Schedule November 2005

• As you saw just before you left for lunch I was reassembling the 'supply side' clutch shaft. Very little problems other than getting the woodruff keys into the slots & the collars. Everything slid on the shaft without problem. As a matter of fact I had to disassemble the shaft unit a couple of times. No problem. It's completed and wrapped in a towel sitting up in the shop.

  Now to bad news.

• The shaft used in this assembly was the one that has many nicks & gouges from trying to remove the woodruff keys and from having been having the clutch removed & rebuilt & reinstalled some time in the past.
• I could not get the first of the clutch units to slide onto the shaft. I though that if I could get past the spot where it was binding that it would slide down the shaft. The unit jammed so tight that I ended up having to disassemble the clutch while on the shaft. I was finally able to 'press' the shaft through the rotor collar. The inside of the clutch rotor collar was scored from being forced over 'jam 'points on the shaft. I used 400 & 600 grade paper to smooth the shaft. I also had to smooth & remove the 'score' lines in the collar.
• Still believing that the shaft was not straight I brought it to Robert at IBM Research asking that he take it down to the shop area and have it tested it. Robert called to say that Bob Erickson had measured the shaft & found it 'true' within 2 or 3 thousandths. I do not know if he was able to rotate the shaft while measuring it. I called Bob, discussed the situation and we agreed to met next Monday at 12:30pm in his shop after his lunch break. I will bring a new bearing & will see if we can identify the cause of this problem.
• The question I thought I have is ( and Grant this is where you need to answer - the thought being to ask Bob to do it if you think it would the safe to do) can the shaft be mounted in a lathe, the cutting tool brought up to but not touching the shaft and then have the tool move down along the shaft removing only those little nicks & high spots without reducing the diameter of the shaft all along its length. Bob said that using 400/600 emery papers to cut down the 'tight' spots will not really work, that I might end up having to hand file spots on the shaft & then use the emery paper to polish the shaft.

1. Terminator resistors - The IBM 360 ohm & 120 ohm terminator resistors, (90 ohm to + 1.5V equivalent or 90 ohm to - 7.5V) were selected for Ground, and +6 V or -6V and -12V supplies. With +5V, Gnd, -7V, and -12V supplies, the resistor values should be:

      -N terminator     128.5 (130) ohms to ground and 300 ohms to + 5 V
      +P terminator     100 ohms to - 7V   and 900 (910) ohms to -12 V
   

2. The signals sourced by the TAU are terminated with the above terminators (located after the last drive) and have large signal swings.

      -N signal  high  +1.5V       (driver OFF)
                 low   -2.12V      (driver ON )
   
      +P signal  high  -4.04V      (driver ON )
                 low   -6.96V      (driver OFF)
   

3. The signals sourced by the 729 tape drives are terminated at the TAU end only and not after the last drive. The TAU receiver provides a 90 ohm termination, but is set up for 0 or 6 mA current. This results in about .5 V signal swing. Following are the expected levels:

      Signal Level   Voltage   729 driver output   driver state  reference
   
       -N   high      +.2 V       0 mA               OFF           GND
       -N   low       -.3 V      -6 mA               ON            GND
       +P   high      -5.7 V     +6 mA               ON          - 6 V
       +P   low       -6.2 V      0 mA               OFF         - 6 V

4. Readback of the signal levels for the 729 tape drives should be done relative to the coax shield. The coax shield is tied to the reference voltage at the receiver. It is AC coupled (10 uF in parallel with .001 uF) at the driver end. The TAU sets the reference voltage (TAU GND for -N and TAU -6V for +P signals) for the ingoing signals from the 729 tape drives. The 729 tape drives set the reference levels for the outgoing signals sourced by the TAU. (yes, the -6 V of all the 729 drives are bussed together by the outgoing coax shields for +P signals and signal grounds bussed by the shields of outgoing -N signals) Fortunately, the outgoing signals are large, so small voltage drops across the shield should not be a problem. We should confirm that the small voltage swing on ingoing 6 mA signals moves at least 200 mV either side of the reference to avoid noise problems. Otherwise, the reference in the tape emulator should be offset to place it in the middle of the signal swing.
5. Signals for all signal lines, ingoing and outgoing, can be received by a differential receiver connected between the coax shield (or offset version if necessary for 6 mA signals per #4 above) and center conductor. A 26LS32 quad differential receiver package should be adequate for all lines.
6. Outgoing 6 mA signal drivers can be made using a 75188 quad RS232 driver package. The driver output should go through a silicon diode (1N4148) and series resistor to the signal line. The resistor helps assure that the output current is 6 mA and also limits power dissipation in the IC package. The 75188 has asymmetrical propagation delays of about 80nS and 230 nS. The 75150 has lower delays of 45 nS and 60 nS, but has only 2 sections per package.

   
     -N Driver        ____________
                     |            |    1N4148      1K *
       TTL level ----| 75188      |------|<|-----/\/\/\----  -N output
                     |____________|                      - 6 mA into -.3 V
   
   
     +P Driver        ____________
                     |            |    1N4148      2K * 
       TTL level ----| 75188      |------|>|-----/\/\/\----  +P output
                     |____________|                       + 6 mA into -5.7 V 

   * Exact resistor values should be checked with actual IC drivers used since the output current limiting can range from 6 mA to 12 mA for the 75188.

What has been done so far makes sense to me and here are two other techniques we could use for smoothing the shaft:

We can cover the shaft with Dykem marking fluid and then slide a bearing onto the shaft. High spots will have the fluid removed and then they can be touched up with a file. Repeat this process until the desired slip fit is achieved along the entire shaft. This technique is a standard process for fitting custom made bearings and is also used in hand scraping items to flatness.

Another option would be to find a centerless grinder grinding job shop that would be willing to spend a few minutes in tweaking in a grinder to just touch grind the shaft to remove the high spots. I've watched centerless grinder operators and know there is quite a bit of "feel" in doing such an operation and hitting .0001" tolerances.

There are many grades of Loctite, including one specifically for bearing retention. It will be very difficult to keep any Loctite in the right place during assembly or to apply it post assembly. The clutches were difficult to disassemble without any adhesives, so to preserve future ability to repair them I think we should avoid using Loctite at this point.

The design of the assembly causes all of the bearings, split rings, and collars to be compressed by the locking nut on the end of the shaft. Making sure this assembly is tight is important for two reasons:

1. 1. it insures the driven plates inside the clutches don't rub on the back covers and
2. 2. it compresses the bearing inner races so they spin with the shaft.

I was also thinking about the sticking of the woodruff keys and suggest we apply a very light coating of Never-Seez to them during reassembly so the next removal is easier. We also need a non damaging key removal technique for the remaining clutch assemblies. I'll give this more thought, but am not aware of any special tools made for this, so I may need to make one.

Grant

http://www.repcoinc.com/carbon_brushes_fld/reference.asp

a very nice tutorial. I couldn't find the basic size we need (.250" round) on their web site so emailed them an RFQ for 100 pieces electrographitic of .250 round with shunt wire or .250 square (stock # REP324W) with shunt wire. In either case we (me?) would need to machine these to the dimensions of the IBM brush, .25" diameter head with .18" stem by tbd length. I've never machined brushes but guess it will be messy like machining graphite for EDM electrodes, something commonly done in machine shops. A little complication is to do this with the spring and shunt lead attached to the brush but I don't think that will be a show stopper with some fixturing.

One never knows what interesting digression will come up next with this project :-).

Grant