Deunan wrote on 2025-06-24, 21:09:

Huh. Usually the screws are from the top, 3 of them. What you have would put the stress of the clamping mechanism on the PCB.
Well, try to undo them and lift the PCB, at least rotate it to make a photo of the components side. Try to wiggle it a bit, might be properly stuck after all these years - the fit is usually pretty good but it should not be so tight as to require any force. So if you can't seem to get it off, put the screws back and try to get a better look from the top. Could be there's screws there too.

EDIT: Do not touch these 3 screws in the rotor. Those are quite likely glued and can be left-threaded. There's no need to mess with the rotor.

Odd. I replied to this message last night and it is gone now.... anyhow, I undid the 4 screws. The PCB was able to rotate slightly, but it didn't want to pull off. At least not easily. So I put the screws back.

To look at the other side will, I think, involve a lot of disassembly?

That upper metal plate has to go to get a good look (and screw access, if required). It's hard to tell from just one angle but this is what it should take:
- The closing lever/arm needs to be removed, by pulling it to front. It might be quite stuck but it should only take fingers to pull it. Do not use any tools, if you can't remove it there might be another way, not risking any damage to plastic parts.
- Undo two screws that hold the front panel to the case, with those removed (and lever gone) you should be able to tilt it to front and unhook from the bottom.
- Remove two (?) screws that hold the head shield and remove that as well.
- Another 2 screws near the front, on the sides, that hold the big metal mounting plate to the case. The right one is simple, the left also holds down the WP sensor.

That should get the plate off. Might be another screw somewhere closer to the heads. Do not remove the whole clamping rod assembly, that should only connect to the plate. There might be a screw or some sort of a notch under the PCB, if so you'll need to unbolt that as well. Not all the plugs have to be disconnected, usually if you unplug one side the PCB can be just flipped to the side and out of the way.

It might seem like a lot of parts need to go but it's all very simple to put back together. The only things you don't want to touch is stepper motor and head assembly. And track 0 sensor. The rest is like LEGO blocks, just pay attention to the screws as these might be different length or thread. When putting it back together route the wires as they were before, to avoid anything getting pinched. Take more photos if required. I think I see one of the 3 screws I mentioned, around the spindle, on one of the previous photos with the heads. So those are probably there after all.

Deunan wrote on 2025-06-25, 09:53:

- The closing lever/arm needs to be removed, by pulling it to front. It might be quite stuck but it should only take fingers to pull it. Do not use any tools, if you can't remove it there might be another way, not risking any damage to plastic parts.

I seem to have failed at the first task. Haha. I can't get that lever off. I pulled quite hard. Just so I understand you, the metal rod that the plastic lever is attached to, should stay and I am trying to pull that lever off the rod?

Yup, the rod stays and the lever should come off. But it can be a very tight fit - as not to come off by itself when the lever is moved. Please note the plastic piece with a spring is supposed to prevent people closing the drive without a floppy inserted, which not great for the heads but will not cause any damage if there are no vibrations, like during transport for example. It can be bypassed by pulling that little plastic bit (that spring connects to) out of the way to let the rod turn. Some people force this and damage this mechanism permanently. I'm mentioning that because while trying to get the lever off you might also damage the plastic bit (if it wasn't already, which is often the case with used drives).

Maybe you can just unbolt the front and it will come off with the whole metal plate. Perhaps there will be enough wiggle room to unhook the bottom of the front cover. Another way is perhaps to remove the metal clips that sit on the rod and prevent it from moving in/out. You want to move it out at least enough so that the wide plastic part of the lever is not preventing you from tilting and removing the front cover. There is one I can see (near the plastic bit), might be another one further back. Usually there isn't another near the front, as that would be difficult to get to. Consider getting these clips off with a small flat screwdriver or something of the sort to free the rod to move out a bit.

I haven't worked on one of these Panasonics for quite a while, but I remember also not being able to remove the lever. Was even wondering if it's glued or somehow molded in such a way that it can't be removed. I can't remember what I did exactly bit I somehow worked around that, not wanting to damage the plastic lever. Some very old drives might have a screw that hold the lever in place but in that case the front of it has a small plastic plug to hide the screw. Yours should be one plastic piece, so no way for there to be any screws.

