• Please review our updated Terms and Rules here

XTIDE tech support thread

I have a 286 with 3 multi I/o cards and they all behave the same way with the bios doing this. not being able to detect ANYTHING but IDE HDDs the other computers work with other devices except this one. i have no idea whats causing it to not function properly
20220716_221756.jpg
 

Attachments

  • 20220714_173226.jpg
    20220714_173226.jpg
    1.3 MB · Views: 28
  • 20220717_190510.jpg
    20220717_190510.jpg
    1.1 MB · Views: 29
Hi all, theres a lot to read through here, so Im just going to start over and ask my question straight up:

Im trying to get an XT-IDE rev 4 board working in an XT, but not having much luck.

It seems to be detecting the CF card, although it is showing some gibberish text. According to the problems page this can be caused by the wrong type of device being selected when the BIOS was configured. Ive tried a couple of different device types, but that doesnt seem to be helping. I either get "not found" or random characters. See the attached screenshot for my current settings.

When I am seeing random characters, I can fire up fdisk and try to create a partition. I can see the activity light on for about a second while this happens, but afterwards fdisk doesnt recognise any partitions, and I can go around in circles like this for as long as I want. So it seems like something is being written, but its not "sticking"?

Ive also tried a couple of different CF cards, one is a 32MB Cisco branded card, another is a 128MB somewhat generic "industrial" card. With the 128MB card, fdisk detected this as some 970 odd cylinders.

I tried creating a partition on the 32MB card on my Mac, and during boot XT-IDE will complain about the disk being not bootable, so I guess it sees the partition to some degree? fdisk does not, however.

Ive also been over the board with a magnifying glass to make sure all of my solder joints are looking good, none appear dry or particularly odd.

Edit: Ive also now just tried the R622 version of the BIOS (was previously using R624) along with the "XTCF PIO" mode, and in this case it seems that it was able to read back the model of CF card, although maybe only just. Sometimes It will just read back "UNIG" while other times it might read "UNIGEN FLASH".

Two things which I have done which are perhaps "non-standard", if they matter:

1. Im using my adapter at address 0x380/388 because I want to use the prototype board range (0x300-31F) for some upcoming projects (0x38X is for some SDLC/bisync adapter that I dont think I'll ever have). At this address things seem to be functional, but I have also tried it down at the default 0x30X just incase, but there was no change.

2. I have used mostly 74HCT parts because I just bought a bare board and dont hold any LS parts in stock. Some of these parts are AHCT which are a bit faster than HCT. Most specifically the AHCT parts are the two 04's, the 138, and the two 573's. Is anyone familiar enough with the circuitry to know whether these AHCT parts, maybe specifically the 04's since they are used to generate some delays, are just "too fast"? I havent tried it yet but I could solder in a small cap in one or both of the delay chains to try and slow them down a tad.

Any pointers appreciated.

Thanks!
 

Attachments

  • Screenshot 2023-01-02 at 13.01.06.png
    Screenshot 2023-01-02 at 13.01.06.png
    22.2 KB · Views: 5
Last edited:
Hi all, theres a lot to read through here, so Im just going to start over and ask my question straight up:

Im trying to get an XT-IDE rev 4 board working in an XT, but not having much luck.

It seems to be detecting the CF card, although it is showing some gibberish text. According to the problems page this can be caused by the wrong type of device being selected when the BIOS was configured. Ive tried a couple of different device types, but that doesnt seem to be helping. I either get "not found" or random characters. See the attached screenshot for my current settings.

When I am seeing random characters, I can fire up fdisk and try to create a partition. I can see the activity light on for about a second while this happens, but afterwards fdisk doesnt recognise any partitions, and I can go around in circles like this for as long as I want. So it seems like something is being written, but its not "sticking"?

Ive also tried a couple of different CF cards, one is a 32MB Cisco branded card, another is a 128MB somewhat generic "industrial" card. With the 128MB card, fdisk detected this as some 970 odd cylinders.

I tried creating a partition on the 32MB card on my Mac, and during boot XT-IDE will complain about the disk being not bootable, so I guess it sees the partition to some degree? fdisk does not, however.

Ive also been over the board with a magnifying glass to make sure all of my solder joints are looking good, none appear dry or particularly odd.

