• Please review our updated Terms and Rules here

Safe to open up an mfm hdd to try and fix it?

it most certainly was in drives larger than 5"
You are correct, what I should have said was that there were no commercially available PC hard disk drives which had the technology until the very late 1980s. The closest we had was Quantum/Plus's LPS system.

Well, believe what you want to, but I do observe that the end stop is adjustable. In any case, the drive is trash now, so it hardly matters.
It's still probably salvageable but not with the tools he's got. The mechanical limit is a set screw which can be adjusted but that is not the same as the sensor.
 
I'm opining that there is no sensor. Of course, you could disprove this by locating the relevant circuitry on the PCB. I think that the stepper current is monitored to figure out when the track 0 position is hit. Observe that you need only do this once per power cycle. So step slowly out to the stop and Bob's your uncle. Reports of the sound the drive made when power-cycled would seem to confirm this.

(FWIW, embedded servo was used on a 5.25" floppy drive even before this particular drive dropped off the assembly line.)
My recollection of the Kyocera drives is that they weren't particularly fast, but they were cheap.
 
I'm opining that there is no sensor. Of course, you could disprove this by locating the relevant circuitry on the PCB. I think that the stepper current is monitored to figure out when the track 0 position is hit. Observe that you need only do this once per power cycle. So step slowly out to the stop and Bob's your uncle. Reports of the sound the drive made when power-cycled would seem to confirm this.

(FWIW, embedded servo was used on a 5.25" floppy drive even before this particular drive dropped off the assembly line.)
My recollection of the Kyocera drives is that they weren't particularly fast, but they were cheap.
The track 0 is broken out onto the ST-506 control header as well, it's a non-optional tracking signal. The position must be reported in some way to the controller, so there must be a sense method somewhere. Unfortunately, the track 0 line connects straight to a leg on the microcontroller, just like almost all of the control signals. You would need to decap the uc and read it's ROM to know more from there.

They are definitely not fast drives. Mine is about on part with an ST-225 in terms of average random access time (~55ms).
 
I would expect that Track 0/ on this drive would be an output from the MCU. Doesn't invalidate my speculation. Step out slowly until the motor current increases, step in a couple of steps and call it track 0. Simple.
 
I would expect that Track 0/ on this drive would be an output from the MCU. Doesn't invalidate my speculation. Step out slowly until the motor current increases, step in a couple of steps and call it track 0. Simple.
Track 0 on almost all drives is run through the uc. It's even this way on my ST-412 - the microcontroller keeps track of this information as well, partially for their own seek test.
What makes this theory any more credible than mine? Neither of them are based on real evidence from this drive.

Not to mention, the head stack on my drive and his example of a working drive never contacts the physical stop during the drive's seek test - Thus there is no motor current increase. While plausible, the evidence is still not there.
 
I went ahead and dismantled the stepper. Seems like no big deal to do so, with a bit of effort. Two bolts mount the stepper and 2 bolts hold it together. Easy to pry apart and inspect.

There was absolutely nothing in there that in any way looked like a track 0 sensor.

The only thing it could be, is if the magnetic characteristic of the armature in there is somehow not symmetric, but I kinda doubt that.

Reassembled, the stepper still spins happily and acts just like before.

So, it is still a bit of a mystery at least to me how this drive detects track 0.

I see no electrical wiring that I can attribute to track 0 detection. There is a ribbon cable with that carries the signals needed for the read/write heads. Since that ribbon cable is eventually mounted on the read/write head arm, one could imagine that there could be a little switch mounted on that arm to detect rotation. But, there is no visual indication of such a switch anywhere.

I haven't been able to find any reference to a stepper with built in track 0 detection, either. Probably just my search skills.
 
If we assume for a second that it is in fact the bump up against the stop, that causes a change in stepper current....

Then, given I have swapped the entire controller board with a working drive, and the track zero issue follows the hardware not the controller....

