Sched June 2009

Schedule June 2009

return to main 1401 Restoration Page
go to Team Bios
Contents:
Wed June 3 - General, Thurs June 4 - Tape Team
Wed June 10 - General, Thurs June 11 - Tape Team, Sat June 13 - 2nd Sat
Wed June 17 - General, Thurs June 18 - Tape Team, Sat June 20
Wed June 24 - General, Thurs June 25 - Tape Team, Sat June 27 - 4th Sat

Wed June 03 - general

Present were: Ron Williams, Bob Erickson Bill Flora, Glenn Lea, George Ahearn, Joe Preston, Stan Paddock, Bill Newman, LaFarr Stuart, Ed Thelen, Robert Garner.
Ed Thelen and Stan Paddock examined an enlarged model of a core memory. Soon a difference of opinion arose about how the 1401 memory controller drove the memory, there are different options - soon two quarters were out to make the bet interesting.

Ron Williams was called in to arbitrate the dispute. Ron mentioned that he should get a fee for his expert opinion - who would bid the most? Soon Ed pocketed the quarters -
Ron suggested that Stan's quarter should be glued to Ed's hat as a trophy. Ron even dug out the hot glue gun for Ed to use ;-))
Ed's Trophy

Memory has been a pacing item in computers ever since Zuse (Germany) and Atanasoff (U.S) since the mid 1930s.
"Williams-Kilburn" CRT based memory of the late 1940s was a breakthrough in cost and performance, but it was a difficult and temperamental technology. The Manchester "Baby", Whirlwind, Johniac, Illiac, and other early computers used it for fast random access memory.
For random access memory, the less expensive, more reliable Coincident current core memory technology replaced the temperamental "Williams-Kilburn" memory.
"Core was King" for about 25 years starting with the Whirlwind computer. Large, high current (1/2 amp) vacuum tubes drove the SAGE core memory.
1401 memory was (apparently) designed before inexpensive high current (1/2 amp) transistors were available. As a result "Switch Cores" were used to help drive the X and Y drive lines in the 1401 memory.

This is how one core is selected. another view This is an overview of the system
And a view of the core stack And an image of the backside of the switch

At an other extreme, have some happy gears ;-)) This is the gearbox of an IBM 513 reproducing card punch.

Just look at that wonderful 7 tooth Geneva gear, which goes around twice per card (12 punch rows on card and 2 rows between cards)
Watch a Geneva Gear in action IBM 1402 and the IBM 513

Lyle Bickley (PDP-1 and IBM 1620 restoration) came by with a box of SMS cards which Weird Stuff collects and saves for him to give to the museum restorers.
We listed the loot html and xls, then compared the list with the 1401 SMS list. Four loot card types are used and two for which we do/did not have spares.
Later I compared the loot list and 729 Mod V, Plugging Sort by Card Type. Three other loot card types are used in the Mod 5 tape drives for which we have NO spares :-))
Stan reports,

After you left I decided to run my two test programs. This is what I learned:

The edit test ran the same on the CT machine and the simulator.
On the DE machine, it did not perform the floating $ for small numbers. Ron knows this from his "a penny a day" program.
-----------------------------------------------------------
I ran the stacker select program on both machines. This is what I learned:

On the CT machine, cannot select the center pocket from the punch.
On the DE machine, the punch gets a punch check on the third card punched.
Stan

Thursday June 04 - Tape Team
TAU Debug Status - 6/4/09 - (Sam S. & I)

Last week we noticed that the Overlap STAR would sometimes be ORed into other STARs when the various STAR buttons were pressed on the DE 1401's operators panel. Today we troubleshot that bug first. We found that the a microswitch behind the Overlap button was vibration sensitive. Sometimes the pressing of other buttons on the panel caused the "normally closed" contacts of this switch to open. There are two microswitches behind the Overlap STAR button and one of them was unused. So, we moved the wires from the failing microswitch to the other one. This appears to have fixed the problem.

The DE 1401 was unable to perform reads or writes to tape. The TAU was failing solidly, hanging in the first part (Read Delay) of Read operations. We found that the Delay Counter was unable to toggle the 8 bit.While probing we noticed that Delay Counter bit 4 had an asymmetrical up/down waveform and the trigger +output had a bad up level. We replaced the trigger card. The levels the new card produced were good, but bit 4 still had an asymmetrical up/down waveform and bit 8 still would not toggle.
We then found that a gate that controlled bit 4 was not functioning. We replaced that card and the Read operations stopped hanging, but would complete with errors and the Delay Counter would not stop after the operation.

