Schedule March 2007

return to main 1401 Restoration Page
go to Team Bios

Contents:

Wed March 07 - general, Sat March 10 - 2nd Sat., - Volunteer Day - BBC visits ;-))
Wed March 14 - general, Thu March 15 - Tape Team,
Wed March 21 - general, Thu March 22 - Tape Team, Sat March 24 - 4th Sat.
Wed March 28 - general & core plane repair, Thu March 29 - Tape Team,


Wed March 07 - general

  • Present were Ron Williams, Allen Palmer, Robert Garner, Ed Thelen - and guest ex-1401 Customer Engineers Joe Preston ( JWL1230 at aol dot com ) and Sal Vicari ( DagoSal at sbcglobal dot net )
  • Allen continues to put the rebuilt components back into the 729 frames - a slow process.
  • The 1401 processor was down all day, with Ron Williams working on it. Just before we left at 4:00 it started working - for no obvious reason.
  • Robert Garner tidied up - we can see the top of the work bench again ;-))
  • At lunch we discussed who/how to demo the 1401 system and unit record equipment. Allen suggested the following:
    1. teach theory of 1401
    2. teach to run
    3. teach how to 'clear' 1401 problems - card jam, printer jam, ribbon, tape
    4. teach 'clear' 026 jam - with out sawing off print wires
    5. 077, 083, ...
  • Allen thinks we need a party - for many of us, June 4th is a third anniversary - June 6th is a Wednesday :-)))


Thu March 08 - Tape Team


Sat March 10 - 2nd Sat - Volunteer Day - BBC visits ;-))


Wed March 14 - general

Noon - brown bag lunch (bring your own) with Len Shustk (Chairman of the Board of CHM) noon to 1:30 - free pop & cookies

  • Present were: Ron Williams, Allen Palmer, Frank King, Chuck Kantmann, Joe Preston (new guy), Robert Garner, Ed Thelen.
  • Well - the 1401 came up sick again - It was working again a little before noon - no sick cards found. :-((
  • Here is the new guy, Joe Preston. Doesn't he look innocent - poor guy - little does he know what awaits him. Getting broke in - not so gently - he is still smiling - he clearly doesn't understand what he is getting into - or maybe he does :-))

  • Here is a 729 tape unit stack - fresh from Grant's shop - ain't that purdy :-))) Cleaned, polished. All new bearings, felts, magnetic powder, ... and the slip rings for one of the electromagnetic clutches :-)) Ready for another 40 years.

  • Here is Frank King. Whatcha doin with the 083 sorter, Frank?? Oh bummer !! Another card jam. Hey Frank, ya just can't blame the cards all the time
    Frank related the story of an IBM salesman who had a sorter jam during a demo. The salesman smoothly said "Look, the machine even found a bad card and destroyed it so it won't bother again." The savvy folks in the audience were so amused and amazed by the performance that they bought anyway - probably figuring any organization that was so clever and slick should be able to keep the machinery running. - Maybe our docents will need to have quick "work arounds"!!

  • People who don't dig British humor, or don't watch re-runs of "Are You Being Served" should probably skip down to the next box. As you can see, the Executive Chair is clearly out of place - away from Captain Peacock's desk. Only those privilaged to have a key to the Executive Water Closet are entitled to use this chair. !!

  • Allen has been worrying out loud about the magnets and/or snap switches which sense the position of the 729 tape unit capstans. There is about an inch of "runout" of the shaft of the capstan motor. When the motor is not running, a spring moves the motor rotor (and the capstan) back about an inch - so that the capstan is not near the tape during load and rewind. When the capstan motor is energized (turned on) the 3 phase armature iron become magnetized, and pulls the rotor into the field of the armature - just like a solenoid.

    IBM wanted to verify that the above system worked - so sensed the capstan position as follows:
    The diagram to the right shows the Capstan Motor Assembly. There are two of these magnet based sensors per capstan motor, one to sense capstan in drive position, one to sense capstan in retracted position.

  • This is the "top" view - actually the bottom view as the large metal casting is on the top side, protecting the switches, levers, magnets from harm from above This is the "end" view - as viewed parallel to the axis of the Capstan Motor shaft. The magnet in the center senses the shaft in the withdrawn position. The magnet and Front Switch on the right sense the shaft in the forward/front position.

