• Please review our updated Terms and Rules here

Setting up a CF card in a 5170

Mark2000 said:
Ok, that worked. Didn't know that trick. Thanks! Now if I could only get this cf adapter to report when the disk is active. Not working on the adapter itself or the led attached to the controller.
Click to expand...

Great. Try a different card - I think it was Chuck that found that some CF cards would drive only about 2mA via /DASP (the activity LED). To add, try disconnecting whichever LED you can too.
 
Sigh. I just couldn't leave well enough alone. I've been trying different CF cards so that I can have different system setups on each and I'm running into what seems like compatibility problems. I have one CF1 that works - an Apacer 64mb. The other two - a Sandisk 256mb and a Canon 8mb - come up with errors like the ones here: http://www.vintage-computer.com/vcforum/showthread.php?25286-CF-to-IDE-adaptor-read-errors. Those two are one's I had lying around the house and the problem is probably not fixable.

But the real problem is a new card I just received - a Transcend 512mb industrial. Formatting it in different places seems to have different effects: Fresh out of the box DOS 6.22 and 5.0 would not boot off a floppy. The drive would read for awhile and then the light would just go out. Oddly enough I could boot with 3.3 but anything I would run after that would end with the drive just stopping. Formatted on my Mac the DOS will boot, but on running FDisk I get a runtime error r6003 - which is usually what you get when you run FDisk from a dos shell. Formatting as FAT in Windows returns me to the original problem of no DOS boot. Formatting in the AMI Bios in the 286 boots DOS, runs FDisk, but gives me the r6003 error when I try to partition. All of these formats give me a invalid drive when trying to access C: (if they boot to a prompt).

So is this just an incompatible drive or am I doing something wrong again? If it is incompatible I'm 1 for 4 right now. Anyone know a 256 or 512mb card that is confirmed to work? I'd go Apacer again, but those does seem to be sold in that size anyone, not even on ebay.
 
I just had a quick read through this thread, so maby I missed something. Anyway, if you're having trouble with errors, perhaps it is in the way the partition software is working on a CF which can be a little different than the HDDs it is designed for. I found that overlapping partitions were a problem and that the Ranish Partition Manager solved that problem. YMMV
 
Ranish wouldn't load with even the disks that worked in DOS. I'm going to see if I can run it on Windows XP.
 
Last edited:
Mark2000 said:
Ranish wouldn't load with even the disks that worked.
Click to expand...
After I said that, I had a feeling that it might not. However, you could perhaps partition on another machine first.

In my case FDISK looked like it worked, but when I saw what it had done, I realized that the explanation for why things went bad after a while was that there were overlapping numbers, and Ranish solved that. Perhaps there is other software where you can specify exactly. BTW, the card I am using (which has worked flawlessly for months now) is a Kingston 8GB Elite-pro 133X.

There was some discussion here, and here. Note that pearce_jj points out that "DOS will align paritions on them straddling underlying blocks, i.e. commencing at sector 63" which is where I think problems arise.
 
I just got Ranish working on the 286. Didn't realize until digging in the docs that the latest version was only 32bit. I fixed the errors Ranish was finding in the CHS and the partition size info and reformatted the card. Unfortunately nothing has changed. Still can't boot or run anything from the disks partially working. I also got a divide error and crash when running Ranish with the card that wouldn't allow FDisk to run. What a mess, huh?
 
Mark2000 said:
I just got Ranish working on the 286. Didn't realize until digging in the docs that the latest version was only 32bit. I fixed the errors Ranish was finding in the CHS and the partition size info and reformatted the card. Unfortunately nothing has changed. Still can't boot or run anything from the disks partially working. I also got a divide error and crash when running Ranish with the card that wouldn't allow FDisk to run. What a mess, huh?
Click to expand...

And here I thought I was helping - sorry! I've got a terrible memory, but I seem t recall that I actually made sure that the numbers fit the blocks rather than let Ranish just "fix" it. Anyway, this could be a blind alley. Is it possible that some CF cards have non-standard, or strange, logical structure?
 
