IBM 5150 floppy controller no longer reads disks properly even with known good drive, cable and disk

[compaqportableplus]

So I've been having quite the time repairing another early Rev A 5150 I've recently got. Finally got all the motherboard issues sorted and it's running great now. First truly broken 5150 I've ever gotten and it feels kinda great to have it repaired now. And I could not have done it without the help of both this forum (reading lots of past 5150 repair threads) and minuszerodegrees.net! After the motherboard was working I went ahead and got the Tandon disk drive cleaned and relubricated and it too was working great along with the original floppy controller.

Yesterday I was using the machine to test out some memory cards I had just got (hunting down some bad chips on them) which required several power cycles and all seemed good until towards the end where I put my DOS boot floppy back in and it would no longer boot! Just dropped me straight into BASIC. Tried another disk and same thing. Pulled out another drive same thing. New ribbon and still same thing! Only when I swapped the controller did it finally boot once again from my DOS disk. Being an original earlier revision floppy controller, I refuse to simply swap it out, I must repair it, but I'm just kind of stumped as to what would cause this. The controller was working just fine before.

Here's a pic of the card. It's not the earliest one with the metal chips but a slightly later version.



It seems to still be able to step the heads fine and no 601 error comes up during post, so it's passing the basic tests at least. And the "data error" I'm getting in DOS versus "general failure" implies that it's reading some kind of data but it's getting corrupted within the controller somewhere. I do have a USB scope which isn't the greatest one but it works and was of much help diagnosing the motherboard in this machine. But due to the lack of information out there about repairing these I'm kind of at a loss of where to really even start honestly.

So again, the drive, cable and disk are all known good and working with another controller in the same machine, but when this card is installed no disks will boot and when booting from a hard drive and trying to get a directory of the floppy drive with this controller I get a "data error reading drive A" in DOS.

Any common things to check? Is the 8272 controller chip a potential failure point here? I did some quick probing around on the 8272 and compared it with the working spare controller and didn't notice anything strange. But like I said I just really don't know where to start! And I really hate to just "replace stuff and hope for the best" so I'd like to take a more logical approach to repair here just like I did with the motherboard.

Thanks!

 

[kevju]

Is that a broken trace above the ISA connector? How old are those tants? Can you use your scope to see what the power looks like, and also make sure that you have 16Mhz coming out of the oscillator?

 

[modem7]

I must repair it, but I'm just kind of stumped as to what would cause this. The controller was working just fine before.

Chips can fail in operation.

And the "data error" I'm getting in DOS versus "general failure" implies that it's reading some kind of data but it's getting corrupted within the controller somewhere.

That is what is sounds like. I am guessing that the origin is the following possible error that the Intel 8272 controller chip (uPD765 is the same) can report.

Is the 8272 controller chip a potential failure point here?

I cannot see why a partial failure is not possible.

I did some quick probing around on the 8272 and compared it with the working spare controller and didn't notice anything strange.

Refer to the 'read circuitry summary' diagram that I placed at [here].

Even though you are seeing activity on pin 23 of the 8272 (uPD765) during a read operation, maybe that activity is improper.
Maybe the data window is positioned incorrectly.

 

[compaqportableplus]

Is that a broken trace above the ISA connector? How old are those tants? Can you use your scope to see what the power looks like, and also make sure that you have 16Mhz coming out of the oscillator?

Good eye! Those traces were actually scratched pretty badly so I scraped some solder mask off and soldered over them just to make sure I had a solid connection. Made absolutely no difference.

Tantalum’s are original but are not shorted, certainly not the issue here.

Power is fine, and yes the clock frequency is present.

Chips can fail in operation.

Absolutely!

That is what is sounds like. I am guessing that the origin is the following possible error that the Intel 8272 controller chip (uPD765 is the same) can report.

Interesting.

I cannot see why a partial failure is not possible.

I’ll definitely keep that in mind.

Refer to the 'read circuitry summary' diagram that I placed at [here].

Even though you are seeing activity on pin 23 of the 8272 (uPD765) during a read operation, maybe that activity is improper.
Maybe the data window is positioned incorrectly.

Excellent! I’ll have a look at that.

How would I determine if the activity was improper? Is that something that would be noticeable on the scope? I wasn’t paying too much attention to exactly what the activity looked like in comparison with the other controller, just that there was activity at all and there definitely was. But I can definitely check again if need be.

 

[xp0nentz]

Most likely, it's a partial failure of the 8272. I don't know though, so I'll dive into the schematics of the 5150 and research that.