  • For lunch we went to see Len Shustek speak to a total of maybe 25 volunteers about:
    - Museum finances, money drives, endowments, financial security
    - The hoped for new warehouse in Fremont, CA, and what is planned for it.
    - Plans for the Time Line and Museum - and how to comply with the Fire Marshall :-|
    - Restoration activities, maintenance of the restored systems,
    I felt much more informed about museum activities and plans


Thu March 15 - Tape Team



Wed March 21 - general

Later, Frank King wrote

We have worn feed knifes on both the 1402 reader and the 083 sorter in the 1401 room.
I believe the worn knifes are causing misreading in the 1402 and jamming 
    as well as misreading in the 083.

I have checked with Grant Saviers and a Tool and Die Shop in Morgan Hill; 
both have indicated it would be touchy to restore them to the original new 
specifications. This is because of the hardness of the material 
and the sharp 90 degree .00425 inch step required.
There are 2 other solutions that I believe are worth pursuing 
before we try to have them machined.

1.  Have Robert Gardner check with his contact that has a lot of parts for IBM machines.
     Two of the part numbers are 223082 (steel) and 603850 (plastic).
     I think that punch, reader and sorter are all interchangeable.

2. Ask Dag Spicer to allow us to interchange the knives with punch machines in storage.
     i.e. 2540s,1402s or other late model punches, (Late model being after 1960)
     The punch knives would have much less wear and even if the machines are used later,
     the punch units operates at a much lower speed 
          and would be much less prone to failure.

Please let me know if you think of other alternatives.
Robert, will you let me know if "your guy" has some of these. 


Thanks,

Frank
	
And Thursday, Grant Saviers wrote

I will attempt to make a hardened steel gage with  .0045 go and .0040 no go steps, 
so the knives can be measured.  This will prompt me to get my newer grinder on line, 
so don't hold your breath  :-) .  I'll pick through my odds and ends collection 
of diamond wheels and see if anything is worth trying on the carbide.

While diamond grinding wheels are required for carbide and will need to be carefully 
dressed to get the sharp inside corner on the carbide, a high end grinding shop 
should be able to refurbish the feed knives, particularly since the total material 
to be removed is likely only .001 or .002.  Can the pickers be set to offset the 
lower overall knife height when they are reground? 
   Please confirm or new knives will need to be made.

I'll make some more contacts to try and find a suitable grinding shop, 
and a service that can profile measure the worn ones.

I stopped by the 1401 room today and see that you removed the sorter knives.  
If you would bring them to my shop next week M/T/W, I'll make some measurements.  
Let me know by email what works for you, shop phone 408 929 8413.

Grant
	


Thu March 22 - Tape Team


Sat March 24 - 4th Sat

  • Present were Ron Williams and Tim Coslet
  • Tim Coslet worked on some double size "STAR" (STorage Address Register) cards. He reported that some of the new germanium transistors don't test out much better than the old transistors he has been replacing :-| Ron Williams is wondering if we should go to modern silicon transistors??
    Later, Tim Coslet e-mailed:
    About my comments about Ge transistors.

    I initially tested several with a meter and got what looked like bad readings on many.

    Later I retested them using my curve tracer and they all tested good. The only problem being that the junctions were slightly resistive instead of being "pure diodes" (which was probably what confused the meter). The new Ge transistors you have are perfectly fine! No need to go to Si.

  • Ron Williams found a logic level that was slow - need to trace what load (of 10 in the wired "OR" seems to be doing it) The page of logic showing that was missing.


Wed March 28 - general

  • Present were Ron Williams, Bob Erickson, Allen Palmer, Bill Flora, Glenn Lea, Mike Chaponis, Chuck Kantmann, Ed Thelen

  • First the *GREAT* news - Bob Erickson soldered a broken wire in the upper 8,000 character core stack,
    Details later in this massive message.