Then I think the only explanation is that the bump stop stepper current must be out of spec wrt the controllers ability to detect the fluctuation.

I could confirm this by swapping the stepper on the faulty drive...but enough is enough!

Been fun, glad I got one disk to work.
 
"What makes this theory any more credible than mine? Neither of them are based on real evidence from this drive."

Um, you were saying?

Let's have a look at the stop:
Screenshot_2022-11-20_17-48-38.png

Note that the stop has a little "tab" projecting. It's clear that it must mate with a corresponding recess in the positioner arm. Is there anything else on the positioner arm in that position that might provide an electrical signal?
 
There is a flex cable mounted to the arm with an IC mounted on thr flex cable and wires connecting to the heads. So I assume this IC is a preamp or something. There is no other circuit, no switch.

In the photo you have accurately depicted the stop. There is not a recess on the arm.. a flat surface on the arm bangs up against that stopper.

Seems to me that, within the angle of a single step, that exactly at what incremental angle the arm contacts the head must change the change in current. Stopping early in one step vs stopping late would cause a different current, no?
 
There is a flex cable mounted to the arm with an IC mounted on thr flex cable and wires connecting to the heads. So I assume this IC is a preamp or something. There is no other circuit, no switch.

In the photo you have accurately depicted the stop. There is not a recess on the arm.. a flat surface on the arm bangs up against that stopper.
The IC on the flex cable is indeed a head preamp.
You are correct, there is no recess on the arm. The round protrusion is to ease adjustment of the physical stop - It's main purpose is to prevent the heads from errouneously being driven off the surface of the platters, which would be devastating. Without this stop it would be possible for a malfunctioning controller to effectively decapitate the drive.

Then, given I have swapped the entire controller board with a working drive, and the track zero issue follows the hardware not the controller....
This brings up a great point. If the problem was entirely restricted to the controller, then swapping controllers would have fixed the issue. Something mechanical must have gone wrong.

There was absolutely nothing in there that in any way looked like a track 0 sensor.
The only thing it could be, is if the magnetic characteristic of the armature in there is somehow not symmetric, but I kinda doubt that.
Reassembled, the stepper still spins happily and acts just like before.

So, it is still a bit of a mystery at least to me how this drive detects track 0.

I haven't been able to find any reference to a stepper with built in track 0 detection, either. Probably just my search skills.
I don't think there's any use to the armature being that way as this stepper can make one full rotation for a sweep of the disk - The detection would have to determine between a regular seek and being at track 0.
I do wonder how it detects it's stop, then. It must do it in some way.

The main example of end stop detection I was aware of was the afforementioned CMI 3212 (Mistyped model number, my bad) which includes it as a cam inside the stepper. I've not seen any drives with physical limit switches, external sensors are all optical (Several Seagate drives, all Miniscribe stepper drives, most Tandon drives, etc).

Note that the stop has a little "tab" projecting.
In case you haven't had the displeasure of hearing it before, drives slamming the actuator against their end stop makes a very audible "clack" or banging noise. These drives would be an awful lot noisier if this was their detection method. If you would like, I can even video my unit doing seek tests where you definitely cannot hear it hitting the end stop. I might even be convinced to lift the lid on it again and stick some shims in to show that it stops just early of the mechanical stop.
 
My point was that the drive only has to hit the stop at the power up sequence. Since the sound at the power-up is different, I can readily appreciate that this is done to avoid slamming hard into the stop.

Why would the stop be adjustable otherwise--and it's clearly adjustable. Not only that, but there's not any elastic bumper on it.

There is a way to verify this with a working drive. Insert a shim (a thin bit of cardboard should be sufficient) between the end of the stop and the head carriage. Alternatively, if you're convinced that the stop positioning has nothing to do with it, mis-adjust the stop on a working drive.
 
My point was that the drive only has to hit the stop at the power up sequence. Since the sound at the power-up is different, I can readily appreciate that this is done to avoid slamming hard into the stop.

