Reviving a Seagate ST-225 MFM HDD

[mrmanse]

Hi!

I have a 20MB Seagate ST-225 MFM HDD acting a bit strange, and I'd like to hear your thoughts on this.

The drive is connected to a small Western Digital controller card in a IBM 5160. I believe it's an after-market option, and also, the black cover in front of the drive is 5,25" half height, with no IBM logo.

The drive actually does work, to some degree, and the computer can boot from it, given enough time. A simple dos format gives about 8K bad blocks, all in the last few clusters, which probably isn't that much of a problem, perhaps even fixable. No, the problem here is that the drive constantly seeks for every track, which of course is horribly slow. At computer power-on, the drive gives off the characteristic sound of the heads moving track-to-track from track 0 to 614, and then back again, a few times, then it suddenly finds what it's looking for (i guess track 0), and the POST process finishes without error. I'm currently running Norton Calibrate 6.0 with minimal pattern testing, and for each track the drive can be heard moving the heads all the way from track 0 to track 614 and then back again, which takes about 14 seconds for each track. Norton Calibrate counts every head and track as a track so that gives 614*4 = 2456 tracks. Times 14 seconds per track will take many hours, and I'm still not sure it'll solve the problem. Currently about 100 tracks are done, and no errors found so far.

What do you guys think? My guess is that aging of the drive has caused the original low level formatting to be slightly off, so that when track x is selected, the heads actually move closer to track x+1, which the controller detects and for some reason chooses to search for the desired track in the wrong direction. Every time. Have anyone else experienced this? I have Norton Calibrate 6.0 (it won't run version 8.0, since that requires 80286) and SpinRite 5.0. Do I need anything else? I don't have any other HDD's or controller cards to swap with.

/Måns

 

[retrogear]

A couple observations:

My experience with Norton Calibrate is that it puts a lot of wear and tear on the drive and may expose and 'fix' problems but the problems seemed to return anyway. It claimed to 'recover' bad sectors and if the goal was to try and recover lost data
it was useful but in terms of 'healing' the drive, I don't think so. Others may have a different opinion. The tendency is to use software as a last resort to try and heal a sick drive.

What model is your WD controller? There is info online for WD controllers. Since you don't care about the data, I would initiate a low level format, then high level format, then test. We already know the drive is old and wearing out so don't 'exercise'
it too much. Just use ndd.exe to do a few passes. Don't hammer the drive with Norton Calibrate. Just doing the simple format steps will reveal if there is a hardware issue.

Larry G

 

[Stone]

Or, you could run SpinRite.

 

[lyonadmiral]

Or, you could run SpinRite.

In my experience SpinRite 5.0+ will not work with MFM drives and controllers. 3.1 or older is all that works for me.

 

[SomeGuy]

It is normal for Spinrite, and I believe Norton Calibrate, to perform a slow seek between each current tack and track zero. This results in each track taking slightly longer to process than the last as it goes along. That is normal, although I'm not exactly sure why they do that. Yes, even a ST-225 can take all day.

Now, you should not hear that kind of seeking when using DOS normally. During normal access, that "slow seeking" sound usually indicates the drive is resetting the heads because it has encountered a read error. If running Calibrate/Spinrite, or a fresh LLF does not fix that, then the drive probably does have a serious problem.

 

[Stone]

In my experience SpinRite 5.0+ will not work with MFM drives and controllers. 3.1 or older is all that works for me.

Just to clarify, I use SpinRite II v2.0. IMO, it's the workhorse of the SpinRite clan.

 

[retrogear]

BTW - thank you Stone for recommending Checkit 3.0 the other day. I'm using it to test floppy drives I'm working on and it seems pretty thorough.

To the OP - per Stone's advice, if Spinrite 2.0 can accomplish a low level format I would give it a go and report back here.
I haven't used that software and don't currently have an MFM hard drive that I want to experiment on. I would like to know how it works.
I found it by searching the VCF posts so you should be able to.

Larry G

 

[Stone]

The beauty of SpinRite is that it performs a non-destructive LLF. So the drive gets a new, fresh LLF without losing its data.

 

[mrmanse]

It is normal for Spinrite, and I believe Norton Calibrate, to perform a slow seek between each current tack and track zero. This results in each track taking slightly longer to process than the last as it goes along. That is normal, although I'm not exactly sure why they do that. Yes, even a ST-225 can take all day.

Now, you should not hear that kind of seeking when using DOS normally. During normal access, that "slow seeking" sound usually indicates the drive is resetting the heads because it has encountered a read error. If running Calibrate/Spinrite, or a fresh LLF does not fix that, then the drive probably does have a serious problem.

