Thing as a sound device (https://en.wikipedia.org/wiki/Covox_Speech_Thing). But there are not a lot of supported games, and installing a proper sound card would be better.

Regarding period-ish correct sound cards, I have two options available: a Mediavision Thunderboard and a SoundBlaster Pro 2… I chose to use none of them, as they are part of my collection and I don't really want to use them permanently on a system for now. So I looked for another sound card that could be used with SoundBlaster Pro compatibility, not plug and play, and preferably no TSR program. I found an ES688 AudioDrive, a 1994 card, SB Pro compatible and with an OPL3 chip (a 1:1 copy on my board).

Configuring the sound card was not easy. The driver was a driver pack aimed to support multiple ESS sound cards: ES688 Audiodrive Full install for DOS on this page (https://www.vogonsdrivers.com/index.php?catid … menustate=63,53). Some of the supported cards being plug and play required configuration through a utility. The installer insisted on having the driver loaded by config.sys, and this prevented the system from starting.
With trial and error, I finally found that the driver in config.sys was not needed at all. But an initialization program was required, but not for all arguments—only IRQ and DMA. Changing the base address through the utility would freeze the system. So here is my setup:

• Jumpers on the card set to DMA 220, IRQ 5, DMA 1 (IRQ7 already used by LPT1)
• Driver pack installed
• Line from config.sys removed
• Custom config in autoexec.bat

autoexec.bat

1@ECHO OFF
2PROMPT $p$g
3PATH C:\DOS
4SET TEMP=C:\DOS
5MODE CON CODEPAGE PREPARE=((850) C:\DOS\EGA.CPI)
6MODE CON CODEPAGE SELECT=850
7KEYB FR,,C:\DOS\KEYBOARD.SYS
8C:\ESSUTIL\ESSCFG.EXE /V:1 /I:5 /D:1
9C:\ESSUTIL\ESSVOL.EXE /V:8 /L:8 /W:8 /M:0 /C:8 /S:8
10C:\DOS\CTM.EXE
11SET BLASTER=A220 I5 D1 T2

config.sys

1DEVICE=C:\DOS\HIMEM.SYS
2DEVICE=C:\DOS\SETVER.EXE
3DOS=HIGH
4COUNTRY=033,850,C:\DOS\COUNTRY.SYS
5DEVICE=C:\DOS\DISPLAY.SYS CON=(EGA,,1)
6FILES=10

After this configuration, the sound worked well with both digital and FM synth in games such as Wolfenstein 3D, The Secret of Monkey Island, and Dune.

The system is almost ready but there are still some very small details to fix, like the front panel LED. Let's work on that.

I vaguely remembered that when I used the system for the first time, the front led worked. At least for the floppy drive and the power status. Also, the first time I disassembled the system I found a two-pin connector attached to the HDD. I was pretty sure that this two-pin connector was the HDD activity LED.
When I configured the "Super IO" board I read the jumper settings printed on the back and it mentioned the HDD activity LED. I opened the system, unplugged the card and yes, there is an HDD activity LED connector on "J2"

After connecting the two pins, I powered up the system and the LED started to blink. Great! but the power LED was still off. I started to suspect something and I had to check with a multimeter.

5V—that's a lot for a small red LED. I'm not sure how much this LED can take, but it's burnt. Looking at the schematics it seems like the LED receives the "power good" signal. On an ATX motherboard it is 5V and from what I read in the service manual, it was 5V as well with the original PSU. I replaced the original LED by a new one, with a 200-ohm resistor.

I placed the LED/switch panel back in the case and powered up the system. Voilà! We have a working power LED. This was, for now, the last touch to this system.

As of now, everything works well and the system is very pleasant to use. Loading files/games is super easy through the CompactFlash card and having a sound card and a mouse really makes the system interesting to use.

Final pictures :

(Upcomming)

(Upcomming)