Hrm strange, I have the magicstor running smoothly on a dos/win3.11 system. I set the bios to what the overlay said to use, installed the overlay, and ran through the os installation. Rebooted, and working fine... >.>

Divide by 0 error, makes me think you have ram issues, have you run a memory test yet to rule it out? Could be a bad ide cable too...
 
twolazy said:
Divide by 0 error, makes me think you have ram issues, have you run a memory test yet to rule it out? Could be a bad ide cable too...
Click to expand...

I bet you're right. I didn't catch the software setup either - could even be a misconfiguration of memory manager or caching software.
 
Alright. To answer all your questions:

The Magicstor card runs fine. It's one of the only ones that does.

I've tested my memory through the BIOS and Checkit and it all checks out fine. The Divide error only happens with one of the cards - the Transcend 512mb. Its the same card that causes the floppy to cut out during boot if formatted by Windows. I have no other issues like that any other time.

The IDE Cable should be fine as it's working with the two working drives, but I can replace it.

There are no overlays or mem managers installed. I'm booting from a pure DOS disk with no config.sys or autoexec.bat.

Zombie is telling me in PM that it could be that my controller is BIOSless and therefore there's no support for the cards, but I then don't understand why some cards work fine. I'm wondering if this is a bios problem if an overlay would fix the problem on all the cards.
 
Are you using the XT/IDE Universal BIOS at the moment, and if so is the controllers own BIOS disabled?
 
I don't believe there are BIOS on the IDE card. I'm not using xtide. Would you stick them in the extra u27/u47 sockets or would you replace the main bios - in my case ami for 286.
 
neither will work. Do you happen to have a network card installed on this machine? If you do you just need an eprom, if not you will need some sort of interface card. XT-IDE be one example.
 
OK so the earlier results are direct enquiry of the hardware. It sounds that either one channel is faulty or there is a resource conflict (probably more likely) causing the lock-up.

If you need somewhere to host the XTIDE Universal BIOS, hopefully I will have some kit-form ISA cards available in a few weeks (all through-hole) that provide a 32K or 64K in-system programmable ROM (actually a flash chip).
 
Pearce, that bit about a conflict interested me so I removed my Soundblaster, which was causing problems with the IBM Advanced Diagnostics Disk as well as a Parity Error in CheckIt. I removed it, though, and am getting the same results. This is the board I have: http://stason.org/TULARC/pc/hard-di...B/ACCULOGIC-INC-Two-IDE-AT-drives-SIDE-3.html. If there is a conflict there may be setting to remove it... or make it worse.

I do have a NIC card with an empty bank. I guess I would need some good chap to burn me a chip to stick in it (for which I would pay them) if I'm going to try that solution. One other most likely dumb question out of curiosity - why isn't there an XTIDE ROM that fits the AT's extra empty banks? I know those are for BIOS extensions. Wouldn't it be logical to fill them up rather than sticking yet another card in an ISA slot?
 
You can stick the ROM anywhere, if there is a spare socket on-board that supports an 8KB ROM then that is certainly fine.

Many users here don't have the ability to write to 27Cxx chips and 28Cxx chips are horribly expensive (and again need a board capable of writing to them) - hence why I decided to create a little flash-based board for the job.
 
"Optional ROMs U17/U37

Motherboard sockets U17 and U37 are empty, available to be populated by ROMs, if desired.
Like the BIOS ROMs, they are to be of type 27256 (with speed of 150nS or faster), and like the BIOS ROMs, they are 8-bit ROMs supplying 8-bit data to a 16-bit data bus. U17 supplies the lower 8 bits of the 16 bits and U37 supplies the upper 8 bits of the 16 bits. As a result, the bytes from U17 appear at even addresses, and the bytes from U37 appear at odd addresses. In the 5170 memory map, their combined output appears at addresses E0000 (896K) through to EFFFF (960K-1).
While the ROMs could contain any data, BIOS expansion ROMs can be placed in U17/U37, because a part of the POST looks for the BIOS expansion signature (AA55h) at address E0000."

5170-extrabanks.jpg
 
That is why there's no xtide BIOS 'kit' available in this form (that and because it's a software project I suppose). Of course for anyone with 27C256 chips at hand and a programmer it's a neat solution.

Otherwise, the chips alone will cost you £20 and then we need a programmer or someone to program them. 28C256 chips should work too, but unfortunately they're even more expensive and still need some way to program them.
 
Last edited:
You must log in or register to reply here.
Back
Top