VOGONS


2.88MB ED diskette drive troubles

Topic actions

First post, by nathana

User metadata
Rank Newbie
Rank
Newbie

Hello,

I recently purchased 2 "refurbished" (appear to be heavily used but somewhat cleaned up) 2.88MB Extended Density 3.5" floppy diskette drives. They are slimline model drives, originally packaged in an UltraBay carrier by IBM and intended for use with the ThinkPad 75x/76x line. (I was only in the market for one, but the price was so good I went ahead and purchased a spare to go with it.)

The particular ones I have are manufactured by Alps Electric, model DF328N101A.

I originally purchased these intending to try to get at least one of them to work in an old ThinkPad 770 I have, either by gutting a 1.44MB HD drive from its UltraBay II caddy and replacing it with this, or by putting it in a SelectaDock III after swapping out the UltraBay II slot option in the dock with the original dual-UltraBay I option from the SelectaDock II (they are interchangeable).

However, when I bought these, I also ran across some slimline floppy (26-pin, 1.0mm-pitch FPC) to standard desktop PC floppy (34-pin standard ribbon) adapters. If you free the drives from their UltraBay shells, you find that these are just standard slimline laptop floppy drives under-the-hood. And it was going to be easier to test these first on a desktop PC with those adapters, so that's what I did: tried hooking them up to a motherboard that has 2.88MB floppy support.

However, I can't get *either* drive to work. At this point, I've tried both drives on 3 different PCs, 3 different cables/adapters, and multiple different known-good diskettes.

The symptoms are 100% identical between the two drives: they power up just fine, they perform seek during POST just fine, BIOS does not throw out any floppy drive error codes, it spins diskettes just fine, and the heads seek just fine. But when I ask it to read a disk from DOS, I just get back "General failure". If I try to format a disk from DOS, I get back "Invalid media or Track 0 bad". If I try to boot from a DOS boot floppy, the BIOS tells me there is no boot sector on the diskette. This is even on regular 1.44MB HD media.

I have tried setting the BIOS to 1.44MB instead of 2.88 just to see if that would at least change anything about the drives' ability to read 3.5" HD diskettes, but they behave identically regardless of how BIOS is configured. If I replace either of these ED slimline drives with an HD slimline drive, that HD drive works just fine in the same context (using the exact same adapters/cables, connected to the exact same PC, and reading or booting off of the exact same diskettes).

I have cleaned the heads on both drives, but no change.

At this point, I am baffled. I suppose both drives could be defective, but in the exact same way? This makes me wonder if there is something obvious I'm missing, or if perhaps there is some special consideration that needs to be taken into account with these drives that you wouldn't have to deal with on a slimline HD floppy drive (perhaps the ED density select pin isn't being bridged by any of the adapters I've tried using? could that cause this? but wouldn't at least HD media be readable in this drive without that signal??)

Any ideas or further troubleshooting thoughts would be much appreciated!

Reply 1 of 36, by vetz

User metadata
Rank l33t
Rank
l33t

Could be that these laptop floppy drives is pre-configured to only work with the floppy controller in a Thinkpad.

See here: https://groups.google.com/g/comp.sys.ibm.pc.h … /m/zun9PhBAnjcJ

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 2 of 36, by Zup

User metadata
Rank Oldbie
Rank
Oldbie
vetz wrote on 2021-04-11, 10:49:

Could be that these laptop floppy drives is pre-configured to only work with the floppy controller in a Thinkpad.

See here: https://groups.google.com/g/comp.sys.ibm.pc.h … /m/zun9PhBAnjcJ

Or they use ID0 instead of ID1 as most PC floppies...

I have traveled across the universe and through the years to find Her.
Sometimes going all the way is just a start...

I'm selling some stuff!

Reply 3 of 36, by nathana

User metadata
Rank Newbie
Rank
Newbie
vetz wrote on 2021-04-11, 10:49:

Could be that these laptop floppy drives is pre-configured to only work with the floppy controller in a Thinkpad.

See here: https://groups.google.com/g/comp.sys.ibm.pc.h … /m/zun9PhBAnjcJ

Interesting, but hard to tell whether the symptoms are an exact match with mine or not. The OP of that thread only mentions "formatting", and one of the respondents specifically says he can read/write but just not perform a format (though that person was using a 1.44 drive, so maybe not even the same underlying problem as OP).