Why would the stop be adjustable otherwise--and it's clearly adjustable. Not only that, but there's not any elastic bumper on it.
Units which have bumpers typically do experience regular contact. This does not, more evidence that this is likely not intended behavior.

Track 0 is adjustable in some way on all drives - This is done at the factory for differences in rotary band shapes. The interruptor blades on Miniscribe drives can be rotated to adjust the position of track 0 all the same, this was actually done in the field on occasion, so often so that Miniscribe started to ship drives with track 0 adjustment software you could activate by powering a drive on with the self test jumper in place and removing it when the light blinks. The drive would seek all the way inwards and rattle against it's stop and then pull back the exact number of steps (Typically 615) to stop at what should then be track 0. From there you adjust the blade until the indicator lamp switches off and now you have a valid track 0 position.

I actually wish more drives made the process this simple - Adjusting this on a Miniscribe 8425 is a joy compared to doing it on something like a Microscience HH330.
 
The metal stop bang is so loud... I find it a bit hard to believe.
So I tried the following, to take the lash out of the adjustment.

. Let the drive spin up, bang around and settle at the last step that it can muster.
. Drive spins down
. Leave powered. This locks the stepper
. Adjust stop so that it butts up against the arm. No gap.

This setting I had hoped would eliminate the racket but it didn't really make a dent pardon the pun.
 
The metal stop bang is so loud... I find it a bit hard to believe.
So I tried the following, to take the lash out of the adjustment.

. Let the drive spin up, bang around and settle at the last step that it can muster.
. Drive spins down
. Leave powered. This locks the stepper
. Adjust stop so that it butts up against the arm. No gap.

This setting I had hoped would eliminate the racket but it didn't really make a dent pardon the pun.
You want some gap between the swingarm and the stop. I have always used a set of feeler gauges for this - Your local auto parts store will have them. Typically I will go for 0.0030" but your mileage may vary. Occasionally manufacturer documentation will be kind enough to specify this number, but definitely not always.
 
for fun, I decided to move the stepper from the good drive to the bad drive.

that did not work. same problem.

So I moved the stepper back to the good drive. I had a really hard time getting it to find track 0. eventually,.. by playing with the placement and mounting of the stepper, and the tension in the little steel belt, I got it to work.

Which leads me to conclude... there must be servo information marking track zero.
 
It could well be that there's a "track 0", but not the same track "0" that the external connection sees. Sort of a track -1 that's not accessible through the normal ST506 interface. That could work and save a dollar or two. A logic analyzer might be able to verify that. You can do wonderful things with MCUs.
 
for fun, I decided to move the stepper from the good drive to the bad drive.

that did not work. same problem.

So I moved the stepper back to the good drive. I had a really hard time getting it to find track 0. eventually,.. by playing with the placement and mounting of the stepper, and the tension in the little steel belt, I got it to work.

Which leads me to conclude... there must be servo information marking track zero.
The rotary band's tension and adjustment is absolutely critical. If there was servo information, the drive would have found track 0 with or without your adjustments.
 
for fun, I decided to move the stepper from the good drive to the bad drive.
that did not work. same problem.
So I moved the stepper back to the good drive. I had a really hard time getting it to find track 0. eventually,.. by playing with the placement and mounting of the stepper, and the tension in the little steel belt, I got it to work.
Which leads me to conclude... there must be servo information marking track zero.
If there was servo information, the head mover would be voice coil based, not stepper based. Related information at [here].

When an MFM hard drive is low-level formatted, I expect that the cylinder number is written to each cylinder as part of the low-level format.
For example, section 5.7 of the OEM manual for the WD1002S-WX2 controller shows the breakdown of the low-level format used by that controller.
I can see that the cylinder number is within the IDENT and CYL LOW portions of the IDENT field.
 
track -1! exactly. I think it has to see a specially marked track on the disk that is one notch out from what is intended to be track 0.
 
Back
Top