 

[xp0nentz]

Looked at the IBM technical document for the card, even though the card was a later version. If most of it is still the same except for the controller IC, then I think the issue might be around here?

 

[compaqportableplus]

Most likely, it's a partial failure of the 8272. I don't know though, so I'll dive into the schematics of the 5150 and research that.

I am becoming highly tempted to unsolder the 8272 and add a socket at this point! I might have a board laying around I can pull one off of or I can just order a new one. I have a desoldering station so it’s not gonna be hard to do anyway.

Looked at the IBM technical document for the card, even though the card was a later version. If most of it is still the same except for the controller IC, then I think the issue might be around here?

Very interesting! This gives me some more stuff to look at, thanks! I believe the NEC chip (forget the model) and the Intel 8272 are functionally equivalent since I’ve seen both used on these floppy controllers (both early and later versions).

 

[xp0nentz]

I am becoming highly tempted to unsolder the 8272 and add a socket at this point! I might have a board laying around I can pull one off of or I can just order a new one. I have a desoldering station so it’s not gonna be hard to do anyway.

Very interesting! This gives me some more stuff to look at, thanks! I believe the NEC chip (forget the model) and the Intel 8272 are functionally equivalent since I’ve seen both used on these floppy controllers (both early and later versions).

Sockets are very useful. If you're ready to add a socket, do it. But be careful to not damage anything.

 

[compaqportableplus]

Sockets are very useful. If you're ready to add a socket, do it. But be careful to not damage anything.

Oh yeah, I definitely won’t damage anything, but that is a real concern because I’ve seen some people mangle these IBM boards when trying to remove components.

I already added five sockets successfully with no damage while I was diagnosing and repairing the motherboard in this 5150. Turns out only two were actually needed but oh well. I’ll do a full writeup about this 5150 and my repair adventures with it soon.

Preheating is the trick with the IBM boards. Trying to desolder components from a cold board will result in frustration and eventually damage because the thick internal ground plane soaks up all the heat from your iron. But if you get the general area of the board you are working on nice and warm first you get a good, clean removal.

 

[modem7]

Even though you are seeing activity on pin 23 of the 8272 (uPD765) during a read operation, maybe that activity is improper.
Maybe the data window is positioned incorrectly.

How would I determine if the activity was improper? Is that something that would be noticeable on the scope? I wasn’t paying too much attention to exactly what the activity looked like in comparison with the other controller, just that there was activity at all and there definitely was. But I can definitely check again if need be.

The problem is the complexity. What is shown at [here] is from the data sheet for the uPD765. It shows the low-level format laid down by the uPD765 (and equivalents).

 

[xp0nentz]

Oh yeah, I definitely won’t damage anything, but that is a real concern because I’ve seen some people mangle these IBM boards when trying to remove components.

I already added five sockets successfully with no damage while I was diagnosing and repairing the motherboard in this 5150. Turns out only two were actually needed but oh well. I’ll do a full writeup about this 5150 and my repair adventures with it soon.

Preheating is the trick with the IBM boards. Trying to desolder components from a cold board will result in frustration and eventually damage because the thick internal ground plane soaks up all the heat from your iron. But if you get the general area of the board you are working on nice and warm first you get a good, clean removal.

If you replace the controller and it's still giving issues, it's probably something to do with the data separation I'd assume.

I agree with Modem7, the problem is complexity.

 

[compaqportableplus]

If you replace the controller and it's still giving issues, it's probably something to do with the data separation I'd assume.

I agree with Modem7, the problem is complexity.

Thank you and @modem7 for your insights! I’m going to replace the controller chip as a starting point and I’ll report back with the results!

 

[compaqportableplus]

So, very interesting development, my spare floppy controller failed in the EXACT same way the original did while playing some more with this machine! After maybe a dozen or two power cycles here we are again with the same failure with a different controller. Tried the controller in another 5150 and same issue, so it's now dead too. This is not a coincidence, no way, as I've never had one of these controllers fail before and now within a week two fail in the same way on the same machine?

So now I was suspecting the aftermarket power supply strongly. After all, the motherboard did have two dead chips as well, so I was a little skeptical of the supply from the beginning. The 5V rail does seem a tiny bit hot at 5.2V but I didn't really think that was an issue, but I know PSUs can have other issues a multi meter won't show.