EDIT: Yes, it seems the whole thing can be lifted, see here: https://youtu.be/velEVw_Kdvg?t=160
You can also see the 3 screws that you'll need to undo to remove the spindle motor (and the 4 ones on the bottom most likely as well).

Deunan wrote on 2025-06-26, 21:31:

Maybe you can just unbolt the front and it will come off with the whole metal plate. Perhaps there will be enough wiggle room to unhook the bottom of the front cover.

Well spotted on that video! That was very helpful.

Anyhow, I got the top off whilst leaving the head in place. I had to remove two screws holding the plastic front on, two screws holding the top PCB on, and four screws holding the top lever mechanism in place (one of which was under that top PCB).

I was then able to get to the 3 screws in order to take off lower PCB:

I can't see any leakage or bulging. Can you?

By the way, I bought a other drive on ebay. Different model. Was spares repairs and, unsurprisingly, it has problems. Probably should create another thread for that one. But first will apply what I have learned from looking at this drive to see if I can fix it. Anyhow, the reason for me mentioning it, is that I tried my scope on that drive and it says its running at 6Hz. Which is 360RPM. Thus proving that my scope is measuring properly that this Panasonic drive is running much too fast.

By the way, is the disk hole sensor used for controlling the speed? If so, could there be an issue with that?

No, that would make the motor assembly unable to control the speed if there is no floppy in drive, or if the floppy is hard-sectored for example. All this PCBs has is 2 LEDs, the sensors are on the top and only connected to the upper (main) PCB. For speed control there is one or more sensors, this chip uses 3 hall effect sensors to control the torque and current and it also uses another magnetic sensor for the rotation speed.

So this kind of motor controller uses ceramic resonator to set the frequency. It's pretty accurate and doesn't require any adjustments, so there are none. But these resonators go bad, I've seen that myself. More importantly there's a couple of electrolytic capacitors here, some of which are in important places, like noise suppression from the coils and the speed sensor DC blocking.

Looking at datasheet, there should be a 10uF capacitor that's quite critical. If it dried out it might just caused this problem you have. I bet it's one of the 2 small ones in the part of the PCB with the chip. The 3 bigger ones there are (or should be) non-polarised caps, more difficult to source, but also not as critical. The polarised ones I would replace all, but start with the 10uF one. If you have any ESR/capacitance meter, it should work in-circuit on this cap, seeing how it's in series with a magnetic sensor that's basically a small choke.

BTW this whole motor assembly can be run from external PSU, on the bench. The 5V should not be even needed, as far as I can see it only powers both LEDs and that's it. So you need 12V, GND, motor on signal (MS I bet) and you should be able to switch between 300/360 rpm via S/S. These control signals require low level (so a connection to GND) to activate.

RetroPCCupboard wrote on 2025-06-27, 04:54:

... this Panasonic drive is running much too fast.

Strange, I don't think I've seen this before!

It has been years since I looked at these kinds of things...

Some drives have a little potentiometer on the drive motor PCB to adjust the speed, but I can't see it drifting (or being adjustable)
by this much.

Some drives do support multiple speeds, In the "Software/Images" page of my site I have a link to a page describing a simple
modification of a Panasonic drive to make the speed switchable between 300/360 rpm. I never dug deeper to see how the speed
control actually regulates itself (ie: if it could be a simple component failure) but it might be worth making this modification to
you drive and see of both speeds are "off", and if so by the same percentage - might give a clue as to where to look further.

Deunan wrote on 2025-06-27, 08:55:

If you have any ESR/capacitance meter, it should work in-circuit on this cap, seeing how it's in series with a magnetic sensor that's basically a small choke.

BTW this whole motor assembly can be run from external PSU, on the bench. The 5V should not be even needed, as far as I can see it only powers both LEDs and that's it. So you need 12V, GND, motor on signal (MS I bet) and you should be able to switch between 300/360 rpm via S/S. These control signals require low level (so a connection to GND) to activate.

Many thanks.

I have ordered a Capacitance meter. I will test it on the board before I remove any components. I do have soldering tools, but I am very out of practice.

Going all-in I see. That's the best way to learn, and have some fun while at it. Before your new toy(s) arrive you can check the resonator frequency with your scope. It's that small blue box with 2 leads. You should get 492kHz out of it, try both leads because the chip has one input and one output for the resonator amp. The output will give you stronger and more stable signal level. For this test you only need +12V and GND connected, so you can do it even with the motor PCB removed from the floppy drive completly, and powered by external PSU. That being said if the cables allow it, it would probably be easier to use the PC PSU to power the whole thing.