We found another bad card (bad output level) in the DC Drive circuit. Replacing this card permitted tape operations to complete without errors.

We then booted the Autocoder assembler and it ran successfully to completion.

The bad level on the DC bit 4 trigger could have been with us for a long time, but the other two failures occurred (or degraded to the point of solid failures) over the last week.

If the DE 1401 shows signs of stability, we will move the 729 Emulator to the CT system. We need to position the Emulator monitor in a spot where it is visible from the operators console. Any suggestions on where we should move our table.

Regards,
Bob Feretich

Wednesday June 10 - General

Present were: Ron Williams, Bob Erickson, Frank King, Bill Flora, Glenn Lea, George Ahearn, Don Luke, Stan Paddock, Bill Newman, LaFarr Stuart, Ed Thelen, Robert Garner.
with special guest Shel Jacobs :-))

The gang poses after a lunch of bagels.
Stan was the photographer.
Just look at how those young guys can bend there knees !!!

A frazzled wire near a taper connection was spotted and corrected - but there are still intermittent read errors from the CT 1402 :-((
Robert Garner bought a DKA memory driver SMS from e-bay. It arrived in a box marked "BAD" :-((
Who would have imagined some one would want to use it ??
It is evident that the SMS card inventory lists done before the arrival of the Connecticut 1401 are badly outdated. Also the plugging list of a 729 Mod 5 has not been integrated into SMS inventory list.
Ed Thelen plans to have the lists combined by June 17 - possibly an up-to-date SMS card inventory could then be performed.
There is still some bug in the German 1401 edit command when handling the old (pre 1971) English money system.
There was quite a discussion of 1401 engineering changes leading to starting this web page.
Judith is on "leave of absence" to 12 week internship at Henry Ford Museum, Dearborn, Michigan. - blog

Thursday June 11 - Tape Team
There was no meeting -

Saturday June 13 - 2nd Saturday

Ron Williams, Bob Erickson, Bill Newman and Stan Paddock showed up today for the second Saturday work day.
When we arrived, all of the CHM parking lots were FULL! They even hired people to provide valet parking.
I saw Jim's E-mail on the Saturday work but missed the fact that it has been moved to NEXT Saturday. Well, we had a nice lunch cross the street.
Ron and Bob are close on the trial of the CT system's memory problem.
Bill and I continued our work on removing the "skip printing" wiring that had been done to the DE machine. We removed and installed the last of the revision changes and the machine is operational again as wired from the factory.
We had a return visitor from Australia who knows about the "Sterling" option installed on the DE machine.
The "Sterling" option was to handle the following problem.
Prior to decimalization, the pound was divided into 20 shillings and each shilling into 12 pence, making 240 pence to the pound. The symbol for the shilling was "s" � not from the first letter of the word, but from the Latin solidus. The symbol for the penny was "d", from the French denier, from the Latin denarius (the solidus and denarius were Roman coins). A mixed sum of shillings and pence such as 3 shillings and 6 pence was written as "3/6" or "3s 6d" and spoken as "three and six". 5 shillings was written as "5s" or, more commonly, "5/-". The stroke, /, indicating shillings, was originally an adaptation of the long s.
Various coin denominations had, and in some cases continue to have, special names � such as "crown", "farthing", "sovereign" and "guinea". In February, 1971, the British switched to the decimal system eliminating this problem. On the DE machine, the floating $ option of the 1401 Edit instruction does not work with small amounts . Ron and the Australian visitor believe there might be a conflict between the Sterling option and the Edit instruction. (To be continued)
Stan Paddock
San Jose, CA

Wednesday June 17 - General

Present were: Ron Williams, Bob Erickson Bill Flora, Frank King, Glenn Lea, George Ahearn, Joe Preston, Stan Paddock, Don Luke, Mathias Goerner (down from Berkeley), Ed Thelen, Robert Garner. - The museum was full of INTEL people, so we ate our lunch at tables on the north lawn :-))

Jim (Chip) Hunt, [JimHunt at Pacbell dot net], came for the day. Jim knew Robert Garner from SUN days. Jim also worked on attack submarines, and has wonderful, wild and wooly tales to tell ;-)) (And he seems to have a very practical working knowledge of electronics :-))

Glenn Lea was curious about the state of replacements for the 1402 card pocket clutch springs. Judith has taken a sabbatical, and we were not sure what folks had found. A call to Helander Products revealed that the pictures of the springs and clutch assembly Judith sent did not match the old IBM units available at Helander. We (Bill Flora) will send a sample so that several more spares can be fabricated. Hopefully they have the correct spring stock.