Edit: Ive also now just tried the R622 version of the BIOS (was previously using R624) along with the "XTCF PIO" mode, and in this case it seems that it was able to read back the model of CF card, although maybe only just. Sometimes It will just read back "UNIG" while other times it might read "UNIGEN FLASH".

Two things which I have done which are perhaps "non-standard", if they matter:

1. Im using my adapter at address 0x380/388 because I want to use the prototype board range (0x300-31F) for some upcoming projects (0x38X is for some SDLC/bisync adapter that I dont think I'll ever have). At this address things seem to be functional, but I have also tried it down at the default 0x30X just incase, but there was no change.

2. I have used mostly 74HCT parts because I just bought a bare board and dont hold any LS parts in stock. Some of these parts are AHCT which are a bit faster than HCT. Most specifically the AHCT parts are the two 04's, the 138, and the two 573's. Is anyone familiar enough with the circuitry to know whether these AHCT parts, maybe specifically the 04's since they are used to generate some delays, are just "too fast"? I havent tried it yet but I could solder in a small cap in one or both of the delay chains to try and slow them down a tad.

Any pointers appreciated.

Thanks!
its the cf card. Half the ones i have dont work. here this one definitely works: https://www.amazon.com/dp/B07HM9849Z?psc=1&ref=ppx_pop_dt_b_product_details


i have gotten plenty frustrated with some good cf cards in my drawer which will formst and sort of work. but they end up giving me tons of issues. I cN all but gurantee its the card.

Edit. I was having similar problems with my XTIDE and another board member @Mochatea396 told me this card works fine in his XTIDE (and it turns out it works grea) So I am just passing the info on.
 
Last edited:
Edit: Ive also now just tried the R622 version of the BIOS (was previously using R624) along with the "XTCF PIO" mode, and in this case it seems that it was able to read back the model of CF card, although maybe only just. Sometimes It will just read back "UNIG" while other times it might read "UNIGEN FLASH".
See [here]. That displaying of the model information is not connected in any way with the contents of the CF card. You need to get the model information displaying consistently, then address possible issues with CF content.
 
Im trying to get an XT-IDE rev 4 board working in an XT, but not having much luck.
I assume you have the Glitch Works R4 XT-IDE card https://www.minuszerodegrees.net/xtide/rev_4/XT-IDE Rev 4 - general.htm
It seems to be detecting the CF card, although it is showing some gibberish text
Bad solder joints, Bad sockets, Bad IC/'s, Bad CF adapter, Bad CF card, Bad connection's, Take your pick.
When I am seeing random characters, I can fire up fdisk and try to create a partition. I can see the activity light on for about a second while this happens, but afterwards fdisk doesnt recognise any partitions, and I can go around in circles like this for as long as I want. So it seems like something is being written, but its not "sticking"?
So you can seemingly use FDISK to create a partition but upon reboot there is no partition created, That's not unusual if any of the above faults exist.
Ive also tried a couple of different CF cards, one is a 32MB Cisco branded card, another is a 128MB somewhat generic "industrial" card. With the 128MB card, fdisk detected this as some 970 odd cylinders.
What version of DOS are you using ?
Edit: Ive also now just tried the R622 version of the BIOS (was previously using R624) along with the "XTCF PIO" mode, and in this case it seems that it was able to read back the model of CF card, although maybe only just. Sometimes It will just read back "UNIG" while other times it might read "UNIGEN FLASH".
XTCF PIO mode is used with the Lo-Tech XT-CF adapters not the GW R4
I have used mostly 74HCT parts because I just bought a bare board and dont hold any LS parts in stock
Where did you get the bare board from ?, I have used HCT / HC / LS series in my GW R4 boards with no problems though HCT are preferable.
 
Yes, its a glitch works rev 4 (not 4a). It was bought from a seller on ebay. https://www.ebay.co.uk/itm/394357845492. The only mods made to this board, as I am led to believe, is to include solder bridges alongside many of the jumper/switch locations so that you can cross those parts off the BOM if you so desire.

IBM PC DOS 3.30 if I recall exactly. Something close to those lines. I can get an exact version if it matters.

Ive got a different CF adapter on order because after a bit more research it seems this can play a big part in whether it works or not. Ive also got some HCT parts on order to swap out some of the AHCT parts, just in case. HC would be out of the question because their Vih spec is incompatible with the rest of the machine which is TTL. They may well work, but it could end up being "one more thing" to troubleshoot to ensure wasnt part of the problem.
 