Soldering these PCBs is not difficult provided you have a temperature controlled soldering iron. If not you could damage the traces with too much heat applied, these single-side PCBs are cheap to make but not great to work with unless you have some experience. I can perhaps provide some tips if you need any.

EDIT: I'm pretty sure the cap that you want to check first is marked C11. It's not 10uF but 1uF, which makes it even more suspect as that is well below recommended value, so any further loss of capacitance (and ESR rise) would affect the circuit even more.

Deunan wrote on 2025-06-27, 23:08:

Going all-in I see. That's the best way to learn, and have some fun while at it. Before your new toy(s) arrive you can check the resonator frequency with your scope. It's that small blue box with 2 leads. You should get 492kHz out of it, try both leads because the chip has one input and one output for the resonator amp. The output will give you stronger and more stable signal level. For this test you only need +12V and GND connected, so you can do it even with the motor PCB removed from the floppy drive completly, and powered by external PSU. That being said if the cables allow it, it would probably be easier to use the PC PSU to power the whole thing.

I screwed the top PCB and metal parts back on before putting it on my workbench like this.

I connected the PSU and data cables up and turned it on. The head moves back and forth as expected. I do not get anything from the blue square chip until the motor turns on (by putting a disk in the drive). I get signal from both legs. The scope reads 820Khz on one leg and can't seem to lock onto the frequency of the other. The one it's locking onto though, seems to have a lower voltage, so not sure if that's the correct lead. Neither looks like a square wave, but that is probably a limitation of this scope.... I will upload pictures from the scope shortly when I get back to my PC

Here's screenshots from the scope on each leg of the blue square component whilst motor is running:

Lack of square wave here is not an issue, it never is square on analog parts of an oscillator circuit, should actually be more sinusoidal. The chip will make square out of it internally. The second image is already half-way there, that's probably the output, could be also affected by the scope and probe limitations, as well as your choice of grounding point for the probe. That being said these resonators are more noisy in their output than crystal ones. But this is good enough, if a bit low on the amplitude, but that might just be how it works - hard to tell.

The frequency is a problem though. Quick calculation and 820 (I got 833 from counting the divisions, close enough) by 492 is 1,6(6). This multiplied by 300 gives 500, which is pretty much what you found testing the rpm via INDEX signal. So the resonator is running way too fast, at least according to the datasheet for this chip. I suppose though if this drive used a different configuration of the motor coils it might need a different frequency. I think I see a number "819" on the resonator, wonder if that has anything to do with the frequency. Usually these parts have unusual markings that do not clearly state the frequency so I'm not convinced. This might be a problem, or just so happens that we ended up with these numbers.

I need to refresh my own memory on these resonator specs, because different servo chips can require different frequencies. Some newer 3.5" drives even use a clock signal provided from the main board. In the meantime can you try to figure out the color bands on C2 and C4? These look like green resistors but should be stable ceramic capacitors (in somewhat unusual packages). Or make a close-up photo. The light reflections make it hard for me to read the values, but these sure aren't 47pF, which would indicate the values were also adapted to the different resonator frequency.

EDIT: Actually a frequency of 819200Hz would make sense for this chip as well, considering the dividers it is using. 819200/(8*1024)=100. Which then works nice with x5 and x6 multipliers to get the rotation speed. It just needs a different motor than what the datasheet wants. And the signal level is more or less in the range too, input is higher than 2mVpp.

If you are up to some more scope measurements, probe pins 1, 2 and 3 of the servo chip. Might tell us more about that suspect capacitor.

Just some anecdotal reference: It is well-known that early PS/2 3.5" floppy drives are prone to failure due to degraded SMD capacitors. I took a deep analysis on one of them, and I found that the AC coupling cap in the rotation speed feedback signal failed, so the speed regulating circuit did not observe any rotation at all, and just set the motor driver to "full power". The resulting speeds were 500 to 600 RPM instead of 300 RPM, which looks very similar to the symptom in this thread.