A new? problem? - temperature and vibration sensitivity in a 1406 12,000 character core memory. The 1406 supplies the upper 12,000 of the 16,000 programmer accessible memory location. Suddenly?, there was a problem - different bits of different characters were unreliable under different conditions of tapping SMS cards and temperature. Ron Williams found inconsistencies in the voltage patterns in the switch cores in the address switching matrix.
Normal
Skinny

A Detailed Overview of the memory Ron's study sketch, looks as though he has to struggle with this stuff like the "rest of us".

Details of the temperature sensitive switches three in middle are the series resistors. Like warm ;-))

Jim Hunt had a fine time verifying that the temperature switches worked at different temperatures. He found the middle switch of one spare AKB card with a 50 ohm "closed" resistance, should be about 0.1 ohm :-|
There was a lot more going on - Printers, Tapes, ... , but I got involved with sorting spare SMS cards :-((
Sorry to slight a lot of other folks :-((

Thursday June 18 - Tape Team

Saturday June 20

from George Ahearn
Ed:
Bill Worthington was kind enough to conduct a private tour of the CHM for my visiting relatives.
Thanks Bill

Wednesday June 24 - General

Present were: Ron Williams, Bob Erickson Bill Flora, Frank King, Glenn Lea, George Ahearn, Joe Preston, Stan Paddock, Don Luke, Bill Newman, Jim (Chip) Hunt, Ed Thelen, Robert Garner.
Well folks, I suppose the inevitable Business Controls have become necessary :-((
and the unlikely source was Stan Paddock, who would have thunk it?? ;-))

Ed Thelen was greeted by a delighted group proclaiming that:
not only was he 27 minutes late, but
he had to punch in late :-((
Stan said that
adding WD-40
to the dried ribbon
made it print just fine.

Ron Williams, that paragon of virtue ;-)) had already "punched in".
Does that look correct to you?? Really?? Do I smell a conspiracy ???

The intermittent problem with the 1406 extended memory has been fixed !! Of all the proposed theories, none guessed the actual fact - a physically broken current source card, with an intermittent connection across the break !!! How could that have happened ?? - more guesses - but no IBM exCE would want to suggest that one of their friends could have been so rough - "It must have been some one when off of IBM maintenance."

The goal of this circuit is to give a known, fixed current into a very "interesting" dynamic load of switching magnetic cores. After "Chip" added a patch across the intermittent -

Bill Flora brought in the two spare 1402 clutch springs (see Wednesday June 17 above). One is "right hand turn" and the other a "left hand turn". After an informal "committee meeting" we resolved that Bill should send one to the spring vendor for exact reproduction into both left and right hand version for future "events".

The goal of this circuit is to give a known, fixed current into a very "interesting" dynamic load of switching magnetic cores. After "Chip" added a patch across the intermittent -

Some of "us" may have to eat a little crow. Several years ago ?Mike Cheponis? purchased for us some "contact lubricant". Among many other considerations, should such products be used on museum artifacts? "We" sent the product names to an IBM lab, here is the report. "We" eventually decided not to use the products. So, guess what - This and another can of "SMS Card Contact Lubricant" were out, being use as a contact cleaner on an intermittent Tape Drive Select switch.

Here is Joe Preston, probably "not his best side" ;-)) Joe doesn't make much noise - but good things keep happening near him. Here he is doing something - I have no clue what - but I have confidence that "something" will soon be all better.

"Chip" Hunt was working on a defective operator's panel lamp on the CT 1401. It was burned out, where to find a replacement? We found a box of spare and new lamps, many had spilled out of tagged envelopes. Ed got distracted trying to sort the spilled lamps by glowing voltage and current. Things were going down hill. Ed called Ron Crane, and about 38 seconds things were "all better". Ron had tagged the envelopes so plainly and carefully, all we had to do was open our eyes :-))
Lyle Bickley came in with 60 SMS cards from Wierd Stuff. Lyle has been working with Weirdstuff management for several years - encouraging them to donate vintage artifacts to the CHM Restoration Teams or the CHM itself ;-) It turned out that 17 could be used as spares for the 1401s and 729 Mod 5 tapes.
See Card name vs # in 1401 & 729 Mod 5 vs # in Spares
Ed and Chip placed the cards useful to us in the spares. As we didn't have 5 of Lyle's card types in our spares, our spares situation is now a little better :-))
And a downer - the 519 reproducer has been limping for a while - Bob Erickson and Glenn Lea worked on it - finally Bob decided that there must be a tightness problem associated with the Geneva Gear. The Geneva gear has seven teeth, and the tightness is on a multiple of seven turns of the hand crank. Next time, the transmission is taken apart, again :-((

