• Please review our updated Terms and Rules here

XTIDE Device Compatibility List

blackepyon

Veteran Member
Joined
Feb 20, 2017
Messages
569
Location
Chilliwack, BC, Canada
After noting that there's no thread for listing devices used with the various XT-IDE Universal BIOS (XUB) based cards, I decided to make one.
This thread is dedicated to listing the devices used with the 8-bit cards made to utilize XUB, such as Lo-Tech's ISA-CF rev.2, XT-CF-lite, and Glitch's XT-IDE board series, derivative cards based on them, and so on.

These devices can include Compact Flash cards, DOMs, hard drives, etc, as well as Compact Flash to IDE or SD-card to IDE adapters, and the like.
  • Post pictures of these devices if you can, showing which brands/models you've confirmed work, list benchmarks if you have them, and system make/model/specs if appropriate.
  • Please list which XUB card(s) you are using them with, and if the card you are using is an "off brand" derivative of another popular card (there are a number of cards based on the Lo-Tech ISA-CF rev.2b, for instance), which version or revision of the XT-IDE Universal BIOS you are using, how the XUB recognizes the device(s) in POST.
  • List other relevant information if you can think of it, such as whether or not the devices are used in a master/slave combination, etc, and if they take issue if there's another device on the controller. For instance, some CompactFlash cards don't like to be the master device in a master/slave relationship, unless they are of the same brand name, etc, but may work perfectly well as either master or slave if they're on their own, while CF cards don't care one way or the other.
  • Which DOS version you're running, as well as partition information or custom drive geometry (if you've had to do that for whatever reason), may be useful too.
  • Please place emphasis on which devices you've tested that do NOT work, along with any of the above things you may feel appropriate, so other people know not to buy them for that hardware combination.
  • If you've tested a device and noted it to work on one card, BIOS revision, etc, but not another, please give specific information on your findings, if you can.
Please keep all discussion directly relevant to the topic. This is meant to be sort of a "buyer's guide," and hopefully to collate some data to see if certain brands are more problematic than others, etc.

There's no need to list complete system specs, such as game/sound cards, monitor, other peripherals, etc, if it's not relevant. This is not the place to brag about your vintage computer collection, or how "maxed out" your retro-gaming rig is. Discussion related to device combinations, such as, if a particular CompactFlash-IDE adapter is finicky with certain CompactFlash cards, or if you know of a good place to buy particular card, etc, are more than welcome here, but if you need help troubleshooting issues related to the XUB-based cards themselves (hardware conflicts, etc), please direct those to the appropriate thread: XTIDE tech support thread, or just use a PM.
 
These are the CompactFlash cards and adapters I have. Each has been tested to work on my custom built variants of the Lo-Tech ISA-CF rev.2, and the Glitchworks XT-IDE rev.4, both with XUB 2.0.0β3 (rev. 601) ("custom built" to the "PLUS" form factor for use in my Tandy 1000HX, but otherwise functionally identical).
System: Tandy 1000HX, 8088-2 processor at 7.44MHz, 640KB RAM. DOS 5.0, all are single partitions filling the entire drive, and show up fine.
Benchmarks: Check-It 3, with the XT-IDE rev.4 in hi-speed "Chuck mode."

Canon FC-16M
Canon FC-16M.jpg
Shows up as "Hitachi XXM 2.3.0." 16MB capacity, isn't picky about master/slave configuration.
272.0 KB/s transfer speed (3.20x IBM PC-XT), 1.4ms/1.4ms average seek and track seek times.

SanDisk 128MB
SanDisk128.jpg
Shows up as "FLASHCARD." 128MB capacity, does not like to be master to a slave unless the slave device is of like manufacture.
272.0 KB/s transfer speed (3.20x IBM PC-XT), 1.6ms/1.6ms average seek and track seek times.

SanDisk 512MB
SanDisk512.jpg
Shows up as "STI FLASH 7.4.0." 512MB capacity, does not like to be master to a slave unless the slave device is of like manufacture.
272.0 KB/s transfer speed (3.20x IBM PC-XT), 1.6ms/1.6ms average seek and track seek times.