The fault I had is called "open loop", because the feedback loop from motor power to the speed feedback was interrupted. The tell-tale sign of an open-loop fault is that the actual speed is way above the target speed (it is in your case), and also that the actual speed highly depends on load. Different floppy disks with different amount of internal friction will reach different motor speeds at "full power". So another easy measurement you could do to confirm whether you might have an open-loop fault is to check the index signal with a scope for different floppy disks. If each disk turns at a different speed, it is obvious that the regulation circuit is unable to regulate. This need not be an open-loop fault, though, but an inappropriately high target speed will cause the same symptoms (i.e. if the reference osciallator started to oscillate on the 3rd harmonic instead of the fundamental frequency. The target would be 1080 RPM in that case).

A final hint: While I did not read the thread in detail, I observed references to "speed switching" between 300RPM for DD disks and 360RPM for HD disk. Many 5.25" drives are able to do that, and some computers required the drives to work that way, but IBM-compatible computers require 360RPM all the time on 1.2M drives, no matter what kind of medium is inserted. If a HD 5.25" drives rotates at 300RPM, it is clearly configured (jumpered) in a way that is unsuitable for use in IBM-compatible computers.

mkarcher wrote on 2025-06-28, 15:33:

... IBM-compatible computers require 360RPM all the time on 1.2M drives, no matter what kind of medium is inserted ...

Almost exactly true - IBM AT+ supports 300kbps data rate which allows "300rpm disk" written at the standard 250kbps (DD drive)
to be read/written with the same bit density at 360rpm (HD drive).

Where it "falls short", and the reason I have a note on "Daves Old Computers - Software/Images" about modifying a drive to be
switchable between 360<>300 rpm - is that certain non-PC formats which CAN be read with the PC FDC hardware contain single
density tracks and CAN'T be read at 360rpm.

This is just a "technical note" and won't be a problem for anyone not planning to use something like ImageDisk to read/write non-
PC format disks. If you are only using floppy disks created on/for a PC, keep it at 360rpm.

DaveDDS wrote on 2025-06-28, 17:00:

mkarcher wrote on 2025-06-28, 15:33:

... IBM-compatible computers require 360RPM all the time on 1.2M drives, no matter what kind of medium is inserted ...

Almost exactly true

Yeah, I possibly should have written that the BIOS of IBM-compatible computers and nearly(?) all operating systems for IBM-compatible computers that contain their own floppy drivers requires this. Custom software like IMD specifically made to read/write foreign disks may of course relax that requirement.

DaveDDS wrote on 2025-06-28, 17:00:

is that certain non-PC formats which CAN be read with the PC FDC hardware contain single density tracks and CAN'T be read at 360rpm.

Do you happen to know why this is the case? The 300 kbps mode clocks the FDC in a way that it samples the data line at a rate suitable for disks written at 125kbps FM / 250kbps MFM in a 300RPM drive. Is the issue related to signal conditioning filters on the drive or the FDC that can't cope with single-density signals at 150kbps FM, but can do so at 125kbps FM?

mkarcher wrote on 2025-06-28, 19:17:

Do you happen to know why this is the case? The 300 kbps mode clocks the FDC in a way that it samples the data line at a rate suitable for disks written at 125kbps FM / 250kbps MFM in a 300RPM drive. Is the issue related to signal conditioning filters on the drive or the FDC that can't cope with single-density signals at 150kbps FM, but can do so at 125kbps FM?

Yeah, I too should have been a bit clearer (trying to be brief and not fully trusting 10+ year old memories)

I worked with and tested A LOT of PC compatibles during the course of developing ImageDisk.

And just to be clear, ImageDisk talks directly to the FDC hardware, it has nothing to do with BIOS (in fact, on my main
disk imaging system I have B: (a "loose" connector I connect to appropriate drive types - incl 8"!) set to type=None
in BIOS!

And as I recall, I did encounter a few PC compatibles that could read single-density on an HD drive - but MANY of them
couldn't.

What I eventually decided was that these PC's used less capable data-separators which couldn't handle a
bit-stream of 150kbps. But ... more (but not all) of then could handle a bit-stream of 125kbps.

Kinda made sense (at least to me at the time) - the PC was never intended to use single-density (125kbps) and those
that could just happened to use an off-the-shelf data-separator designed long enough pre. that it could do it.
On the other hand as far as I know there were never any actual/common drives that were used for SD at 360rpm
and very few data-separator designs ever needed to do 150kbps so they didn't!

