I have the strangest problem on my PC using the XUB in a NIC with a 16-bit multi I/O card (in 8-bit mode) IDE controller and a CF adapter. On boot it never detects the CF card even though the LED on the CF adapter comes on and the XT-IDE BIOS comes up. I’ve checked for exploded tantalums, ran memory and board tests with CheckIt, moved and removed cards, swapped power supplies and multi I/O cards but there’s no obvious fault. Thinking that it could be other hardware I took out all the ISA cards and put them into an XT board and it boots up from the NIC and sees the CF card just fine. All I could think is maybe bad caps on the 5150 motherboard but I’m all out of ideas.
| VCF West | Aug 01 - 02 2025, | CHM, Mountain View, CA |
| VCF Midwest | Sep 13 - 14 2025, | Schaumburg, IL |
| VCF Montreal | Jan 24 - 25, 2026, | RMC Saint Jean, Montreal, Canada |
| VCF SoCal | Feb 14 - 15, 2026, | Hotel Fera, Orange CA |
| VCF Southwest | June, 2025 | University of Texas at Dallas |
| VCF Southeast | June, 2025 | Atlanta, GA |
-
Please review our updated Terms and Rules here
IBM 5150 Can’t Find Drives With XT-IDE
- Thread starter spearce
- Start date
Nope, Using a 16-bit multi I/O card (in 8-bit mode) IDE controller and a CF adapter In an IBM 5150 does not work, The reason being is due to the 'Reserved I/O range in the 5150', See Modem7's website https://minuszerodegrees.net/5160/diff/5160_to_5150_motherboard_differences.htm
That explains a lot since all the standard IDE I/O addresses fall in that range for the 5150. I think I was tripped up on this because I was experiencing memory errors and lockups with the WD MFM controller that was there previously and thought this was all related to bad caps or a power supply problem. Mysteriously, that problem seems to have ceased for now at least. Presumably now I’ll need to look into 8-bit IDE controllers or actual XT-IDE cards that support non-standard addresses and I won’t be able to use the rear mounted CF adapter like in my XT (as I’m out of slots).
Sounds like the same problem I had with my 5150 and XT-IDE card. It never worked no matter which CF card I'd try. But it did work fairly reliably in my '486 motherboard.
Still lacking in a solution for my 5150, I recently acquired a PicoMEM card from TexElec. That card works rock solid in the 5150. And as a bonus, it also provides additional memory and Wifi networking among many other features. If you're willing to look for a different solution, I suggest going that route. And here's a thread started by Freddy (creator of PicoMEM) https://forum.vcfed.org/index.php?threads/picomem-project-pi-pico-on-an-isa-board.1241199/
Still lacking in a solution for my 5150, I recently acquired a PicoMEM card from TexElec. That card works rock solid in the 5150. And as a bonus, it also provides additional memory and Wifi networking among many other features. If you're willing to look for a different solution, I suggest going that route. And here's a thread started by Freddy (creator of PicoMEM) https://forum.vcfed.org/index.php?threads/picomem-project-pi-pico-on-an-isa-board.1241199/
That’s interesting because an XT-IDE card with a configurable I/O address should work as most start at 300h, so maybe it’s a ROM problem? I’ve been looking at the PicoMEM and it’s interesting but definitely overkill for what I’m trying to solve. I may get one for my XT later even though I have a working set up with a multi I/O + NIC xub boot ROM, it depends on where development goes.
I’ve been looking at what’s available these days as so many XT-IDE cards are somewhat pricey when converted to CAD. Thankfully I do have that BIOS version so I can boot cards with ROMs without issue, such as a boot ROM in my NIC. The other option I’d forgotten about is Blue SCSI v2 as I have a Seagate ST-02 card that’s picky with drives no matter the BIOS version and a Future Domain TMC-850 that I’ve had some issues with in my XT. It seems like that’s at best only a slightly cheaper option but both cards are confirmed to work with it.
I'm not sure what caused my problem on the 5150. Whether it was a conflict with the I/O address or an IRQ. I doubt that since the XT-IDE card would post just fine, but it would almost always garble the CF card name and wouldn't be able to read/write to it. I actually thought I might have had a defective XT-IDE until I tried it in my '486. There it works nearly flawlessly. Though it sometimes garbles the CF card name when posting and then the drive is unreadable (this is a rare occurrence, but reseating the card fixes it).