These little slimline drives have zero jumpers on them that I can see in any case...

Zup wrote on 2021-04-11, 11:54:

Or they use ID0 instead of ID1 as most PC floppies...

I don't see how that could be the issue, given that the drive actually responds to commands but just can't seem to successfully complete them. If it were a matter of ID mismatch, I'd expect the drive to simply be a brick and not show any life at all, since it would never think it was being addressed.

I'm starting to think that the only way I can conclusively rule out faulty hardware is to actually find somebody local with a 750/755/760/765 ThinkPad who would let me try the drives in their system...what are the odds of that...*sigh*

Reply 4 of 36, by vetz

User metadata
Rank l33t
Rank
l33t
nathana wrote on 2021-04-12, 11:07:

I'm starting to think that the only way I can conclusively rule out faulty hardware is to actually find somebody local with a 750/755/760/765 ThinkPad who would let me try the drives in their system...what are the odds of that...*sigh*

I own a 750C and a 760ED. Thanks to your post a found a NOS Thinkpad (FRU P/N:66G5060) 2.88ED floppy drive in Japan for 15 USD, so I bought it. It doesn't hurt to upgrade the floppy drive on one of those laptops. I can atleast see if I get the same error on a Thinkpad.

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 5 of 36, by nathana

User metadata
Rank Newbie
Rank
Newbie
vetz wrote on 2021-04-12, 11:41:

I own a 750C and a 760ED. Thanks to your post a found a NOS Thinkpad (FRU P/N:66G5060) 2.88ED floppy drive in Japan for 15 USD, so I bought it. It doesn't hurt to upgrade the floppy drive on one of those laptops. I can atleast see if I get the same error on a Thinkpad.

My drives were not NOS, but very clearly heavily used (dents, scratches, etc.) and then cleaned up. I'm guessing your basically new drive will work just fine in your ThinkPads. The question is whether my drives are actually mechanically or electrically defective, or whether I am just missing something about making them work while they are interfaced to a standard PC...

Reply 6 of 36, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
nathana wrote on 2021-04-11, 04:57:

I originally purchased these intending to try to get at least one of them to work in an old ThinkPad 770 I have, either by gutting a 1.44MB HD drive from its UltraBay II caddy and replacing it with this, or by putting it in a SelectaDock III after swapping out the UltraBay II slot option in the dock with the original dual-UltraBay I option from the SelectaDock II (they are interchangeable).

I'm probably not reading closely enough, but were you able to try either of the drives in your 770?

I'm interested if you do get yours working as I've got a normal HD Mitsubishi MF355F-250MG from '93 that's developed similar symptoms (seek during POST is fine, but can't format disks) at some point over the last year. It had a couple of leaky capacitors on the +5V line, one looks to cover the two motors, the other the drive electronics. I've had a go at replacing those, but it doesn't seem to have made any difference.

Reply 7 of 36, by vetz

User metadata
Rank l33t
Rank
l33t
nathana wrote on 2021-04-12, 11:57:

The question is whether my drives are actually mechanically or electrically defective, or whether I am just missing something about making them work while they are interfaced to a standard PC...

True, were the adapter you bought expensive? Is it the one on EBay for about 5 USD?

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 8 of 36, by nathana

User metadata
Rank Newbie
Rank
Newbie
snufkin wrote on 2021-04-12, 12:27:

I'm probably not reading closely enough, but were you able to try either of the drives in your 770?

Not yet. I'd like to find a "donor" 1.44 UltraBay II drive first that I can use, so I don't have to sacrifice the only good one I have. (I can literally find ZERO information about anybody opening up one of these UB-II drives up, so I don't know how easy or hard it is to get inside the casing "non-destructively".)

I did, however, just run another test: I found a DD (720k) disk, and tried to read that.

Interestingly, one drive can read it perfectly (was able to XCOPY off the entire contents without any difficulty). The second drive can kinda-sorta read parts of it: I got a directory listing back, but "sector read errors" and the like while actually trying to read a file off of it.

