Friday, December 19, 2014

Ruined disk heads removed from Diablo drive (external drive, not primary 1130 disk), console printer adjustments


The main problem appears to be sticky behavior of the margin rack which should be popping to the right when a carrier return begins in order to be slammed left as the carrier reaches the left stop. I am also a bit uncertain of the adjustment for the linkage that will release the rack, as that might not be ensuring the release is made every time.

In hand cycling to test this, I saw the machine begin to operate poorly in another area I thought already resolved - tab movements. The tab mechanism on the rear of the carrier should be locking up until it is released by a set pin at some target column, but instead is not locking so the tab movement ends prematurely or is only a few spaces long.

This appears to be the sludgy old lubrication still causing problems, alas. I am working with 99% Isopropyl Alcohol to clean some of it out, everywhere I can get to in these mechanisms, and then replace the lubrication with an appropriate light oil (Mobil 1).

I removed the tab overthrow stop and gang clear plates from the rear of the carrier, which leaves the escapement, backspace and tab mechanisms clearly visible and accessible. I was able to carefully observe the operation of the mechanisms which led me to the likely cause of my erratic tab movement.

The tab torque bar is twisted by the cam and levers on the right rear of the machine when the tab operation is selected. The bar pushes the tab lever toward the rear, pulling the escapement and backspace pawls out of their racks to allow rightward movement of the carrier. The tab latch should set, keeping the pawls out of the way until the tip of the tab latch strikes a set pin on the tab rack, which pivots the latch, releasing it.

The latch is not setting - but if I push the linkage to rotate the tab torque bar myself I can make it reliably latch. Thus, I think that I can shorten the linkage, which has a turnbuckle for adjustments, to ensure that the tab latch is pushed far enough to reliably engage.

The failure of the margin rack to be released to the right when a carrier return operation begins is a similar situation - a lever pulls a linkage that releases a latch on the right of the margin rack. I will adjust that turnbuckle to make it shorter, thus moving the release arm further on the margin rack release mechanism.

After adjusting the tab torque linkage, the tab latch is setting reliably. Whenever a tab is commanded it latches and moves to the right, only resetting when it hits a set pin on the tab rack. Excellent behavior. One major problem resolved.

I shortened the linkage to release the margin rack but it isn't so reliable. It appears that the range of movement is too short for the lever that pulls on the linkage. I am going to dig further into the mechanism and adjustments to sort out why and what to do about this.

I did find that if I am doing a right tab and it runs into the right margin, it isn't releasing the tab latch. I need to see how this is supposed to occur before I can sort it out. It may be related to the problem with the rack release, or not.

I did find a spring under the mechanism, undoubtedly fell off the typewriter somewhere. I need to figure that out and replace it, needless to say. My first spots to check will be the carrier return and tab sections of the operational shaft, about where it fell, as these could be the cause of my two remaining problems.

After investigating the theory of operation manual, I realized that the tab latch will remain set when the carrier runs into the right margin stop. It is when the carrier return operation begins that this is resolved, because the escapement torque bar is turned by the carrier return mechanism in order to keep the escapement and backspace pawls out of the escapement.

If that didn't happen, the carrier would make a ratcheting sound as the escapement pawl skipped over all the teeth on the escapement rack. The twisting of the escapement torque bar also swings part of the tab latch mechanism out of the way, releasing the latch. Therefore, the right hand margin behavior is not a problem.

I reached the end of the evening without a clear diagnosis of the problem with the carrier return mechanism. I will study the theory, parts and adjustment information, plus review some video training on Selectrics, preparing to tackle this again tomorrow.


I removed the disk heads from the Diablo drive that came in the CHI rack - they show major damage from a head crash and are unusable as they are. They have crashed into the surface of a disk, ruining both the disk cartridge and the heads.

Head crash damage visible on the heads - scrapes on metal plate and magnetic oxide wedged in scratches

Rectangular slow beside upper holes, to side, with brown iron oxide coating from disk packed over ceramic head inside
You can see lots of scraping on the shiny metal surface of the head, caused by various particles which were embedded in the disk surface which rotated under the head. Also, you can see significant brown ferromagnetic coating material from the disk surface has been packed in the rectangular gap which exposes the ceramic part of the head and its metal poles that do the actual reading and writing.

First, the oxide material has to be cleaned out of the gap, then the condition of the underlying ceramic and electromagnet poles must be assessed to see if they can be saved or need rebuilding/replacement.

The metal plate has to be removed from the arm assembly and the internal ceramic, coil and pole magnet assembly which is the functional part of the head has to be removed from the plate and set aside.

Next, the metal plate would have to be polished to remove the scratches, perhaps with a new layer of the chrome-like surface added.

After removal of scratches, the metal surface must be returned to the curvature it needs to fly aerodynamically on an air cushion at a target height above the rotating disk surface, without wobble and certainly without oscillating vertically in a way that might bring it into contact with surface irregularities or the disk itself.

When all else is done, the ceramic/coil/pole part has to be put back onto the plate at the right alignment and depth, the plate affixed by microwelding to the arm in the proper orientation, before it could be put into the drive and used again.

As you can tell, rebuilding a head requires special equipment, skills and quite a bit of care. Years ago there were several services which did this processing, taking damaged heads like these and returning assemblies that were rebuilt to be as good as new. Once the arm and head began to be built into the disk drive itself, users no longer removed and replaced heads as a maintenance activity - without that process, the need for head rebuilding services evaporated.

If I am lucky I will find a service that still is able to do this. There may be a tiny market handling repairs to systems that must remain in service even though they are decades old and no longer have spare parts availability. Defense systems are just one example of situations where really old technology is still being used with the need for repair/rebuild services supporting an ecosystem. The cost of replacement of the technology with more modern tech is so high in these cases that is justifies high charges for service and rebuilding, even at the extraordinarily low volume of such activity.

If anyone knows of a service or person who can rebuild the heads, please let me know. If you know of someone who has replacement heads for the Diablo Series 30 disk drives or something compatible, also pass along the information.


I will try to solder on a replacement microUSB connector onto the fpga board. Digilent replied but does not do repairs. They offered to process a warranty claim but it is clear damage by me, not a defect, so that is off the table. I ordered the receptacle from Digikey and should have it in a couple of days, slight delays due to heavy holiday shipping volumes and of course a few actual holidays where packages stop moving.

No comments:

Post a Comment