  • Allen Palmer and Glenn Lea continued re-assembling the other 4 tape drives. Allen took about 12 reference pictures.
    Some jobs in life just plain, unfortunately, involve getting down and under :-(( 1401 Customer Engineers were on the floor a lot!! Nature of the game. And Allen, who we love to tease as being Mr. Neat bought this neat kit

  • Mike Cheponis went back into the inner sanctum to find a missing page of logic (see Saturday Mar 24) and found the master, and reproduced it for the working copy schematics. There was some discussion of new, weak germanium transistors (see Saturday Mar 24). There is the general feeling that using modern silicon transistors in the SMS logic would be workable.
    Later, Tim Coslet e-mailed:
    About my comments about Ge transistors.

    I initially tested several with a meter and got what looked like bad readings on many.

    Later I retested them using my curve tracer and they all tested good. The only problem being that the junctions were slightly resistive instead of being "pure diodes" (which was probably what confused the meter). The new Ge transistors you have are perfectly fine! No need to go to Si.

    Ron Williams, Bill Flora, and Chuck Kantman found a card that was excessively loading a driver, causing a slow rise time which may have been the cause of some of our unreliability.
    Chuck Kantmann is in the background Ron says the bottom trace should be the mirror image of the sharp upper trace -

  • Doug Engelbart (Mr. Mouse, Mr. Window, Mr. Networked Computer) and Alan Kay (many of us bow reverently) came through with a group of Japanese students.
    Doug Engelbart with Bob Erickson :-)) The core stack Bob is fixing is in the lower right of the picture. Doug Engelbart and Alan Kay (back to camera) signing autographs at door to 1401 room after visit.


    OK - I better warn you

    - if you aren't a techie, you might not enjoy the following techie core memory stuff

  • So - 85 year old Bob Erickson had just returned from a cruise and trip to Australia. He and Ed Thelen were chatting about it. It turned out that a moderate quantity of Foster's Beer had given Bob gout in the little finger of one hand for a while - all better now - but sadly another phase of his life (beer) is out-of-bounds :-(( I (Ed Thelen) can empathize :-((

    Then Bob wondered out loud what to do next. I suggested making plastic covers so that people can see into the insides of 1401 equipment he has done such a good job with the 077 collator - but Bob figures that the solvents in the glues are tough on his lungs

    - how about fixing the broken wire in the 1406 extended memory.

    Great - any programmer is delighted getting more memory ;-))

    But no one present had ever heard of a field repair of a core plane. (Did Bob say he fixed a 704 core plane at Los Alamos?? - I will call Bob tomorrow.)
  • Core memory, quick memory refresher,
    • The game is to make cores as small as practical to increase switching speed and decrease required drive current
    • Each core in the regular kind of core memory has four wires through it
      1. A half select wire in the X direction,
      2. A half select wire in the Y direction
        - each half select threaded through every core plane
      3. A write inhibit line, going through all the cores in its own core plane.
      4. A sense winding through its own core plane threaded in a diagonal way
        - to minimize noise from select currents and imperfect core hysterisis curves

  • A little history -
    Mid 2006, the 1401 and 1406 (extended memory) was working well enough for Ron to determine that the top 8,000 character memory stack had a periodic single bit error that was not caused by a bad logic card - likely an open core stack half select wire - :-(( - There are other problems in that core stack, see here.

    Sept 1, 2006 - Bob Feretich and Ron Crane got down and under the low 1406 unit and probed with an ohmmeter and determined which terminal in which core plane was showing open. By great good luck, it was the bottom plane - most easily damaged, and most easily accessible.
    Here is their log entry And here is the red dot they placed indicating the area of the defect :-)) You can see the array of interconnecting half select current wires bridging between the core planes. Aren't you glad for cheap, fast, miniature, "easy" semiconductor memory - you can buy a gigabyte for $50 and use it as a bookmark !!

  • Bob talked with Ron Williams who was working on a 1401 problem - Ron asked if Bob wanted power shut off - Bob said no, he wouldn't interrupt anything. (The core half select windings are DC isolated from power supplies by the switch cores, see figure above,

  • but you must admit that most people would drop power fearing a slip of a probe or wrench could bring an overcurrent and smoked components. I know that I would !!)

    So with power on, and Ron working on the attached 1401, Bob Erickson and I started looking at the 1406 and taking out the sick 8,000 character core stack and associated drive circuits :-|

  • Glenn and Ron helped release the chassis from the slide assembly, and we used Ron's carpeted desk top to support the pin side with out straining the pins or shorts - (the unit is still hot and running) Bob getting serious !! Did I mention Bob is a trained watch maker and still uses his jeweler's lathe? The 4th wire on the right of the closest core plane had a break at the terminal. (The end terminal is not used, the 5th terminal connects to the 4th wire.)

