Print

[stepleton]

Hi folks,

A party over on another forum (although not a stranger to us) has a sick Widget with limited motion of the heads. On power-up it attempts to move the heads slightly but fails to achieve full travel. The power-up routines therefore fail and the Widget just spins its big wheel helplessly.

The Widget owner has replaced most of the electronics, so the problem is likely mechanical. I suspect that the arm mechanics have got some sticky lubricant and that some more attempts to get the arm to move (e.g. by resetting the drive repeatedly) might free up its travel. I've seen something like this work before on Widgets of my own. But either this Widget is stickier than most, or something else is wrong.

What would you do next? (Assume that you don't have a UsbWidEx.)

• It's hard to hear anything over the spinning platters, so if there's something else in the way of the arm (e.g. a dislodged coarse alignment graticule), the owner might not be able to hear anything knocking against it.
• Suppose the owner pinched and gently twisted the arm servo shaft while the disk was turning --- this to overcome additional stickiness. We all know that you are never supposed to touch a ProFile stepper while it's moving, but could this be slightly less damaging for the Widget?
• Is there any way to drip fresh lubricant into the mechanism somehow? There's an Adrian's Digital Basement (YouTube) episode where Adrian revives an old PC drive this way, but I think this drive had a stepper mechanism.

Anything else to check or do?

[AlexTheCat123]

I had a similar issue with sticky heads on one of my Widgets and I started by trying to reset the drive repeatedly just like what you mentioned. But unfortunately that didn't work and I ended up going with the "move the shaft manually" idea. After a few twists (with the disk spun up of course), things seemed to become a lot less sticky and the drive came to life the next time I powered it up! Moving the head servo didn't seem to cause any damage whatsoever, so maybe give that a try if there aren't any other ideas. I'm not a hard drive expert or anything, but I can't think of any reason why doing this would cause any damage. As long as the heads are floating, you shouldn't be able to do any damage to the platters. The worst case that I can think of is some data potentially getting overwritten if the heads are trying to perform a write as you move them. But that's easily remedied by doing a LLF and/or reinstalling an OS. Once again, I'm not a hard drive expert, so I could be wrong about something, but this process seemed to work just fine for me!

[ried]

Thank you, stepleton for bringing this to the community. Thanks to AlexTheCat123 for the quick reply, too. Indeed, this Widget behaves as described, and I'm not sure what to try next. I've changed out the Controller, Read/Write and Servo boards as well as the motherboard - although I have since read that motherboards are specifically tuned for each drive and that originals should be used, if possible, so did put the original back on. There was no change to the drive's behavior with any of these swaps, so inclined to believe it's a mechanical issue as stepleton described.

Here are two videos recorded this morning. Be warned, the bearings are pretty loud and thus the videos are also.

Full sequence (long and boring):
https://www.youtube.com/watch?v=zG82oF_TZfw

Close up (short and to the point):
https://www.youtube.com/watch?v=pyaMPeMUrAs

As you can see, it attempts to move the heads every second or so, and its motion is extremely limited. When it reaches the extent of its motion (before falling back to its baseline), the green LED often gives a bright blink. Usually, this happens 1-3 times in a row, and then it'll stay off while subsequent attempts continue, then suddenly repeat the pattern. Occasionally, the LED will blink every second for 20, 30 or more seconds. But the motion remains limited.

I'm using an ArcaneByte drive emulator (based on Cameo/Aphid) attached to a 2-port parallel expansion card to use BLU 0.9. As you can see, the drive never gets to a state where it can be responsive to BLU or NeoWideEx to i.e. reset the servo for performa low-level format.

What to do, friends? I'd like to save this old drive, if possible. Thank you.

[compu_85]

This is what my widget did initially. My suggestions:

With the drive spinning, manually move the head arm stop to stop a bunch of times. (Yes, grab it with your fingers, and rotate it) The voice coil has very little torque. If it gets gummy at all, it won't work. Doing this can get the bearing "moving again".

The 3 adjustment pots are *very* on the motherobard sensitive. On my list of things to do is to do a proper factory set up of these. I've not done one yet, I've just tuned them by ear on my drives. The movements your is making look just like what my drives did when those adjustments were off. I would mark each one with a pen, then slowly adjust them a little bit until the drive began sounding more normal. This is harder to do with loud spindle bearings of course. But as you get close the drive's sound will change, and the head will start moving differently.

