Magnetic Drum

The heart of this machine is a magnetic drum, 7” long and 6.5” diameter that rotates at 3600, 3700 or 4000 RPM, depending on what source material you look at. A tonewheel at the end of the drum is used as the system clock and 3 additional heads read pre-recorded timing data off the drum, used combinationally in defining the machine’s operation cycles.

Then there are 64 heads, corresponding to 64 Tracks (staggered 0.075” apart), and each track (.040”) is divided into 64 Sectors of 32 bits each (only 31 bits are made use of). 64*64=4096 words of memory. Addressing is in the form of Track:Sector.  The first 6 bits select the head via a diode matrix and the second 6 bits select the sector.

In addition, there are three recirculating registers, with independent read and write heads for each, spaced a single sector apart. These tracks correspond to the Accumulator, Program Counter and Instruction Register and get continuously updated. An additional head for the Accumulator is used during multiply and divide instructions.

The good news is that as long as at least one of the 64 long tracks is usable, I will have 64 words of mem to work with and can get the machine off the ground. That still means I need functional short registers and intact timing tracks.

Monday, Jan 2, 2017: With car unloaded, I surveyed the drum. The motor and belt were carefully removed. First problem- the drum was ‘stuck’. Not wanting to risk damage, I carefully removed the bars containing the (offset) head assemblies. From what I could discern, there were many heads in direct contact with the drum surface. If someone had tried to rotate the drum in this condition it would have likely destroyed all of the tracks. Having removed all the assemblies except one, I made the mistake of checking to see if the drum was still locked. Much less resistance this time, and a ½” scratch through the oxide on one track. Dammit. This is the kind of goof that hits you in the pit of the stomach. Further inspection revealed two other scratches down to aluminum that were not my fault and those heads had already been moved to open positions. In retrospect, it was a good thing that nobody had tried to power up the machine of they would have destroyed the drum on the first revolution.

Tuesday, Jan 3: With delivery of the machine proper still some time off, I prepared a checklist for my order of ‘resuscitation’ of both the drum and machine and continued work. For the drum, I made 0.004” spacers out of strips of notebook paper, punching a hole in each for the head assembly arm screws to pass through. This put the 3 timing and 64 ‘long line’ heads up out of harm’s way. Two layers of kitchen foil made a non-magnetic 0.0015” feeler gauge that I then used to get the recirc register heads set. This is a real trial and error process since the head mounting screw must be loosened before the adjusting screw can be set, but tightening the mounting screw after adjustment affects the overall gap. With heads cleaned and elevated and the recirculating registers gapped I put the drum back together and set it aside. Module work would be next.

In the interim, work progressed on the rest of the machine with the plan of installing the drum only after all other systems were ready to go.

Friday, Feb 10: The time had come to sit down with the drum and get the heads adjust as best I could. This was no easy task as the location of the first and last bars prevented head adjustment with the bars secured; other bars could only be adjusted with the adjacent bar removed. In order to adjust a single head, both top and bottom allen head screws must be broken loose, then ‘snugged up’. A center adjusting screw raises and lowers each head. If either of the securing screws is too loose when the adjustment is made, it will throw off the setting when torqued. If too tight, you can’t adjust the height. After several tense hours of carefully removing bars, elevating heads, installing bars, adjusting for .001″ clearance and removing the bar again for access to the next row, the drum was initially set. 

The fresh v-belt was installed with the massive motor and the assembly moved out to the machine. Interestingly, a pair of spacers mount between the drum and rubber isolators to tip the motor forward enough that it doesn’t contact the sheetmetal of the rear module assembly. I suspect this motor might be a replacement during the refurb and that originals may have had a smaller case.

The moment of truth: Direct AC connection for spinup. And…light bearing squeal. Exactly what I didn’t want to hear. There is no practical way to replace the bearings on this drum without full tear-down, and there’s simply no practical way to achieve the tolerances in this assembly without specialized equipment and adjustment. And the odds of achieving that on a ‘first try’ without a head crash? Nada. But I got lucky- I removed the belt-drive pulley and was able to see the right side bearing and partial shield! I was able to shoot some silicone spray with PTFE into the bearing and reassemble. This eliminated the bearing noise and was enough for me to get the rest of the machine powered up.

12 Feb: Still no bearing noise, but I decided to remove the belt and pulley again and wipe out the old grease that had now become ‘liquid’. Clearly, the long-term fix was to get proper grease back into that bearing, but it would involve some ingenuity. Using a modified turkey baster “flavor injector” did the job perfectly. Grease forced in under pressure and exactly where it needed to be.  Unfortunately, there’s no way to gain access to the left side bearing, so I’ll cross my fingers it behaves itself.