So this leads me to believe that at least ONE of my drives is in SOME way defective...perhaps head alignment? (No service manuals available for these @#$@# drives though! And nobody apparently with any experience with them!)

As for the drive that can read 720k disks just fine, it's making me wonder whether there's something about the way that this drive model does density selection (& signals it to the controller) that's not quite the same as a normal PC diskette drive.

vetz wrote on 2021-04-12, 12:31:

True, were the adapter you bought expensive? Is it the one on EBay for about 5 USD?

Not sure which specific listing you are looking at. There are various PCBs from China that are roughly $6.50 USD before shipping. But I didn't want to wait that long.

I have tried 2 different ones that both perform identically for me, just slightly different form-factor:

https://www.ebay.com/itm/392362275195
https://www.ebay.com/itm/333390079817

Reply 9 of 36, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
nathana wrote on 2021-04-12, 12:43:

I did, however, just run another test: I found a DD (720k) disk, and tried to read that.

Interestingly, one drive can read it perfectly (was able to XCOPY off the entire contents without any difficulty). The second drive can kinda-sorta read parts of it: I got a directory listing back, but "sector read errors" and the like while actually trying to read a file off of it.

Huh, again, that's similar to my 355F. Can get a directory listing from most 720k and only a few 1.44M, but can't reliably read data from either. Testing with IMD's alignment test then the alignment seems ok (after I discovered how Mitsubishi did track 0 detection), but on any given track the read will only be able to read a few sectors, and it changes which sectors it can read.

On my drive I suspect a rotation speed issue at the moment. Checking with IMD, it gradually approaches 300rpm, but with some quite sudden jumps along the way, as though it's hitting 300rpm on average, but isn't stable. HD are 18 sectors/track, DD are 9 sectors/track, so maybe HD disks needs the rotation speed to be more accurate? I need to actually get my 'scope out though and check it, which won't happen for a few days.

Reply 10 of 36, by nathana

User metadata
Rank Newbie
Rank
Newbie
snufkin wrote on 2021-04-12, 13:02:

Huh, again, that's similar to my 355F. Can get a directory listing from most 720k and only a few 1.44M, but can't reliably read data from either. Testing with IMD's alignment test then the alignment seems ok (after I discovered how Mitsubishi did track 0 detection), but on any given track the read will only be able to read a few sectors, and it changes which sectors it can read.

I have found zero HD diskettes that either drive can read. They cannot even find track 0 on these disks form what I can tell.

On the perhaps-physically-fine drive, IMD alignment test gives it perfect results on all tracks of a DD diskette. On the iffy drive, it's not consistently able to read all 9 sectors on any given track on either side of the disk.

In IMD, *both* drives get a complete failing grade when alignment test is conducted with HD media (all '?' question marks).

So I think I'm fighting 2 different problems: probably some physical issue with one of the two drives, and then some kind of density select issue between drive and controller with both (so they're constantly stuck in 250kbps rate mode). The latter would explain why neither HD nor ED media work at all.

Reply 11 of 36, by weedeewee

User metadata
Rank l33t
Rank
l33t

verifying the continuity of the little format/WP/density/Changedisk switches should be easy enough... if one is able to get access to them.
however... replacing them if faulty might be a problem, though it might be possible to open'm up and clean'm, thus fixing them.
Did it ,last year, on an ibm usb floppydrive. all three switches in there were giving a bad contact. though the only effect I recall was that it didn't recognize a disk change. it should also have had issues with either always being write protected or never being.

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 12 of 36, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Good idea on the HD/DD select. I've just tested both the HD/ED switch, and the WE/WP switch on my drive (it's not ED, but the loading bit and motor control have all the PCB markings for ED, and a signal is sent back to the control PCB). When the pins are pushed in they read open circuit, so I assume they're supposed to be NC switches. When the pins are up then the reading is erratic (this is with the switch unsoldered and on its own), from ~9k ohm -> 1k ohm. Normally I'd expect a switch to either be short or open.

Just tried cleaning with IPA, but no joy yet. May just solder it shut for HD and see if that makes a difference.

