Log of November 2006 events

Back to Home Page

Tuesday, November 14,

Tuesday, November 14
These are preliminary notes of the RAMAC Tuesday November 14th meeting -

Next meeting December 5th, (November 28th canceled)

Items: Reversibility, Custody, coaxial plug/sockets, Head Pistons, Track Position Potentiometer, retain/record
Off-line, common vocabulary, Human Interface

Present were Al Hoagland, Dave Bennet, Joe Feng, Jack Grogan, Ed Thelen.

Next meeting November 28th.

  1. There was quite a discussion of "reversibility" as viewed in restoration circles.
    After some comments, it seemed a good idea to have Dag Spicer or Allison Akbay give us the views of museum people in general and CHM in particular.
    Al Hoagland will talk with Dag Spicer about his presenting the CHM perspective of artifact care, logging, "reversibility", and other relevant topics next meeting November 28th. (this meeting canceled!)

  2. There appears to be some question among the meeting members about the custody of this RAMAC. All agree that this RAMAC is owned by IBM, on loan to someone.
    One members believes that it is now on loan to CHM, and under "custody" to CHM - that CHM employees are responsible for restoration decisions. That 1st level responsibility (for restoration decisions?) would be with Robert Garner, next up the line would be Dag Spicer, ...
    Others (and Allison Akbay who Dave Bennet and Ed Thelen spoke with later) believe that the machine is on loan to Al Hoagland, with restoration responsibility in Al's hands.

  3. One of the external interface connectors that had been on the RAMAC when delivered to Santa Clara University had coaxial plug/sockets imbedded in it.
    Dave Bennet said that he called to original vendor of the interface connector, and the vendor said that connector had been out of production for many years.
    Apparently the Santa Clara students decided that coax signal connections were not required, and replaced the no-longer-manufactured connector with a connector identical with the three other connectors. The signals previously shielded in the coax are now conducted by 8 foot open wire. picture here
    The location of the removed multi-coaxial connector is not known. Dave Bennet will make inquires about its location.
    (Joe Feng says that he can "see" electrical noise from the nearby little "universal" motors in Tim Robinson's Babbage machines, and the compressor start up. Coax connections as in the original, as opposed to the open wire connections set in place by the students, would likely improve the signal to noise ratio of the read/write heads at external electronics. (Actually Joe is doing just fine with the existing open wire connections.)

  4. There has been rework on the head pistons (which push the heads toward the disk).
    Dave Bennet says that as received at Santa Clara, there were no head pistons.
    Apparently a version was made that did not work out?
    Then Dave Bennet made some pistons from drill bit shanks. These drill bit shank pistons are in current use.

  5. The open wire (track 80) of the Track Position Potentiometer - has been a cause for considerable "debate".
    A point of reference, this is the "Almadin" track pot as photographed by Dave Bennet.

    "Our" pot, picture here, has the "random" wiring (coming out of the back of the circular pot in the center) encased in a rubberoid "potting" compound - making physical access to the connections very risky and difficult (we all think)

    a) if we use the "Almadin" pot, will "our" display be sufficiently historically accurate?
    b) is this pot any better than "our" pot?
    c) could we swap or borrow this pot? Legal implications? ... ?

    Hitachi Data Systems (which bought IBM's disk business) has a 350 RAMAC with a track pot - all above questions apply.

    Dick Oswald would like the track pot question settled before he finalizes his switched ground servo design.

    (Working around a defective contact in the track pot increases the complexity of his design.

  6. There has been some pressure to retain/record what was written on the RAMAC as it arrived at CHM. There were questions about why write to more than one test (maybe in CE area) sector.
    Many options. Joe Feng has recorded in 100x oversampled form most (80%) of the RAMAC - and has an off-line program to convert the wave forms to BCD or ASCII?
    How to write on the RAMAC? From a long (100x8 bits?) shift register? Who do design and make? Who to design and make an on line read/write capability?

----------------- off line ----------------

I (Ed Thelen) think we should adopt a common vocabulary -

Ed's view of the RAMAC Servo world. (Ed is not a servo guy. Fifty years ago, he did fine in Servo 101, but did not take the follow-on course, deciding that digital computers were easier and "hot".)
"Name" Comment
Analog Servo To me, analog servo is what they teach in Servo 1 or maybe Servo 101 - and includes the system used in the 350 !!

Anything can set up the set point - target, including the 305 setting up the relay trees to set ground at a point of a pot.

Other alternatives to setting up a target (set point) can be manual (decimal ;-) switches) and a Digital to Analog converter similar to that used by the students.

Nation Instrument's LabView can also be used to set target (set point) and sense and control the various contacts and digital outputs to operate air valves etc.

LabView provides an easy development environment, nice GUI (Graphical User Interface), and coded to handle the RAMAC logic and I/O to/from the RAMAC can do a complete loop in about 50 milliseconds.

The above is using realistic code and I/O running in a reasonably fast lap top.
The above paragraph means that there is a maximum of about 50 milliseconds between a contact, and the response to the contact.

I think the convenient LabVIEW will slow a worst case seek from say 800 milliseconds to maybe 1000 milliseconds - barely noticeable by a human.

IBM 305 Implementation
or Switched Null or ...
A variant of analog servo (above) using relay trees or equivalent to cause a zero voltage to appear at the potentiometer (slider) spot at the disk or track to enable accurate (low detent wear) insertion to define the exact disk or track.

It was noted that the relay tree, with wire contacts, was a significant RAMAC failure source.

Using +-15 volts and semiconductor switches, the switching failure rate should be minimal.

The servo amplifier drives to cause an input of zero volts, a tachometer is used for velocity feedback.

Digital Servo a digital machine (the students used a PIC) can read the target voltage and slider voltage and the tachometer voltage (or a summing junction) and cause the output of driving voltages (or wave forms in the case of the students)

A much faster computer can do a very competent job to gain very high performance, using many sorts of optimizations, as in modern disk drives.

Human Interface

The following is an example LabVIEW "G Code" of driving a RAMAC. The large internal box conceals some complexity of the handling of 12 states or modes of driving to a disk and track. There is no attempt to do reading or writing to specific sectors. Creating the logic and user interface below took about three weeks. This was my first LabVIEW project - and includes interfaces to my new NI 6008 USB interface unit. The logic currently works in a loopback mode via wires connected to the NI 6008
To the left is the "Front Panel" which is set up for linear disk and track pots. You may notice the user adjustable slope and offset values for the two pots. The present disk and track positions are shown. A user selectable disk and track *or* random desired disk and track are also shown - along with the switch to select between user or random modes To the left is the control input logic, user interface logic, mode logic, and output logic. The mode logic is in the large box in the left center. Each of the 12 modes or states is another set of wirings in this box.

Please Note: The above logic and interface unit do *NOT* handle sector select nor reading and writing. The data rates and timing requirements for reading and writing are far above the capabilities of the above example and interface unit.