Memory Technology Company 128MB
MTC 128.jpg
Shows up as "CF CARD." 128MB capacity, does not like to be master to a slave unless the slave device is of like manufacture.
268.0 KB/s transfer speed (3.15x IBM PC-XT), 1.7ms/1.7ms average seek and track seek times.

Memory Technology Company 512MB
MTC 512.jpg
Shows up as "CF CARD." 512MB capacity, does not like to be master to a slave unless the slave device is of like manufacture.
272.0 KB/s transfer speed (3.20x IBM PC-XT), 1.4ms/1.4ms average seek and track seek times.

Note: my comments on the master/slave configurations might be due to the dissimilar CF-IDE adapters I'm using, further testing may reveal differently. See pictures below.

Each of these work, and have master/slave jumpers:

CF-IDE 40pin adapter V.D2
PC bracket CF-IDE.jpg
These guys are common enough. No +5v pin-20 is present, but the CF seems to be able to pull power from DMACK on IDE pin-29, I think? Which is convenient, because that pin is pulled to +5v on the XT-IDE rev.4.

HX-218
HX-218 CF-IDE40.jpg
These guys are a few bucks a piece from China, and they're all over the place. I have no idea why they have a 3.3v power jumper, but these pull power just fine off of pin-20. I've got another one of these on order, and I'm hoping to see if the master/slave issues MIGHT be resolved by having identical adapters (my custom built XT-IDE card has internal and external IDE headers), though I wouldn't bet money on it.

CF-IDE44/2.0mm adapter V.H2
44-pin CF adapter.jpg
Generic 44-pin laptop-sized CF-IDE adapter (with a straight-through 44-pin to 40+power adapter). Passive device. Works just fine, if that's what you got.
 
Last edited:
Based on the results from my first round of extensive QA I'm going to create a bootdisk with James Pearce's DISKTEST.EXE on it and go through my pile of cards again, running a cycle or two of the "mediatest" function before I declare whether the card is actually good or not.
 
Watch out on those little "HX-218" adapters (they show up with other numbers too). I usually buy 5 or 10 at a time to make sure I get enough good ones for whatever I'm doing. They're universally poorly built. I've had several XT-IDEs come back to the shop when the problem turned out to actually be one of those cheap-o CF adapters.
 
Watch out on those little "HX-218" adapters (they show up with other numbers too). I usually buy 5 or 10 at a time to make sure I get enough good ones for whatever I'm doing. They're universally poorly built. I've had several XT-IDEs come back to the shop when the problem turned out to actually be one of those cheap-o CF adapters.

I imagine they got sloppy soldering the CF connector, but that's generally the thing you have to check when getting stuff from China. When they work, they work. When they don't, it tends to be sloppy assembly. Fortunately, all of mine have worked.

Well, they're certainly cheap enough that buying 5 or 10 at a time isn't a significant expense.
 
That was my attitude -- I wouldn't use them in something for an industrial control customer, but they're fine for hobbyist stuff, and they're so cheap you can get a few every time to make sure you get enough working adapters. I've never bothered diagnosing one, but it seems to usually be stuck or open bits, often in the upper byte (some will work in 8-bit mode). They are of course "cheese grade" throughout :)

Is there interest in setting up a compatibility list page on glitchwrks.com? Something like with the TestFDC registry, where people can submit stuff and it gets hand-curated? I could set up a text export to the XT-IDE repository on Github so it's available there, too.
 
Is there interest in setting up a compatibility list page on glitchwrks.com? Something like with the TestFDC registry, where people can submit stuff and it gets hand-curated? I could set up a text export to the XT-IDE repository on Github so it's available there, too.
That's a good idea. Soon as we can get a dataset together. The problem, as has been pointed out, is that lots of people have reported on what works for THEM, but it's just all over the place.
 
That was my attitude -- I wouldn't use them in something for an industrial control customer, but they're fine for hobbyist stuff, and they're so cheap you can get a few every time to make sure you get enough working adapters. I've never bothered diagnosing one, but it seems to usually be stuck or open bits, often in the upper byte (some will work in 8-bit mode). They are of course "cheese grade" throughout :)

Visually at least the soldering quality of all of my CF adapters is pretty sketchy, but I'd have to say the 44 pin ones (same PCB as pictured earlier) look even worse than the 40 I have (Also that same PCB). But they all seem to work, at least with the cards that appear to be most compatible. Again, I've decided I'm not going to call a device compatible until it's passed a few rounds of media tests because I definitely am seeing on some cards the initial *appearance* of functionality, IE, they'll take a format and boot an installed DOS, but when you use it a while flaky behavior starts becoming evident.