The original GW R4 boards are easy enough to build though obviously soldering experience and parts familiarity are required, I use DOS 6x but have used DOS 3.3 with no problems, I have had zero problems with HC series with the GW R4 in my PC 5150 or XT 5160.
 
HC is fine as long as everything puts out at least 3.1ish volts for a logic high, and it all may very well do so the vast majority of the time.

But TTL is allowed to output as low as 2.4V for a logic high, and HCT is specifically intended to cater for that.

I'd just rather not be chasing down a problem caused by a marginal signal. :)
 
There is always a problem with "Borderline" IC's even with the best of manufactures, It's just something to keep in mind, Just a tip, Whatever CF card you use wipe it raw / Zero out the whole CF card First before trying it.
 
Some 74LS04's arrived today, so Ive swapped them in and there is no change. Its still very random whether the model of the card is read correctly (still mostly just get gibberish, although it is much more consistently able to read the model when I use PIO mode), and fdisk still naturally doesnt work.

FWIW I actually have PC DOS 3.10 installed, not 3.30 as I thought.

Still waiting on delivery of some LS/HCT573's and a different CF adapter to see if either of those will help.
 
A bit more investigation today, having received a replacement CF-IDE adapter (although tbh I think this one is about as rubbish as the first).

I decided to take a look at what was being stored on the card by fdisk, and noticed a bit pattern which kind of looked like bits 4 and/or 5 of every second byte was being flipped or corrupted somehow. So I figured maybe I had a dud 245 buffer or 573 latch (since those are the only things in the way between the system bus and the CF card), so I replaced all 3 of those parts but to no avail. I did notice the pattern seemed to disappear after replacing U1, but this still doesnt seem to be enough to make it happy. The 573's are still AHCTs while I wait for some LS/HCT replacements to arrive.

At least now I dont see gibberish text when the PC starts up, but the model number is still not entirely correct, and string data stored in the MBR still appears to be corrupt.

Before and after screen shots of the contents of the MBR attached for those who are interested.
 

Attachments

  • Screenshot 2023-01-05 at 12.03.21 PM.png
    Screenshot 2023-01-05 at 12.03.21 PM.png
    64.2 KB · Views: 4
  • Screenshot 2023-01-05 at 4.39.49 PM.png
    Screenshot 2023-01-05 at 4.39.49 PM.png
    62.2 KB · Views: 4
Hello there - I believe this is my first post here!

I'm working on an IBM PS/2 model 30 8086. The HDD unfortunately has gone so I built a Glitchworks Rev4 XT-IDE card. I need some help to find out what ROM to flash. The pre-built one I found on https://www.xtideuniversalbios.org/ does not seem to have the required module for my version of the card, if not mistaken. My ROM is 8KB but I have something larger if required.

I've read the instructions on how to build my own ROM but I was hoping for someone to point me to the right direction so I can get started. What modules shall I add? BTW, the card works fine on a Pentium with one of the pre-built ROMs, so I know all is in order :)

Thank you for your help!
 
A little update on my progress: a friend sent me his version of XTIDE, "XT" from 2013. That worked! I could see the drive, FDISK it, format it. I could install DOS 3.3 via SELECT. I can see the files and access the disk.
But I cannot boot from it. XT-IDE freezes on Booting C>>C and I have to power cycle.

I tried the current XT version available on the website but that does not recognise the drive at all.

So I am still a bit lost here :)
 
Hello there - I believe this is my first post here!
Welcome to these forums.

Consider adding your location to your profile:
Click on your profile name at the top-right of the screen, then select {Account details}, and there you will find a field named {Location}.

A little update on my progress: a friend sent me his version of XTIDE, "XT" from 2013. That worked! I could see the drive, FDISK it, format it. I could install DOS 3.3 via SELECT. I can see the files and access the disk.
But I cannot boot from it. XT-IDE freezes on Booting C>>C and I have to power cycle.
That sounds a like the very first issue described at [here].

I tried the current XT version available on the website but that does not recognise the drive at all.
It is highly probable that it, the XTIDE Universal BIOS (XUB), needs to be configured to suit the 'Glitch Works XT-IDE Rev 4'. See the {XTIDE Universal BIOS (XUB) - 'Device type' setting} section of [here]. Based on that, and assuming that your card is jumpered for hi-speed mode, the 'controller type' setting in the XUB would be configured for 'XTIDE rev2 or modded rev1'.