Deunan wrote on 2025-06-28, 10:30:

I think I see a number "819" on the resonator, wonder if that has anything to do with the frequency. Usually these parts have unusual markings that do not clearly state the frequency so I'm not convinced. This might be a problem, or just so happens that we ended up with these numbers.

Sorry for slow reply. Had some family stuff, which meant I didn't have time to look at this earlier.

You are right, it says 819. Though googling doesn't seem to find a match for the markings on it:

Deunan wrote on 2025-06-28, 10:30:

I need to refresh my own memory on these resonator specs, because different servo chips can require different frequencies. Some newer 3.5" drives even use a clock signal provided from the main board. In the meantime can you try to figure out the color bands on C2 and C4? These look like green resistors but should be stable ceramic capacitors (in somewhat unusual packages). Or make a close-up photo. The light reflections make it hard for me to read the values, but these sure aren't 47pF, which would indicate the values were also adapted to the different resonator frequency.

I am not sure how to read thw bands. Though they seem identical:

Deunan wrote on 2025-06-28, 10:30:

EDIT: Actually a frequency of 819200Hz would make sense for this chip as well, considering the dividers it is using. 819200/(8*1024)=100. Which then works nice with x5 and x6 multipliers to get the rotation speed. It just needs a different motor than what the datasheet wants. And the signal level is more or less in the range too, input is higher than 2mVpp.

If you are up to some more scope measurements, probe pins 1, 2 and 3 of the servo chip. Might tell us more about that suspect capacitor.

OK. I will have a look at that.

I tried to measure C11 with my new multi-tester. It thinks it is a diode. Though, it does mention a capacitance:

RetroPCCupboard wrote on 2025-07-03, 12:13:

I am not sure how to read thw bands.

So these are 47pF after all. I was not sure if there was a band or not based on the previous photo but this one made it clear: yellow (4), violet (7), black (0) = 47*10^0 = 47pF. The rest is tolerance and voltage rating and such, nothing we care about here.

RetroPCCupboard wrote on 2025-07-03, 12:13:

I tried to measure C11 with my new multi-tester. It thinks it is a diode. Though, it does mention a capacitance:

I really hoped the inductance of the sensor would prevent any bogus readings here but I guess there is a protection diode inside the chip, to limit the pulses, and the forward drop is probably measured using DC. Or the cap itself could be so unformed by all these years of not seeing a DC bias that it itself turned into a diode of sorts. Aluminum oxide in these is a good insulator only one way (unless it's non-polarized cap). This one shows about 4V forward drop, a bit less than usual (about 6 volts) but not impossible I guess. The 100uF is obviously not possible unless the cap has such high ESR that it confuses the meter.

Tell you what, if soldering is not something you are comfortable with then desoldering stuff, without causing PCB damage, is probably even worse prospect. So rather then removing C11 it might be easier to just tack another 1uF capacitor to it, on the bottom of the PCB. Lead length is not cricital, you can keep them longer but same size to make the soldering easier. This would not be permanent, only for testing. The idea being if the cap has dried out and lost capacitance then adding another one in parallel will fix the issue, and should provide the chip with clean pulses from the speed sensor. And it's easier to do that than removing the old cap.

Try to get the scope measurements of X1 and the pins I mentioned first, before any soldering.

Deunan wrote on 2025-07-03, 13:05:

RetroPCCupboard wrote on 2025-07-03, 12:13:

I am not sure how to read thw bands.

Tell you what, if soldering is not something you are comfortable with then desoldering stuff, without causing PCB damage, is probably even worse prospect. So rather then removing C11 it might be easier to just tack another 1uF capacitor to it, on the bottom of the PCB. Lead length is not cricital, you can keep them longer but same size to make the soldering easier. This would not be permanent, only for testing. The idea being if the cap has dried out and lost capacitance then adding another one in parallel will fix the issue, and should provide the chip with clean pulses from the speed sensor. And it's easier to do that than removing the old cap.

Try to get the scope measurements of X1 and the pins I mentioned first, before any soldering.

OK. I will get my scope out.

I think removing the caps though shouldn't be a problem. I have one of those desoldering guns that sucks air through a tube (the pump is in the box that sits on the workbench). I bought it for recapping motherboards. I haven't tried recapping yet, but I have practiced removing components on some dead electronics. I think I got the hang of it.