So far I still have to say that my favorite device is this 44 pin IDE to SD adapter.

SD_Converter.jpg

The Amazon listing calls it "GODSHARK" brand, but it looks like this exact same unit is sold under multiple names. I've seen it on eBay for under $7. This just happens to be the listing I bought it from. Identifies as "SINTECHI HighSpeed SD to CF Ad". I've run multiple media tests ("disktest.exe mediatest", default 4MB test file size) and it has been rock solid. All tests are with an old 2GB "Microcenter"-branded SD card installed. (I will try swapping in a different card and see if that makes a difference.)

Checkit3 disk benchmarks:

Transfer Speed: 548.0Kps
Rating: 6.45 times IBM PC-XT
Seek: 1.4ms, 1.4ms

DiskTest 2.3:

Write Speed: 457.65 KB/s
Read Speed: 503.81 KB/s
8k random, 70% read: 14.9 IOPS
Sector random read: 58.3 IOPS

Test platform is a Tandy 1000HX motherboard with a 7.16mhz V20 CPU, 624k+96k RAM, MS-DOS 6.22. XTIDE BIOS ID's as v2.0.0b3, 2013-10-22 (it's the newest version I've actually found in compiled form), it's the "IDE_XTP.BIN" variant with 80186 instructions. Host adapter is a homespun XT-CF-Lite v2b mostly-clone. (I'm using the "pickier" of my two prototypes, which may have a noisier bus or tighter timing requirements than the original. I will also test any cards that fail on it on the seemingly slightly more forgiving prototype that is a closer copy of the reference design.)

I'm thinking of buying a couple of the cheapest listings for this board off eBay and trying them. If they work as well as this example seems to I wouldn't hesitate to recommend this over a CF card. The quality of construction of my unit at least looks superior to the CF adapters as well.
 
XTIDE BIOS ID's as v2.0.0b3, 2013-10-22 (it's the newest version I've actually found in compiled form)

The team abandoned the Google Code site quite a while ago and moved to the new xtideuniversalbios.org domain. For some reason, they do not directly link to their SVN builds on the main page, but the newer builds are here: http://www.xtideuniversalbios.org/binaries/
The latest is R602 with a build date of March or April 2019. (I am not at a machine to check which month..but it is far more recent then 2013 either way)
 
The team abandoned the Google Code site quite a while ago and moved to the new xtideuniversalbios.org domain. For some reason, they do not directly link to their SVN builds on the main page, but the newer builds are here: http://www.xtideuniversalbios.org/binaries/
The latest is R602 with a build date of March or April 2019. (I am not at a machine to check which month..but it is far more recent then 2013 either way)

Ugh. Yes, thank you for throwing out that secret link again; I know at one point I'd found it and I believe I'd actually been running that version for a while, but I inadvertently downgraded to that older version (which is apparently "R566") when I switched to IDE_XTP. It's mildly confusing to tell them apart because both ID as v2.0.0b3 when booting. I went ahead and upgraded to this version and ran through a sample of the disks I'd already tested again; I don't think it made any difference in compatibility but might as well be running the latest.

As an addition to the "ground rules" of this thread I would probably recommend running at least a couple cycles of DISKTEST.EXE's "mediatest" function before declaring a disk "compatible". (The default 4MB file size is probably adequate, using "maxsize" on all but the smallest cards would take forever in an XT.) In testing my heap of cards I found a depressing number that will let you install DOS and boot your system (at least some of the time), but in extended use will introduce read/write errors at random that will express themselves as random system flakiness. Outside of an asterisk or two (which could be CF adapter related) the cards that are "bad" failed this test consistently across the two computers/host adapters I tested, and in no case out of a batch of identical cards (same size/manufacturer/ID string) were there inconsistent results. (IE, all the cards repeatedly tested either good *or* bad.)
 
