curiousmarc
patreon
Clutch job on the big IBM tape drives
🕑 Added 2022-05-13 06:41:23 +0000 UTCThe IBM 729 vacuum column tape drives at the Computer History Museum need our help again. I am working on a tape that won't load because of a mechanical problem, and Ken is working on one that won't load because of an electrical problem. These are major repairs, they have been going on for over a month.
We think my problem comes from a bad magnetic clutch on the right reel. Here is the partially disassembled right reel shaft, at the back of the machine. It's temporarily suspended with zip ties because I slid the back casting support out.
The reel shaft is activated by 3 magnetic clutches. The two ones on the right, with the grooves for the motor belts, are the forwards and backwards motion clutches. But the problem clutch is the one on the far left, with the crown gear. This clutch is used only for slow motion of the reel during loading and unloading.
Taking that shaft out is a major operation, it takes about a day, particularly from an old machine where everything is stuck. But I eventually got it out and on the bench. It's easier to see the three clutches here.
It was another battle to take the clutches out without breaking anything, as grease had caked between the bearings, keyways and the shaft over the years. It took another day of gentle rubber hammer persuasion, but here they are. You can see the copper slip rings used to energize the clutch, with the carbon brushes that I repaired years ago (this video here https://youtu.be/7Lh4CMz_Z6M , in which you can see how that tape drive works).
I disassembled the clutch and I think I can see the problem. The black magnetic powder is everywhere, which is kind of expected, but it has leaked into the bearing, which is not good at all.
Normally there is a felt seal that prevents that, which I am pointing at in the drawing below.
It looks like the wrong felt seal was installed in this clutch, maybe during a previous repair. It had the incorrect diameter and thickness, for another version of the clutch. So it didn't fit and they put it at the wrong place.
Now the powder made it into the bearing, via the cavity that would normally be occupied by the seal. The bearing is also sealed, but the powder is very abrasive, so it eventually chewed up the bearing seal and seeped in. You can see the ragged top seal of the extracted bearing in the photo below.
To make things worse, this is a non-standard bearing, that you can't find anywhere. Fortunately, the old timers had found out that you can replace it with two stacked bearings, which are available. Here I am pressing the new bearings back in. Feels more like a car repair than a computer repair!
I then re-filled the clutch with fresh new magnetic powder, changed all the seals, and verified operation of the clutch (it takes about 250 mA to operate it). Here I am putting the shaft assembly back in. You can see how the main motor belts go over the two other clutches.
There is a lot more to do to reassemble the tape, but we should be able to test it next week, and see if I repaired it. If yes, then I can finally make the video. If not, I might make a video anyway. It's not everyday that you do a clutch job on a tape drive!
Marc
[Edit: I have made this post public as people from the museum requested to have it available as part of the restoration documentation]
Comments
Chris Faehl
This was really a fascinating process.
CuriousMarc
False. They were actual octopuses dressed with a white shirt and a tie.
Marcwolf
Is it true the old IBM techs are a hybrid clone between an octopus and a human?
