• Please review our updated Terms and Rules here

XTIDE & Compact Flash cards

Mike_Z

Mike_Z

Veteran Member
Joined
Dec 1, 2013
Messages
1,713
Location
Near Milwaukee Wisconsin
I have installed a Glitch Works XTIDE card in my IBM XT. I'm using a Syba IDE to CF adapter. I have attempted to use a dozen of so different CF cards, with various results. My method of setting up the CF is to boot the XT from floppy, use FDISK to set one partition and make it active. Then use FORMAT C: /s. Then I attempt to reboot the XT from the CF card. I have 5 CF cards that no matter what I do will not boot ever. Two must be completely bad because I can not read anything when looking at the CF via a WIN7 machine. The other 3, although will not boot, if I boot from floppy, I can see command.com on the CF card. Then if I transfer programs to the CF, some of the programs will run and some will just crash.
The other 6 or so CF cards work a little better, 4 will boot sometimes and will run some programs, but will crash after a while. The last two cards, both SANDISK, a 256 MB and a 1GB Ultra II will boot about 80% of the time (the Ultra is better) and will run all programs started from floppy or the CF card.
I have used DOS DEBUG to look at the MBR and it looks OK. I can see the boot code and error messages. The first partition 16 bytes
80 01 01 00 06 0E B0 B6 30 00 00 00 80 A2 07 00 (this is for the 256Mb card).
The boot sign is there 55 AA, but yet there are times when the CF just will not boot. I have ordered a different kind of IDE to CF adapter, maybe that is my trouble. Any suggestions? Right now the machine is not reliable, thanks Mike
 
before you format, do a fdisk /mbr
Most CF cards don't have a friendly master boot record.
 
If you boot from floppy, can you see your C: just fine? If so, try "fdisk /mbr", this does require MS-DOS 5 or later.
 
No, I have not used /mbr. I'll give it a try. I was under the impression that, that switch would repair the MBR. I think that mine is OK, but who knows.
Orange, excuse me but what is a DOM? Thanks MIke
 
Last edited:
Based on my pretty miserable luck when it comes to CF compatibility with XT-CF controllers I strongly recommend running several cycles of the "mediatest" function of this DOS disk testing program. I doubt your problem has anything to do with the MBR; that's not a thing that's going to be intermittent, it's either going to boot or not. Intermittent issues with booting, file corruption, and crashes when running programs loaded off the CF are all smoking guns that your cards have compatibility issues.

As I've droned about far too often I personally find these SD to IDE adapters far more reliably compatible than most CF cards, and it's also about the cheapest solution. But of course your mileage may vary.
 
Mike_Z said:
No, I have not used /mbr. I'll give it a try. I was under the impression that, that switch would repair the MBR. I think that mine is OK, but who knows.
Orange, excuse me but what is a DOM? Thanks MIke
Click to expand...
I use a SanDisk Ultra on my 1000SX and it took FDISK /mbr to get it to post.
 
Mike_Z said:
I'm using a Syba IDE to CF adapter.
Click to expand...
As for the 'unreliability' symptoms. A possibility: Looking at a photo of a that adapter online, there is a jumper (JP1 - CF POWER SOURCE) that the user sets according to how the user powers the adapter: either via the power connector on the adapter, or via pin 20 on the adapter's IDE connector. The default setting is for power via the power connector on the adapter. Is that jumper set according to how you are powering the adapter? One might think that if the jumper was set incorrectly, that the CF would not power at all and therefore, would not work, but I'm sure that six or so months ago, someone reported 'strange' symptoms, and it was related to them not powering the adapter properly, or maybe the jumper settings printed on the card were opposite to actual.
 
I read about IDE DOM, looks like another version of IDE-CF. I don't have a DOM, but I do have an old IDE hard drive. I connected a Western Digital WD800 to the XTIDE. Used FDISK to partition and activate 256 MB, then formatted to have the system on it. It would boot sometimes and sometimes not, much like the CF cards. This leads me to believe that the problem is on the XTIDE card. I want to continue testing the WD for a while. I would have thought that the WD should work just fine. I have a couple other IDE drives and may try those. Mike
 
Modem, I have the jumpers set all on 1-2, which according to the data sheet that came with the adapter is the default settings. JP1 is power from external (CNS) or 44 way SFF ICE connector (CN4). JP2 is +5 volts and JP3 is master/single. I had the +5 volt jumper on the XTIDE card closed. Mike
 
Seems that the WD800 would boot better from a warm boot rather than a cold boot. But I did get a error message, Bad or missing command interrupter. When the boot would fail from a warm boot I could use ALT CTRL DEL and it would re boot, yet when the boot would fail on a cold boot the machine would hang. I do not remember exactly if the CF card behavior was the same. Mike
 
I tried 50 boots with the WD800, it failed 6 times. Then I tried the DISKTEST that was suggested. Once on the WD800 and once on the SanDisk 256MB card
...................................WD800.................CF
Write speed...........121 Kb/sec.......109
Read speed............233......................218
8k random...............8.7 IOPS.............6.0
Sector random......14.1......................7.4
Access time.............71 ms................136

I would have thought that the CF card would have done better. Mike
 
Wow. Those are kind of amazingly bad results, performancewise. My Tandy 1000 with an 8-bit XT-CF paired with a PATA->SD adapter, when throttled down to 4.77mhz (instead of its normal 7.16mhz) scores:

WS: 360.25kb/s
RS: 378.56kb/s
8K Rand: 24.9 IOPS
Sector Random: 45.7 IOPS
Access time: 22ms

... Anyway, though, that's not really important. The thing you want to do with disktest is run "disktest mediatest". That will fill the test file with various bit patterns and see if they can be read back correctly. If you get errors here then it's probably pointless to keep trying with that device.
 