Here's some cards I'm pretty confident putting on Santa's naughty list.

Cisco OEM-labled Unigen Flash, 512M and 2GB

UNIGEN_FLASH_cards.jpg

I'm lumping these two sizes together because they behave identically. Show up as "UNIGEN FLASH". Only tested as master with no slave.

At first glance these cards seem to perform well; they will partition with FDISK, take a format, and boot at least most of the time. Testing them with DISKTEST MEDIATEST, however, reveals a massive error rate:

unigen_mediatest.jpg

It's a shame these cards don't work because on benchmarks they're the best performing thing I tested, just edging out the CF-SD adapter on DISKTEST IOPs.

Tested on both the previously mentioned Tandy 1000 HX motherboard + homebrew combo serial/XT-CF-lite card and a Tandy 1000 EX with a simpler XT-CF-lite clone. Same results with firmware versions R566 and R602, both XT and XTP variants. Combo card used "CF-IDE44/2.0mm" adapter from previous page (tried two different units), 40-pin card tested with HX-218 (previous page) and a nameless 40-44pin adapter on short 40 pin cable with CF-IDE44 fitted.
 
Tested cards:
1GB SanDisk CF card
2GB SanDisk Ultra CF Card
4GB SanDisk Ultra CF Card

Adapter used:
Syba CF2IDE adapter (master/slave jumper, power from IDE or external, and voltage)

Network card:
3Com EtherLink III 3C509-TP

System:
Packard Bell Pack-Mate 28 Plus (prior to the Micro Firmware BIOS upgrade)

The system worked with the CF cards, but, after a while, it started to slow down every 2 months.
 
Tested GOOD on the same test platforms described in the last post.

Cisco OEM-label Simpletech Flash 8.0.0 128MB (3x)

128msti_flash.jpg

Tested 3 identical units, master only, no slave. Identifies as "STI FLASH 8.0.0". Performance was consistent between the three cards and all tested host adapter/carrier combinations.

FDISKs and formats with no issues, reliably passes MEDIATEST. IOPS scores seem oddly lousy in DISKTEST; with XTP firmware typical scores:

Write Speed: 374.75 KB/s
Read Speed: 475.17 KB/s
8K random, 70% read: 5.1 IOPS
Sector random read: 5.6 IOPS

These cards were also quite slow when connected to the USB-PATA adapter I used to zero them before starting. In actual use they don't feel much if any slower than the faster benchmarking disks, however. Might make a difference on a machine faster than an XT.

Checkit 3.0 benchmarks (XTP):

Transfer speed: 543.9K/second, 6.40 times IBM PC-XT
Average/Track seek: 1.3ms, 1.3ms
 
Im using r620 XTPL edition on my v20 equipped EX, running DOS 5.0 and my CF-IDE is running in 8-Bit BIU mode:

DiskTest, gives me the following results on my:

4GB Seagate ST1 MicroDrive:

XT-IDE ID: ST640211CF
Write: 334.37 KB/s
Read: 525.14 KB/s
8k Random: 13.5 IOPS
Sector Random: 18.5 IOPS


1GB no-name Flash Drive:

XT-IDE ID: CF Card
Write: 266.32 KB/s
Read: 441.38 KB/s
8k Random: 12.7 IOPS
Sector Random: 18.8 IOPS


IMG_9198 copy.JPG
 
No, however both CF's have been operating without issue for months, I tend to swap between them when testing stuff.

That's probably good enough, then.

Those Unigen cards really screwed me. I probably would have noticed how flaky they were sooner and made the connection if I'd set them up from scratch in the machines and *hadn't* also been in the middle of general hardware QA on the new boards. But instead I got the "brilliant" idea, since they were just exactly the right size to do it and the cards seemed nifty and fast, of using DD on the unix box to clone the working DOS install I had on the CF-SD adapter over to the pair of 2gig cards. Despite how obviously flaky they look under the microscope the fact that they flip bits on the regular can *mostly* go unnoticed, or at least not be obvious to pin on the CF card. Stuff will just crash at random well after it was loaded into memory and you'll pull your hair out and start howling about what *must* be wrong with your RAM board. Sigh.
 
Back
Top