[Damn springs... I now have two springs, two pegs, two contacts, one housing, and no clip to hold it all together. Time to go searching]
[HAha, found it, small clear plastic cover]

Reply 13 of 36, by weedeewee

User metadata
Rank l33t
Rank
l33t
snufkin wrote on 2021-04-12, 14:19:

When the pins are up then the reading is erratic (this is with the switch unsoldered and on its own), from ~9k ohm -> 1k ohm. Normally I'd expect a switch to either be short or open.

Yep, a switch should read an infinite resistance in one state and close to zero in another state, measured out of circuit.

[Damn springs... I now have two springs, two pegs, two contacts, one housing, and no clip to hold it all together. Time to go searching]
[HAha, found it, small clear plastic cover]

Hahaha 😀 yeah, those tiny bits are very finicky 😁
I used some dishcleaningsoap&water and a toothpick to rub it clean and some IPA to rinse.

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 14 of 36, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Right, that seems to have at least partly worked for me, although it may just have been alignment. Only did the ED/HD switch (who needs working write protect anyway?). Cleaning the switch made it so that Win98 would at least open the drive and could slowly load a few text files from an HD disk. Computer sill wouldn't boot from it and DOS would give a general failure error when trying to access a:. Rechecked the alignment in IMD and found that whilst the beep test sounded good and it could read data on all sectors, there were a few alignment positions that all gave a good reading (I may be misunderstanding what the beep pitch means in the alignment test). Picked one and could then access the disk and boot from it.

So I think there was a switch issue (I couldn't access this disk in Win98 before, it would ask to format it), followed up by needing better alignment. As it is now, it's very slow to copy files off a disk, so I'm going to go and read the IMD instructions and see if I can improve the alignment in case that's still an issue.

Just to deviate from the thread slightly, alignment of this drive is fun. The bottom read head has a fin on it that moves along above the control PCB. The control PCB has an optical sensor on it which can be blocked by the fin on the read head when the head reaches 0. The position for track zero is set by the exact location of the control PCB in relation to the drive chassis, and the PCB only has a couple of screws holding it in place. So alignment is done by slackening the screws, moving the PCB a bit, then tightening them back up again. Of course, I didn't realise this when taking the PCB off in the first place to get at the faulty caps (definitely faulty, they read a few picofarads), so didn't mark exactly where it was because who on earth would make alignment dependent on the PCB location?

Reply 15 of 36, by weedeewee

User metadata
Rank l33t
Rank
l33t
snufkin wrote on 2021-04-12, 16:01:

(who needs working write protect anyway?).

Anyone who ever experienced the Change Disk signal not working.
Insert another disk, OS has no clue disk has changed and happily writes cached data back to disk...

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 16 of 36, by nathana

User metadata
Rank Newbie
Rank
Newbie

I'm having a difficult time finding any write-up (much less a good one) on how the different forms of density select work (how it can be signalled between the drive and the controller). I can find references here or there to "host density select" vs "media density select" (which is the same thing maybe as "media sense"...?), but no actual documentation on the difference OR how they manifest / are implemented. I can make educated guesses by trying to read between the lines, but it's frustrating that nearly every resource out there that even mentions this in passing simply seems to ASSUME that the reader is already familiar with the difference, and I can find NOTHING that goes out of its way to explicitly explain anything. UGH.

And that's just on the DD vs. HD density select front, to say nothing of what might be different once you throw ED into the mix!!

From what I *think* I can gather, most PCs expect the floppy drive to tell the controller what bitrate (DD 250k / HD 500k / ED 1000k) to switch into via the density select line(s), and these PC clone-compatible drives are looking to the holes (& their exact location/offset) punched on the side of the disk opposite the write-protect hole in order to determine what the type of media is, while other drives (mainly models intended for use in PS/2 machines) either were just blindly told what bitrate to use by the controller, OR perhaps still performed media sense and somehow signalled that to the controller, but it was still the host's ultimate call what bitrate to use...?

Let's say that this is the issue, and the problem in my case ultimately boils down to the fact that at least one of the density select pins is always being seen by the floppy controller as being in the position representing DD media (whether that's low or high, I don't know, because *where is this documented*). Can I maybe just work around this via hardware mod to the floppy cable (forcing both DD/HD density select pin and ED density select pin, from the perspective of the floppy controller, to what media I have in the drive)? Or is there possibly a way that *through software* I can force an override of the media type by instructing the floppy controller to switch to the bitrate that the media I'm using was written with??