There was lots going on, groups working all over the place. The 13 folks seemed to have had a busy, productive, fun day :-)) I missed much of the action.
We may need more group reports, to go along with the "new" Time Clock. ;-))

Most people have left to fight he commute traffic. It had been a busy, productive day. Ed wants to leave -
- he has to punch out
- leaving that shameful late arrival for all to see :-(( ;-)) A tale - One of the Watsons got irritated about the casual working hours of the corporate staff at Armonk - and forced them "on the clock", punching in and out. !!!
Anyone got more details?

Thursday June 25 - Tape Team
from Bob Feretich and Sam S.

TAU Debug Status Report 6-25-09 - (Sam S. and me)

First, we verified that 5 Tape Autocoder still ran on the DE System. No problem, it worked.

Then we disconnected the 729 Emulator and moved it to the CT System. To place the Emulator where we needed it, we had to move the two CT tape drives a couple tiles to the right. When Sam tried to attach the power cable to the rightmost drive, one of the pins got pushed in. Further examination found that it had been being held in place with electrical tape. While Sam replaced the pin with one from the donor drive, I coupled the two data cables together (bypassing the drive) and tested the Emulator connection.

I found that multiple errors were occurring on the READ Bus. Bit A was shorted to ground, and other bits had intermittent errors. Close examination of the short (drive-to-drive) data cable revealed that the head end biscuit connector was broken all the way across. The crack went through the cluster of READ Bus contacts. Pressure on the biscuit connector opened the crack allowing the contacts to move. Several were partially pushed in. One was partially out and shorting to its neighbor. We replaced the cable with the one that was stowed under the floor.

Once the repairs were made, we tested basic tape operations, then ran the 5-tape Autocoder application. The basic tests (just some reads and writes from the TAU control panel) worked without errors. The Autocoder runs to a successful completion 30% of the time. A machine check occurs 60% of the time with a "all zeros with bad parity" OPCODE being loaded. The other 10% of the time the TAU hangs in a Backspace operation.

We did not try using any of the 729s on the CT system.

Ron W. tried using the left 729 on the DE system. It would not load tape.

NOTE: THE CT SYSTEM'S TAPE CHANNEL IS NOW TERMINATED BY THE 729 EMULATOR. EMULATOR POWER MUST BE ON FOR THE TERMINATORS TO WORK. EMULATOR POWER IS ATTACHED TO THE POWER OUTLET OF THE NEAREST TAPE DRIVE, SO IT WILL POWER ON/OFF WHEN CT SYSTEM POWER (The wall breaker, not the Ops Panel switch.) IS CYCLED. DO NOT TURN THE POWER SWITCH ON THE FRONT OF THE EMULATOR BOX OFF.

Regards,
Bob

from Sam Sjogen

Hey, I'm not "just" a software guy! Ironically, today I ended up spending most of the day fixing a 729's power connector.
Chris, note carefully that the power connectors on these drives are fragile. In rearranging our drives while moving our emulator from one 1401 system to the other we ended up with some damage to the pins on one drive and it took the rest of the day to repair it.
Using an "organ donor" drive I was able to obtain some pins and retention clips to do the job. Note that the pins are of various lengths, with the most precious being the one used for ground (longest, so as to make contact before the others). There seems to be a mix of signal and power pins, the latter being thicker than the rest and requiring different extraction tools.
We didn't have a pin extractor of exactly the right size, even for the signal and ground pins, so some improvisation was employed. Most of the retention clips were damaged in the extraction process. Each clip has three grab points, you're doing OK if you keep two of the three intact during an extraction. As you don't have spare drives, you might figure on using any extra power cables you're able to obtain as spares just for pins for the power connectors on your drive.
Cheers,
-Sam

Here are a few photos illustrating the power pin issue. Note in the close-up that the retainer clips are missing their teeth. http://gallery.me.com/redwoodcat#100111

Saturday June 27 - 4th Saturday

Go to July 2009
go to Team Bios

This is how one core is selected.		another view		This is an overview of the system
And a view of the core stack		And an image of the backside of the switch