If you haven't heard this already, this is what my best running widget sounds like: https://www.youtube.com/watch?v=YwHR8Tflvuo

-J

[ried]

With the drive spinning, manually move the head arm stop to stop a bunch of times. (Yes, grab it with your fingers, and rotate it) The voice coil has very little torque. If it gets gummy at all, it won't work. Doing this can get the bearing "moving again".

Okay, just did this. There is no physical resistance to full stop-to-stop movement that I can feel. None whatsoever. I can also feel the servo motor moving the arm at the beginning of its sweep, but as you describe it has very little torque and doesn't seem interested in moving the head much (if at all) past that first 25-50% of its range. If I hold the head at 80-100% of its range, for example, it doesn't feel like the motor is doing much of anything in that area. It is certainly active across the first ~50% of it's range, but drops off quickly after that.

The 3 adjustment pots are *very* on the motherobard sensitive. On my list of things to do is to do a proper factory set up of these. I've not done one yet, I've just tuned them by ear on my drives.

I need to look into this. Sounds like that is the next step. Is there a guide to tuning these? Better yet, what should I specifically avoid doing, if anything?

If you haven't heard this already, this is what my best running widget sounds like: https://www.youtube.com/watch?v=YwHR8Tflvuo

That sounds great! Thanks for the good example. Hopefully we can save this one, too.

Edit: Found this very helpful post:
https://lisalist2.com/index.php/topic,15.msg37.html#msg37

However, working with an oscilloscope is beyond my hardware and skillset at this point. Hmmm...

[compu_85]

I'd say, as you adjust the pots, avoid leaving them at an extreme. If turning it one direction doesn't "improve" the situation, stop. Also note they do interact with each other.

-J

[ried]

Understood. Okay to start with Tacho Offset, LED current, Tacho gain? Seems a lot easier than disassembling the board and adjusting A', B', B, A amplitude.

[compu_85]

I think I left the LED one mostly alone. If you get a long, thin screwdriver, and remove the shield piece from the back of the motherboard, you can adjust them with everything hooked up.

[stepleton]

I need to look into this. Sounds like that is the next step. Is there a guide to tuning these? Better yet, what should I specifically avoid doing, if anything?

<...?

Edit: Found this very helpful post:
https://lisalist2.com/index.php/topic,15.msg37.html#msg37

However, working with an oscilloscope is beyond my hardware and skillset at this point. Hmmm...

I think the calibration method is described in the Widget ERS document on Bitsavers. This is basically the Bible of Widget, and (unfortunately) an oscilloscope with X-Y capability is required. Look for key waveforms near the back.

[ried]

Well, this is not something I was expecting to do:
https://imgur.com/W6nsTCK


But adjusting B' slightly did result in this additional movement:
https://www.youtube.com/watch?v=g4196_uTkl8

Update: Nice to see the internal disk appear as a startup option:
https://imgur.com/a/zNYmuZB

Not working yet, of course. I suspect I don't have the knowledge or tools to fine-tune this enough to actually make it usable. But this is progress, it seems!

Update 2: The drive is seen by BLU 0.9 now, and responds to the Exercise / Test Disk function, for example. It goes through the physical motions of the exercise, and eventually generates an Error E803:
https://imgur.com/a/1AAhphs

[sigma7]

Update 2: The drive is seen by BLU 0.9 now, and responds to the Exercise / Test Disk function, for example. It goes through the physical motions of the exercise, and eventually generates an Error E803

In BLU 0.90, Error E803 is reported when a Widget's command response is abnormal -- something other than (theCommand+2).

Now that the Widget code is available one might be able to figure out what causes that. In this case I'd guess it is "sorry, I can't do that Dave"

edit: BTW, since this forum isn't designed to generate revenue, there is no problem mentioning or linking to other relevant forums.

[compu_85]

You need to adjust the pots a little bit more. You're getting closer.

[ried]

Well, after adjusting the trimpots with both the original motherboard and a replacement, I have restored full movement to the head servo.

https://www.youtube.com/watch?v=R3ktz9bICaE

That said, there seem to be several other trimpot adjustments that I likely threw out of spec during the effort, so it's still generating an Error 81 upon boot. I've tried both exercising the disk and a low-level format through BLU and NeoWideEx, adjusting the trimpots further, then trying again. Rinse and repeat. Those attempts have all failed with the disk remaining unresponsive.

It seems that any further adjustments will need to be made by someone with an oscilloscope (and the skills to use it) in order to bring all of the variables into spec. Even then, a working Widget is not guaranteed. I was hoping that my own manual adjustment would be enough, but alas I am not the Widget Whisperer... this time.

Thanks for all the kind advice, everyone. Greatly appreciated.