Schedule January 2007

Wed Jan 03 - general,

Thu Jan 04 - Tape Team from Bob Feretich

TAU Debug Status - (Ron W. & I)

The "tape stretching" that was reported yesterday was not a problem for us. We "loaded" tape about a dozen times today. The tape was smoothly drawn into the columns every time except once. That time the supply column stuttered once then loaded.

We observed that the TAU's skew and error indicators were stuck on. This was a new problem that cropped up sometime in the last three weeks. We traced the problem to two bad cards (a 3-input NOR on one card and a 2-input NOR on the other). These two NOR gates are cross wired to form a set-reset latch. The outputs of each of these gates were stuck minus (-5.8 volts). Either of these gates having its output stuck minus would cause the observed failure. It is curious that both of these cards (card bags labeled 02b3c15 & 02b3f15) failed within three weeks of each other. Coincidence???

During the last work session I fixed a card punch error in the Move Tape (5040) diagnostic, but was unable to verify the fix. Today, we loaded and verified this fix.

We set the sense switches to continuously repeat the Move Tape diagnostic and ran it for nearly the entire length of a tape. (about 10,000 fifteen-character records) For this diag, a record is written, the tape is backspaced, the record is read and compared, and then the sequence is repeated with a different record.

Results of the long diagnostic run:

• Sixteen error clusters occurred. Each cluster consisted of three to eight consecutive failing records.

• Bad spots on tape caused some of these error clusters. (Fewer errors occurred as we moved deeper into the tape.)

• Eight of the clusters showed a write/read data compare failures on the first record of the cluster. No other data compare failures occurred. Other errors were parity or skew errors.

• On most data compare errors, a single bit was picked or dropped. Picking the A-bit was most common, but sometimes the A-bit was dropped. Sometimes the entire record was trashed.

• Scoping the OR of the received clipped data bits shows that all bits of a character arrive within a 2.5 microsecond window. We were running at 556 cpi density (about a 22 microsecond character frame). I don't know if this is within specifications, but at least one skew error was reported.

• Frequent CPU A-Reg errors occurred. (TAU read data is delivered to the A-Reg.) The error indicator glowed dimly during the entire run. These errors were not being reported in the TAU error indicators nor printer report.

• Printing an error report line often turned on the "Printer Check".

• The diagnostic finally crashed when the 1401 CPU branched into the print buffer (201- 332) and loaded a Group Mark as an opcode. The error report showed that a longer than 15-character record was read just before the crash. Data read from the tape is transferred directly into the print buffer. The program starts at about address 400. It is possible that the erroneously large incoming record trashed the beginning of the program, but a Group Mark with a Word Mark should have protected the program.

We disparately need spare TAU SMS cards. There were no TAU cards in the Repaired Cards box. Only two TAU cards were in the BAD Cards box. I found a batch of bagged bad TAU cards in the rolling shelf unit. It looks like they were removed from the Bad Cards box so that they would get higher repair priority, but put aside and forgotten. I placed them back into the Bad Cards box (they are all labeled TAU). We would appreciate getting these cards repaired ASAP.

Regards,
Bob

Wed Jan 10 - general

Thu Jan 11 - Tape Team

TAU Debug Status 1/11/07 - (Ron W. & me)

We started out a few steps back again today. Characters that we read were received with their 8-bits set to zero. The failure was consistent so it was not too difficult to troubleshoot. We isolated the problem to a jumper wire on the back of the 729's read amplifier gate. The pin socket that was crimped onto the wire was failing. Wiggling the wire at the socket made and broke the connection. There was even a state that passed an attenuated signal. This may have been responsible for some of the intermittent errors that we saw in previous weeks.

We then ran the Move Tape diagnostic (5040) for a good portion of the tape without error. Previously this diagnostic seemed to cause a large number of 1401 A-Reg parity errors. None of these errors occurred during this run.

We ran the Card-to-Tape (5000). Eight to sixteen errors were reported each time we ran the test. The most common error seemed to be TAU compare errors with the TAU's A-Reg set to 0000001 and its B-Reg set to 0000000. Neither of these characters are legal, so we think that the TAU is sampling data when it not supposed to. These errors occur on the read back part of Write instructions. The error is probably occurring in the read back timing. The primary difference between this test and Move Tape is that the tape operations in Move Tape occur more quickly so the tape flows smoothly. Card-to-Tape incurs a card read delay between each record, causing the tape prolays to chatter. We checked the Write Delay and the Write Disconnect Delay. They look ok. The root cause is still TBD.