The fact that you are seeing issues with a WD800 attached certainly suggests a problem with the XT-IDE card, unless of course, the WD800 (or IDE cable) has issues.

Mike_Z said:
Modem, I have the jumpers set all on 1-2, which according to the data sheet that came with the adapter is the default settings. JP1 is power from external (CNS) or 44 way SFF ICE connector (CN4). JP2 is +5 volts and JP3 is master/single. I had the +5 volt jumper on the XTIDE card closed. Mike
Click to expand...
But when you are back to using the Syba IDE-to CF-adapter:

If the photo below matches your unit, then,
JP1 set to 1-2: Power for adapter/CF is being provided via 'external' (which will be connector CN5) or via the 44-pin connector (connector CN4)
JP1 set to 2-3: Power for adapter/CF is being provided via pin 20 on 40-pin IDE connector (e.g. XT-IDE card supplying power on IDE pin 20)

It sounds like you need to have JP1 set to 2-3.

Untitled.png
 
Ran the media test three times with no errors.
Your picture is cut off and can not see the jumpers. My datasheet says that JP1 2-3 is reserved.
 
I took a step before branching off into experimentation land that may or may not be something I would recommend, but I mention it here for completeness. I purchased my first 'XTIDE - IDE cable - IDE/CF adapter - CF card' setup as an assembled, tested, money back guaranteed unit from a guy on eBay. It worked every time, but it wasn't what I wanted. As I worked towards what I did want, I learned a lot about cards, cables, adapters, CF cards, and the various IDE/ATA standards. When something I tried didn't work, I swapped back to my original "tested working" assembly.

As I understand it, the original CF card standard was supposedly designed as an IDE HDD work-a-like. But like a lot of standards, there is wiggle room in the IDE standard, not to mention that there are at least 3 IDE standards. Some manufacturers interpret the standards differently. That's why you are cautioned to place your newest IDE HDD in the master position. You're also told not to mix and match masters and slaves from different manufacturers. Because they perhaps didn't interpret the unclear parts of the standard in compatible ways. Sometimes even drives from the same manufacturer don't play well together. One site I came across said WD drives are some of the worst in that regard. Try testing with a drive from another brand.

Next, CF cards were not originally meant to be a PC's boot device. Some do it better than others, some don't do it at all. The cards that don't boot aren't necessarily defective. They likely held images from your camera just fine. We're just pushing them into a place they weren't really tested for. DOMs (Disk On Module) are another story. They were intended to boot a device, and they do. But they just don't look like, or swap like, a CF card. And they wouldn't really do what i wanted to do.

Finally, one of the first types of IDE/CF adapters I tried was a fancy expensive ($80 apiece IIRC) part from Addonics. What they had going for them was that they came in a "sled" that mated perfectly with a 3.5" to 5.25" half height drive bay adapter. My plan was to stack two of these in the right hand bay of my 5160. But neither of my XTIDE cards would even identify the CF card when using the Addonics adapter. The next adapter I tried was from Syba. They worked as far as I could tell. I didn't test them extensively because right off I found they identified a CF card as Master regardless of how the Master/Slave jumper was optioned. Since part of my plan was to have two "drives" exposed and swapable in the front of my 5160, that wasn't good enough for my purpose. So I dropped back to the bare bones IDE/CF unit that came with my eBay rig, ordered two more online, sorted out some funny cable issues (a cable select 40 conductor IDE cable that wasn't marked as such), and it all "just worked."
 
Last edited:
Could use a design review - I came up with a slightly different way of doing the 16-bit to sequential double-word cycle. The odd/even byte cycle is handled in hardware via manipulating the bus clock signals. This avoids the need for the 'delay-line' write strobe. The way I have in mind is a rock-solid approach that in the read mode, strobes the IDE port and latches the whole word, then the clocking enables the byte outputs via a registered clock delay (flip-flop chain.) That way, each byte is held solidly without having to rely on any data-hold timing. The write cycle works in reverse, storing the bytes, but NOT strobing the IDE write until after the last byte written, allowing the data to be solidly held in the latch, again not relying on any delay-line effects. The byte address increment issue is solved via latching the A0-A2 lines upon the FIRST byte written, 'freezing' them for use on the 2nd byte store and subsequent IDE write. This eliminates the hassle of doing a 'bit-shift', as the IDE address lines have been held in the first byte write cycle. My idea is to have it all done based on the bus clocking only, thus not even having to worry about hold times or transition delays through logic. I'd love feedback, and anyone is free to use this. (If you wire it up before I do, please let me know how well it works!) From what I researched, this is fully transparent to the XT-IDE bios out there.
floppy-IDE.jpgXT_IDE_clocking_read.jpgXT_IDE_clocking_write.jpg
 
You must log in or register to reply here.
Back
Top