To do that configuration, you use a DOS program named XTIDECFG.COM. See note 2 at [here].
 
Hi @modem7

Thanks, appreciated. I am familiar with the XT-IDE configuration utility and I suspected that "device type" had to be on XTIDE rev 2 but the PS2 ROM I downloaded from the website did not have the relevant module - same for the XT version I believe. I appreciate I should build my own version but there are quite a few options and I was hoping for some directions before investing an awful lot of time experimenting! :)

My R625 which works on my Pentium machine is set to "XTIDE Rev 2". That works on the Pentium but not on the PS/2. The card is jumpered for high-speed.
What parameter do I need to change to match the "compatibility" mode of the card? I'll try your version indeed, thanks!

The boot issue is weird. Somehow it works now. PC-DOS 3.30 does not have the FDISK /MBR command available but somehow I must have changed something on the CF when I installed MSDOS 6.22 (that did require the FDISK /MBR) because after removing the partition before installing PC-DOS 3.30, the CF started booting correctly.
 
I appreciate I should build my own version ...
But only if you need to. That hasn't been established yet. No one reading this has yet indicated that there is something special about the PS/2 model 30 8086 that requires a custom build of the XUB.

What parameter do I need to change to match the "compatibility" mode of the card?
Presumably, you are experimenting. It requires two actions:
1. Change the jumpers on the Rev 4 card to the 'compatibility' setting (i.e. compatible with XTIDE Rev 1); and
2. Reconfigure the XUB: setting the device/controller type to 'XTIDE Rev 1'.

My R625 which works on my Pentium machine is set to "XTIDE Rev 2". That works on the Pentium but not on the PS/2. The card is jumpered for high-speed.
So the subject card (plus IDE device) now fully works on the Pentium (booting 'and all'). But when you move the card (plus IDE device) to the PS/2 model 30 8086, the card does not work. Is my understanding correct?

What exactly is the symptom?

When the card is fitted in the PS/2 model 30 8086, does the RAYXTIDE tool at [here] report any problems with the XUB ?
 
But only if you need to. That hasn't been established yet. No one reading this has yet indicated that there is something special about the PS/2 model 30 8086 that requires a custom build of the XUB.
Ok. I kind of assumed that because there is a pre-built "PS2" ROM which does NOT support "XTIDE Rev1" or "Rev 2" so I thought one option was to build my own PS/2 version with that module included.

Presumably, you are experimenting. It requires two actions:
1. Change the jumpers on the Rev 4 card to the 'compatibility' setting (i.e. compatible with XTIDE Rev 1); and
2. Reconfigure the XUB: setting the device/controller type to 'XTIDE Rev 1'.
Fantastic, that is what I wanted to know, thanks.

So the subject card (plus IDE device) now fully works on the Pentium (booting 'and all'). But when you move the card (plus IDE device) to the PS/2 model 30 8086, the card does not work. Is my understanding correct?

What exactly is the symptom?

When the card is fitted in the PS/2 model 30 8086, does the RAYXTIDE tool at [here] report any problems with the XUB ?
Let me find out exactly as things kind of changed after yesterday's tests. Thanks for pointing out that utility, I'll run it on my PS/2 and report back.
 
Ok. I kind of assumed that because there is a pre-built "PS2" ROM which does NOT support "XTIDE Rev1" or "Rev 2" so I thought one option was to build my own PS/2 version with that module included.
I just realized that I made the USE_PS2 define with the assumption that PS/2 machines are a superset of the AT class machines. That's probably a mistake?

The USE_PS2 define is for support of the software controlled HDD LED. It also changes the XUB splash text to say "PS/2" instead of "386", AT", "XT" or "XT+". You can try using that define together with the options used for the regular XT build in your custom build. This is assuming your specific model actually has the software controlled HDD LED (do all PS/2 machines have it?) and that you care about that functionality. It will work fine without it - you just won't see when the drive is being accessed. You will most likely need to undo the change I did to ModuleDependency.inc.

BTW, don't make the mistake of thinking that the Compatibility mode jumper setting is a safer setting than Hi-speed mode - because it isn't. You'll just reduce the performance. At this point, "Compatibility" just means that you can use very old versions of XUB and also hargle's BIOS with your XT-IDE card.
 
Back
Top