I almost think I may have either a slight voltage issue in the 5150 and the CF cards just so happen to be the most voltage sensitive to device in the computer, or I'm having bad connections with the ISA sockets. I need to get the meter or the scope out to check it. But I haven't since the XT-IDE has been so useful in the '486. Maybe I should just "Deoxit that Socket!"
I almost think I may have either a slight voltage issue in the 5150 and the CF cards just so happen to be the most voltage sensitive to device in the computer, or I'm having bad connections with the ISA sockets. I need to get the meter or the scope out to check it. But I haven't since the XT-IDE has been so useful in the '486. Maybe I should just "Deoxit that Socket!"
That's pretty interesting, I could reliably reproduce the garbled device name in my XT if I connected a regular IDE drive that needs 16-bit transfers but I'd be surprised if you had multiple CF cards that didn't support 8-bit mode as that's unusual. How are you using it in 486 and which XT-IDE card is it? From what I've read it's usually not recommended to use the 8-bit XT-IDE cards in machines with 16-bit ISA slots as there's a performance penalty but using the ROM with either the onboard IDE interface (drive disabled in BIOS) or a 16-bit ISA or 32-bit VLB card can work.I'm not sure what caused my problem on the 5150. Whether it was a conflict with the I/O address or an IRQ. I doubt that since the XT-IDE card would post just fine, but it would almost always garble the CF card name and wouldn't be able to read/write to it. I actually thought I might have had a defective XT-IDE until I tried it in my '486. There it works nearly flawlessly. Though it sometimes garbles the CF card name when posting and then the drive is unreadable (this is a rare occurrence, but reseating the card fixes it).
I almost think I may have either a slight voltage issue in the 5150 and the CF cards just so happen to be the most voltage sensitive to device in the computer, or I'm having bad connections with the ISA sockets. I need to get the meter or the scope out to check it. But I haven't since the XT-IDE has been so useful in the '486. Maybe I should just "Deoxit that Socket!"
After a lot of searching and pricing things out I've decided to go with a Blue SCSI v2 Desktop card since it's somehow cheaper than any XT-IDE card, even older ones. I found a video on YouTube where a guy uses it in his 5150 with a Seagate ST-02 card (BIOS updated from v3.0 to v3.3) and I have the same hardware and a ROM programmer. From there I'll swap some hardware with my AT and convert it from its current mechanical SCSI drive set up to XT-IDE using the multi I/O card currently in the 5150 and an XT-IDE ROM in the NIC but it'll run fully 16-bit.
The card I have is an Lo-tech XT-CF-lite rev.2 that I picked up from Tex-elec ( https://texelec.com/product/lo-tech-xt-cf-lite-rev-2/ ). Seeing as it is fairly reliable in the 486 I don't think it's the card's fault. I've tried at least 6 different CF cards (64MB, 256MB and multiple 1GB cards from different makes, and an 8GB card).
When I pulled my 486 out of 2 years ago, I was saddened to find the drive was toast. And after having the trouble with XT-CF-lite card in the 5150 I tried it in the 486 with success. I've disabled the onboard IDE controller and plugged the card into an 16bit slot. Yeah may not be the fastest, but I really haven't noticed it much. I'm just glad to get the computer working again and able to load some of the programs I used to have on it at the time.
I've been looking at those Blue SCSI for another project I have. I've acquired a Macintosh SE with a failing hard drive. Adrian (Adrian's Digital Basement) had an episode where he used a Blue SCSI plugged directly into the back of the Mac. I figure this would be a good solution if I'd like to put that computer on display too.
When I pulled my 486 out of 2 years ago, I was saddened to find the drive was toast. And after having the trouble with XT-CF-lite card in the 5150 I tried it in the 486 with success. I've disabled the onboard IDE controller and plugged the card into an 16bit slot. Yeah may not be the fastest, but I really haven't noticed it much. I'm just glad to get the computer working again and able to load some of the programs I used to have on it at the time.