You may be right, I'm afraid. Calibrate needs about 1 more hour to finish, and then we'll know. I've repaired ST-225's before with calibrate, and a slow track-to-track seek before moving to next track I don't recognize, but calibrate does a fast move to track 0 and back between each track formatted. I guess it uses track 0 as a reference and want make sure to do the formatting om the right place on the surface. Like this:

A. start at track 0
B. jump to track n
C. format track n
D. jump to track 0
E. jump to track n+1
F. format track n+1
G. jump to track 0
H. etc.

I believe I can hear the B and E jumps, and they are NOT succeeded by a slow track 0 - 614 - 0 seek, but i think D and E fast seeks to track 0 are. So the problem could lie with track 0. That even makes sense since the track-to-track noise is heard also at POST. As soon as calibrate finishes I'll see what happens if I simply connect power to the drive without cables to controller, if it does the same then at least it should'nt be a problem with the controller.

Btw, the controller is a Western Digital WDXT-GEN2 PLUS

 

[Chuck(G)]

I have an XT clone with the original Quantum Q540 drive in it. The other day, I was looking for some data and couldn't find my backups for the machine. So I hooked up the system and--"0 hard disks" message. Gave the HDA a couple of raps with my hand and the disk started spinning and calibrated. Rebooted and everything was fine.

But I'm not stupid--one of these days that machine isn't going to boot. Spinning rust doesn't last forever. So, using Interlnk/Intersrvr, I copied all the data from the Q540 to another system using a parallel port "Laplink" cable. The curious thing was that the XT was running IBM PC-DOS 4 and the destination machine was running Win98SE on a FAT32 volume--and the copies of the Interlink files came from an MSDOS 6.22 disk.

Worked just fine, so at least I'm good if I have to replace the Quantum.

My recommendation is to back up your data and do a plain old LLF on the hard disk and transfer your files back. Should be only a couple of hours, tops.

 

[SpidersWeb]

With a DOS boot disk and FORMAT/FDISK/DEBUG on it:

A:> debug.com
- g=c800:5

<hopefully here the Western Digital Format Utility starts up from ROM>

Then FDISK, then FORMAT C: /S etc
I'd use an interleave of 5 on a 5160, but I'm sure Norton Calibrate or similar would've already mentioned the best setting?

I like this method for two reasons:
- doesn't waste time trying to read anything (since you've high level formatted by the sound of it anyway?)
- if the controller requires a configuration track, it's built in utility will write it correctly

Downside is the drive will be completely blank afterwards.

Just my two cents. Everyone has their own preferences and favorite tools.

 

[woodchips]

Ah, interesting. I had some ST225 20MB drives to test and the first off the pile did exactly what yours is doing, recalibrating back to zero between every track. Took ages. I have a WD WX1002 XT contoller and just used the low level format routine in the BIOS.

What happened next is a bit hazy but I think I manually rotated the stepper from one side to the other, don't think the disc was rotating. Won't damage it anyway. After that the drive formatted correctly taking 10 or 15 minutes from memory.

 

[mrmanse]

Ah, interesting. I had some ST225 20MB drives to test and the first off the pile did exactly what yours is doing, recalibrating back to zero between every track. Took ages. I have a WD WX1002 XT contoller and just used the low level format routine in the BIOS.

What happened next is a bit hazy but I think I manually rotated the stepper from one side to the other, don't think the disc was rotating. Won't damage it anyway. After that the drive formatted correctly taking 10 or 15 minutes from memory.

Manually rotated the stepper? How? Do you mean opening the hard drive?? Oh, I really wouldn't want to do that.

The Norton Calibrate procedure did very good, if any I'm afraid, and the drive is still moving track-to-track at certain operations. I did also do another low level format (calibrate is low level format) with debug and G=C800:5 as SpidersWeb suggested, and while it finished very fast (only 10-15 minutes), it too didn't improve the drive. So now I'm running a level 5 spinrite 5.0 session, currently at about 60%, and I can hear the drive stepping like before, but the behavior is slightly different from calibrate. Don't know why, could be Spinrites way of doing things.

But I have another question. The Western Digital WDXT-GEN2 Plus controller has settings for 4 different drives, and the currently selected is setting 3 (4 heads, 615 tracks). But hey, the drive only has 614 tracks?? Is the controller jumpers set wrong? Is it looking for a track that doesn't exist? And how on earth wasn't that detected for 20 years??? There is no option to set it to 614, but I can set it to 612. I can certainly live with the slight loss of capacity if it solves my problem. What do you think? Can anyone with the same configuration of controller and HDD confirm working settings? Thanks!

 

[Stone]

As mentioned previously, SpinRite 5.0 is *not* what you want to use. By then the useful MFM functions had been depreciated out of it. SpinRite II 2.0 is the ideal program.

 

[SomeGuy]

the currently selected is setting 3 (4 heads, 615 tracks). But hey, the drive only has 614 tracks?? Is the controller jumpers set wrong? Is it looking for a track that doesn't exist? And how on earth wasn't that detected for 20 years??? There is no option to set it to 614, but I can set it to 612. I can certainly live with the slight loss of capacity if it solves my problem. What do you think? Can anyone with the same configuration of controller and HDD confirm working settings? Thanks!