I put in a different known-good PSU and (reluctantly) grabbed another floppy controller from another 5150. Was another later silver-bracket controller though so pretty common/cheap card and not a huge risk. I figured if something was killing the floppy controllers it was unlikely to be the motherboard (considering it has no shorts and works fine) and far more likely the PSU. Well with the known-good PSU and FDD I did dozens of power cycles and no issues! Controller is still working. So I think, hopefully, I've narrowed this down to a bad PSU. And now I have two floppy controllers to repair. This does give me more hope that maybe the 8272 is probably the issue, as I have a feeling it will be more sensitive to voltage issues than the 74-series logic chips. I kinda just wish I had binned this PSU from the beginning now, dang. Shame too because it seems like a nice unit but if it does really turn out to be the issue I don't think I could ever trust it again.

Guess I better order TWO floppy controller chips now!

 

[xp0nentz]

So, very interesting development, my spare floppy controller failed in the EXACT same way the original did while playing some more with this machine! After maybe a dozen or two power cycles here we are again with the same failure with a different controller. Tried the controller in another 5150 and same issue, so it's now dead too. This is not a coincidence, no way, as I've never had one of these controllers fail before and now within a week two fail in the same way on the same machine?

So now I was suspecting the aftermarket power supply strongly. After all, the motherboard did have two dead chips as well, so I was a little skeptical of the supply from the beginning. The 5V rail does seem a tiny bit hot at 5.2V but I didn't really think that was an issue, but I know PSUs can have other issues a multi meter won't show.

I put in a different known-good PSU and (reluctantly) grabbed another floppy controller from another 5150. Was another later silver-bracket controller though so pretty common/cheap card and not a huge risk. I figured if something was killing the floppy controllers it was unlikely to be the motherboard (considering it has no shorts and works fine) and far more likely the PSU. Well with the known-good PSU and FDD I did dozens of power cycles and no issues! Controller is still working. So I think, hopefully, I've narrowed this down to a bad PSU. And now I have two floppy controllers to repair. This does give me more hope that maybe the 8272 is probably the issue, as I have a feeling it will be more sensitive to voltage issues than the 74-series logic chips. I kinda just wish I had binned this PSU from the beginning now, dang. Shame too because it seems like a nice unit but if it does really turn out to be the issue I don't think I could ever trust it again.

Guess I better order TWO floppy controller chips now!

Ok, so we probably have the cause down! Replace the controller ICs and it should work again! (hopefully)

 

[compaqportableplus]

Ok, so we probably have the cause down! Replace the controller ICs and it should work again! (hopefully)

That’s what I’m hoping! Will report back with results once I replace them.

 

[xp0nentz]

That’s what I’m hoping! Will report back with results once I replace them.

Alrighty.

 

[Istarian]

Supposing that the controller is determined to be the problem, it's always possible that the chip was already well on it's way to failing.

This unfortunate situation might simply have been the last straw.

We are talking about hardware from the 1980s here and it's four decades (40 years) into the future for anything manufactured between 1980 and 1986.

Depending on what kind of "scope" you have it might be possible to use it to inspect the output of your power supply.

Not that I would ever use the sketchy PSU for anything important unless/until you can verify that it's working properly.

 

[compaqportableplus]

Supposing that the controller is determined to be the problem, it's always possible that the chip was already well on it's way to failing.

This unfortunate situation might simply have been the last straw.

We are talking about hardware from the 1980s here and it's four decades (40 years) into the future for anything manufactured between 1980 and 1986.

Depending on what kind of "scope" you have it might be possible to use it to inspect the output of your power supply.

Not that I would ever use the sketchy PSU for anything important unless/until you can verify that it's working properly.

Certainly possible, but two different cards has me very skeptical. And these are not particularly unreliable cards/chips either.

Obviously anything can fail at any time for any reason, but something’s definitely up here and the power supply is the biggest suspect at this point.

I have a USB scope that isn’t the greatest but it works. I’m sure I could do some tests on the PSU with it but I’m more concerned with getting these floppy controllers repaired first.

And yeah I just don’t think I could ever trust that PSU now, if that does in fact turn out to be the issue. Will probably just gut it and install a new power supply board into the case.

 

[Istarian]

You could also use the original PSU board as a practice board for power supply repair if you can identify the probable fault/issue. Consider it an opportunity to practice using the scope to observe the behavior of power supplies.

 

[compaqportableplus]

You could also use the original PSU board as a practice board for power supply repair if you can identify the probable fault/issue. Consider it an opportunity to practice using the scope to observe the behavior of power supplies.

That’s a great idea actually!