  • Fortunately Bob had plenty of supervision - none of whom had ever even heard of a field repair of a core plane. Left to right are Ron Williams, Mike Cheponis, Bill Flora, Chuck Kantmann, Bob Erickson. Bob brought the large spool of #33 enameled (red) wire in case he needed more. (He used 2 inches. ;-)) He also brought some cores.

  • Bob has fished the "V" shaped wire under the vertical wire with the broken end at the pin, and is about to pull it gently back through the frame. If he stresses the near by core too hard, it will break, and there will be hell to pay! AH - the end is free - now all he has to do is strip off the enamel from that little bitty wire, solder the extension wire onto the terminal, thread the extension wire through the hole in the frame, and do some fancy soldering!!

  • Bob is wrapping an already stripped and trimmed #33 wire around terminal #5, for wire #4, preparatory for soldering. Soldering the extension wire to the terminal. Next Bob will thread the extension wire through the correct hole in the frame. Another moment of truth - scraping/stripping the enamel off of the #33 half select wire. Too much tension and you break a core, you can also cut the wire making the next attempt even tougher!!

  • Soldering is complete. Ron Williams reported the error caused by this broken wire is now fixed. Re-assembling the 1406 extended memory unit. CONGRATULATIONS !! Ron and Allen congratulating Bob. Does Bob look a little tired?? Sports announcers love to talk about stress - Yeah - I bet Bob knows about stress. One slip in the past two hours and there would have been almost fatal damage to a core memory 1960 price about $55,000 !! (about 4 houses at the time)

    Added November 1, 2010 - factory repair of core stacks.


  • Jim Somers showed up with chilled apple cider ;-)) I think Jim is trying to explain about proper celebrating in a museum. A bottle opener is one tool we don't have !! Allen struggling with a wrong tool. This wasn't the correct tool either, but it worked ;-))


Post script:
I checked with Ron Williams and Frank King to make sure that I was on the straight and narrow. Ron Williams said

Yes - the problems caused by that broken wire are fixed - but there are still problems in that upper memory:
  1. The "B" bit does not work from 8,000 to 15,999. The write inhibit and sense cards seem OK.
    It would be good to use a current probe to verify circuits are working as expected.
  2. The "4" bit does not work from 12,000 up.
    Same equipment would be useful to check this situation also

Frank King said

He had never heard of field repair of 1401 core planes - send the stack back to the factory.
  1. He had tried to use factory authorized tools to fix a much larger (capacity and cores and wires) Large Capacity Storage. He had poor results, bent wires, ... and was much frazeled. He told his supervisor that on the next problem, send the memory back to Kingston, or send out a factory person to try.
  2. Part of the special equipment that came with Large Capacity Storage was a gold soldering "iron", something you would not leave out on a work bench.


PostPost script: - is there such??
John Van Gardner, of near Atlanta, Georgia e-mailed

Ed,
The guys did a good job on the 1406 array. I still have my Ungar Imperial soldering iron with the gold plated iron tip and a roll of .010" diameter plastic rosin core 60/40 solder used to work on 2361 LCS memories.

If they never heard of core array repair in the field they never read my story 43.5 1401 Print Buffer Surgery. My failure was in the middle plane and I had to split the array apart, but I cheated. I turned the power off ;-))

Van Gardner
More of John's stories can be found here here and computer stories in general here. :-))

Kristin Abkemeier, who has interviewed Bob Erickson for a story e-mailed

... I just checked out the project website and saw the photo essay on Bob Erickson fixing the core plane! Amazing! This totally has to go into the story. I wish I could have witnessed this. How many core planes on the planet still get fixed by someone who has firsthand knowledge from when they were the latest technology?

If this was Japan, Bob would be a living national treasure. (I hope he had a great time on the Queen Mary 2, which I saw sail into SF the weekend after I last visited y'all.)

And hey, if you're going to have a party in June, don't forget to invite me!


Thur March 29 - Tape Team



Go to
April 2007
go to Team Bios