Thanks for any leads! The lack of clear info on this subject (at least that I can find) is driving me nuts!!

Reply 17 of 36, by nathana

User metadata
Rank Newbie
Rank
Newbie

Curiouser and curiouser.

Now both drives "work", insofar as they can both reliably read 720k-formatted disks. Not sure what was up with one drive reading them unreliably earlier. Best guess/theory is that I might've either not had the flat ribbon cable quite all the way seated in the drive properly, or accidentally had something shorting out...regardless of how, I suspect there was signal loss between drive and controller.

So now they both behave identically again, and I'm back to thinking that likely neither one is defective.

What's more interesting is that I ran some writing tests, and though both drives can READ a 720k disk *perfectly* (zero errors on any track from either head via IMD Alignment Test!), if I try to WRITE to one, it wipes whatever sectors it touches. A format appears to work initially, but the disk is unusable after that (same symptoms as if I had a formatted HD disk in the drive). If I copy a file to the free space on a disk formatted on another machine, and then try to copy the file back off, I get sector read errors on the file's first sector. Not only that, but once I eject and re-insert the disk, I can no longer get a directory listing, because when it updated the FAT, it essentially wiped it. If I ask IMD to do a low-level 720k format, then after that, not even IMD Alignment Test can find any readable sectors!

So in summary: it can only read DD formatted disks, not HD or ED, and it cannot write to any disk without destroying its contents. It really seems like the most logical conclusion is that there is some sort of disagreement between the floppy controller and the drive about what bitrate & recording mode to use.

Anyway, I managed to find another 770 "parts" drive for cheap. It's on its way. I'll gut it, stick one of these drives in that shell, pull my 770 out of mothballs, and try the drive with that machine. If *that* doesn't work, I'm-a go insane...

Reply 18 of 36, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

I've spent lunch tracing some connections. I think the DD/HD/ED switch set are signals in to what looks like the main controller on the drive, and that then generates any signals sent back to the motherboard. Pin 2 on the floppy connector seems to sometimes be listed as 'Density Select', but I think it's actually 'Reduced Write Current' and is a signal to the write heads on the drive. When low then it's set to write DD, when high for writing to HD.

So this may actually be a red herring, and the switch only matters for setting write strength when writing to the disk.

I think means the motherboard has to work out the disk density based on the transition rate on pin 30, RDATA, which looks like it spits out data whenever the disk is moving under the read heads. Transition rate looks to either be 2/4/6us for HD and 4/8/12us for DD (I assume it'd be 1/2/3us for ED). So other than for writing, the DD/HD/ED signal is just internal to the drive. Maybe. I haven't been able to find any good documentation either, probably don't know the right words to search for.

FWIW, on my Mitsubishi, the HD/ED switches are both pulled to +5V with no disk in, and pulled to GND via a 22k resistor if the pin is pushed in. So for DD they'd both be pulled to GND. For HD or ED then only one would be pulled to GND. There isn't a separate Eject switch (there's a pad on the PCB for one), so I guess the drive controller knows a disk has been removed if both pins are pulled to +5, and can signal Disc Change (pin 34) back to the FDC. Which was probably what Weedeewee found.

Not sure this gets you any further forward though. Sorry.

Reply 19 of 36, by weedeewee

User metadata
Rank l33t
Rank
l33t
snufkin wrote on 2021-04-13, 14:39:

FWIW, on my Mitsubishi, the HD/ED switches are both pulled to +5V with no disk in, and pulled to GND via a 22k resistor if the pin is pushed in. So for DD they'd both be pulled to GND. For HD or ED then only one would be pulled to GND. There isn't a separate Eject switch (there's a pad on the PCB for one), so I guess the drive controller knows a disk has been removed if both pins are pulled to +5, and can signal Disc Change (pin 34) back to the FDC. Which was probably what Weedeewee found.

All I found on the drive I once examined was that there were three switches, and all three needed cleaning.
Also it was just a DD/HD drive, and still had three switches, one for RW/RO, one for DD/HD, and I guess the third one was for disk change.

I can only assume that an ED drive has a fourth switch, since the hole for DD/HD is different from the hole/indent for ED.

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port