I've been looking at those Blue SCSI for another project I have. I've acquired a Macintosh SE with a failing hard drive. Adrian (Adrian's Digital Basement) had an episode where he used a Blue SCSI plugged directly into the back of the Mac. I figure this would be a good solution if I'd like to put that computer on display too.
Throughput is limited by the 8-bit interface, but you still get the advantage of not having seek times compared to a spinning disk (and compatibility). It mostly depends on how much data you need to transfer, and especially in DOS, that's often not that much. These are slow machines anyway, a second hitting the disk won't kill us.
I have used both the internal (50-pin) and external (DB-25) variants of the BlueSCSI with a Mac Classic. Both work great. The latter is nicer on a compact Mac since you don't need to open the case.
Ensure that the BIOS settings on the motherboard are correctly configured for your hardware. Specifically, check settings related to the ISA bus, IDE controllers, and any specific settings for boot devices. Even though you've tested different cards, conflicts can still occur. Make sure that the IRQ and I/O addresses for the multi I/O card and the IDE controller don't overlap with other devices or system resources.
If your XT-IDE BIOS is outdated, it might not be fully compatible with your setup. Check if there's a newer version of the XT-IDE BIOS available and consider updating it.
If your XT-IDE BIOS is outdated, it might not be fully compatible with your setup. Check if there's a newer version of the XT-IDE BIOS available and consider updating it.
Ah, I had strongly considered getting this card because it more or less met all my requirements but as with all the others it was more expensive than BlueSCSI since I already have SCSI controllers on hand. As others have mentioned in this thread the performance bottlenecks using an 8-bit card may not be as big of an issue depending on what you're running and would rival a mechanical drive anyway. For a 32-bit OS I could see there being some penalty but maybe it's more comparing using a CF card on a wider bus as opposed to an old drive where CF should win regardless. I ended up going through with my plan and am using a 16-bit multi I/O IDE interface in my AT with the drive disabled in the BIOS and using the XT-IDE BIOS in the NIC and it works really well -- but I don't own any XT-IDE cards.
For the PC I went with BlueSCSI + Seagate ST02 and with a lot of mucking around got it going properly. I do find that the main target for BlueSCSI is old Macs so it could be a good choice for your setup and should just work out of the box. For my set up there were problems because the ST02 could access the virtual hard drive but wouldn't boot from it because it defaults to 255 heads per cylinder and the onboard BIOS hated that.
I think you're probably correct on the point about the 8-bit bus as long as its with a single process OS like DOS. I've read that CF and SD aren't necessarily great choices with multiprocess OSes anyway where there's a lot of writes but I can't speak to that because I haven't used that kind of setup.
I ended up going with the BlueSCSI v2 Desktop (internal) and it's pretty solid but took a bit of trial and error because of compatibility issues with the controller I'm using. After trying some settings from other legacy machines from the documentation I figured out how to get it going and now everything just works and the performance seems comparable to XT-IDE in a PC/XT class machine.
This ended up being a known compatibility problem with a standard 16-bit IDE controller in an 8-bit slot with the 5150 PC as the IDE I/O addresses fall into reserved space and so it doesn't work. If I'd had a regular XT-IDE card with configurable I/O addresses then it would've been fine but I was just using the XT-IDE BIOS. Moving the same multi I/O card to an AT and using an XT-IDE ROM works great, just as it does in an XT.
That depends on how many writes the OS actually does. A memory-starved system constantly swapping (or a task switcher doing the same) is definitely not good for flash storage, but such setups are unlikely to be fun to use anyway.
On the other hand, people are using regular SD cards in Raspberry Pi and similar systems. The amount of writes caused by a single Linux system upgrade dwarfs anything a lowly PC/XT does in a week of regular use. Most cards should hold up fairly well; especially when only a small part of the card is used (e.g. 500 MB on a 4 GB card), any halfway decent wear-levelling algorithms has all the space it needs.
Hmm, I wonder if there is a decent TRIM utility for DOS... to mark areas not covered by partitions as unused.
I can speak to this one after using Raspberry Pis for over a decade and unfortunately SD card corruption is a very real problem with any meaningful writing frequency. It’s definitely improved as they’ve put better SD adapters on them but even with a model 4B I boot and run it off a regular drive with a UAS compatible USB adapter. I’ve never heard of anything supporting TRIM support in DOS but that would be interesting.