The specs for the ST-225 are:
CYLINDERS ________________________________615
HEADS ____________________________________4

The cylinders are numbered 0 to 614.

 

[mrmanse]

As mentioned previously, SpinRite 5.0 is *not* what you want to use. By then the useful MFM functions had been depreciated out of it. SpinRite II 2.0 is the ideal program.

The specs for the ST-225 are:
CYLINDERS ________________________________615
HEADS ____________________________________4

The cylinders are numbered 0 to 614.

Thanks to both of you! I've now located SpinRite II 2.0 and am presently running a extra thorough scan, currently at 2%. It takes about 13 seconds per track (out of 2455) so the whole process should take about 9 hours. However, my efforts so far have had some effect. The slow seek track-by-track from 0 to 614 and back again is heard not as often as before, that's good, but the downside is that the slow seek at POST is changed into a machine-gun ta-ta-ta-ta-ta-ta sound.

So, to sum up the current status:

Low level format through controller: Did nothing except clearing all data, as expected. No strange sounds, no healing effects.
Boots, reads, writes in DOS without obvious problems: mostly, has about 20K bad clusters grouped at the end of the drive (and they are increasing)
Slow seek track 0 - 614 - 0: at every track with calibrate, at some tracks with dos format and spinrite 5 (but with spinrite II 2.0 it hasn't so far)
Machine-gun: only at POST (or drive power-up)

I'll get back later with more updates.

 

[Trixter]

The beauty of SpinRite is that it performs a non-destructive LLF. So the drive gets a new, fresh LLF without losing its data.

Assuming the track can be written to. Consider:

1. Read track containing data
2. LLF track
3. Write test pattern
4. Test pattern fails
?. Where to put track of data that was read?

I believe what Spinrite does in that situation is try to relocate the data to free sectors on other tracks and then rewrite the cluster chain to compensate, but Spinrite isn't thorough enough to test those tracks first for validity. So it just dumps the data into untested tracks if those tracks are ahead of where it was testing.

Spinrite is mostly snake oil. The only time I pull it out is when I want to put an MFM in active use that 1. has an obviously wrong interleave, 2. the defect list was removed from the drive so you don't know what to enter in the controller's own LLF utility, and 3. I'm too lazy to back everything up (ie. common data) but don't really care if it's saved or not. I was a paying customer for years until I learned more about hardware, and Gibson.

 

[mrmanse]

Well, it's about time to give up, I guess. I've run the HDD controller low level format, norton calibrate 6.0, spinrite 5.0 and spinrite II 2.0, norton disk doctor 6.0, and dos format all of them a couple of times. The drive hasn't died, nor has it magically repaired itself. I guess I was naive, but my previous experiences with ST-255 is way better than this. The drive works well in DOS, except for the last 5 or so cylinders, which are mostly dead. And at power-on there is that clicking ta-ta-ta-ta-ta noise, but only then. Better than nothing, I guess.

The process of trying to fix the drive has left me with still some questions, I hope someone can explain it to me.

The drive is supposed to have 615 cylinders, but only 614 are accessible. That is not an error from starting the count from 0, but only 614 cylinders are in fact reachable to any of the applications I've used. Calibrate formats 2456 (614*4) cylinders, and spinrite checks 41752 sectors (614*4*17). Norton diskedit 6.0 lets me select and edit any physical sector from track 0 to track 613 (which is 614 tracks). So, the only explanation is that there must be a reserved engineering cylinder (as spinrite calls it). Where on the disk is such a cylinder located? before track 0 or after track 613 (after the 614th track)? I'd expect it to be the latter, since the surface have many defects in that area. And, is this track responsible for the ta-ta-ta-ta noise at power up? Is accessing this track part of the drives internal power-up process? Is there any way to investigate that part of the drive and possibly give it a shot with low level formatting? I'm not looking to repair bad blocks here, but to get rid of the ta-ta-ta-ta would be nice. Can anyone explain this? I've googled a lot without finding anything useful. Thanks

 

[Stone]

The drive works well in DOS, except for the last 5 or so cylinders, which are mostly dead.

Why don't you create a partition that doesn't include those cylinders, e.g., 605 to 609 cylinders or something along those lines? Then the obviously bad area won't be in play at all.

 

[mrmanse]

Why don't you create a partition that doesn't include those cylinders, e.g., 605 to 609 cylinders or something along those lines? Then the obviously bad area won't be in play at all.

Yes, that's a good idea! I think it would be easy to find the track numbers of affected cylinders, and to create a partition that doesn't include them. However, the ta-ta-ta-ta noise at power-on probably wont go away by doing that.