Other strange things we observed:

• We were running Card-to-Tape with the read back sense switch (G) on. After several minutes of running, the 1401 jumped to execute a Group-Mark with Word-Mark instruction in the print buffer. We verified that the begining of the diagnostic was not corrupted. We also saw this happen once last week with the Move Tape diagnostic. 
• Twice today the 729 lost Ready status for no apparent reason.
• We ran the tape off the end of the reel. It took a lot of messing with the 729 to reset it after that. (Including a few power off/on cycles.) I am not sure what we did to reset the drive. I have also seen this happen several weeks ago.
• On occasion, the 729 refused to Unload tape. I think this a problem with the mercury switch on the back of the take-up shaft. I adjusted it so that the contacts were horizontal and the problem went away. Might be a fix or a coincidence. We need to wrap the switch's bulb in something tacky so it neither slides nor rotates in its clip.
• We accidentally executed a Rewind operation using the address of the 1402 Card Reader. Several cards got mangled as they were sucked up from the output hopper.

Regards,
Bob

Sat Jan 13 - 2nd Sat

• Present were: Ron Williams, Bob Erickson, Tim Coslet, Ed Thelen, Grant Saviers, Jeff Stutzman and Ron Crane.

• As promised, we weighed a 729 tape drive power supply - with a bathroom scale. It weighs 140 pounds. Well, OK, that is about 2.3 60 pound bags of cement - and just about as easy to handle.

• About one hour after starting the 1401, it began misbehaving :-(( - classic temperature sensitivity. Ron Williams finally traced it down to one transistor in the Overlap system affecting the Storage Address Register (STAR). Chill the transistor and the machine works, heat the transistor and the machine fails. The transistor leaked too much as the temperature rose :-((

  The red line in the right hand screen shot is where the base line should be. The leaking transistor floated the base line (a 'false' value) up (into the 'true' range) confusing the machine.

• Tim Coslet expressed frustration trying to test the double sized SMS cards with out a test fixture to aid making power and signal connections. After some discussion, Grant Saviers will make a double wide test fixture somewhat similar to the single size SMS test fixture made by Ron Williams (with lead in the base to increase stability ;-)) Grant plans to use two SMS sockets rather than the green S-100 BUS socket shown.

• This is an overview of the south west corner of the 1401 room - with some serious folks. Tim Coslet (left) and Ron Crane are fixing sick SMS cards.

• Bob Erickson and Jeff Stutzman worked on the plastic shield for demonstrating the 077 collator. Jeff came up with the ideas of: - use chrome edges to hide the glued corners - use thin lamps under the chrome edges to light the machine

  And no, the 077 was not modified to install the plastic. The brackets are part of the machine to secure the normal metal "skins". Some extenders were crafted to better place the plastic.

Wed Jan 17 - general

Thu Jan 18 - Tape Team

TAU Debug Status - 1/18/07
The new sampling scope that Grant brought in works great. It was essential for troubleshooting the intermittent bug described below.

The WriteTapeMark (WTM) instruction was not working. When executed the CPU hung waiting for the TAU to complete the instruction. I replaced a card in the WTM logic to fix the problem.

I erased the entire tape, then executed a read record operation from the TAU control panel. One of the manuals recommended this technique to troubleshoot intermittent errors. A specific error occurred frequently. The error indicators were: Error - on
A Reg VRC error - on
R/W Reg VRC error - on
CBA8421
A Reg = 0000000
B Reg = 0000000
R/W Reg = 0010000

The A, B, and R/W registers are reset to zeros at the beginning of a character time and bits are set to one based upon the character received on the Tape Channel Read Bus. This is interesting because data from the 729 passes through the A-Reg or B-Reg to get to the R/W-Reg. I set the scope up to trigger on activation of the Error latch and started looking for the cause of its activation. I found a 1 microsecond glitch that occurred 85 microseconds before the Error latch activated. The glitch originated in the 729 and was passed through the A-Reg, but was too narrow to trigger the A-Reg latch. Either the R/W-Reg latch was faster or the logic along the path stretched it slightly.

The glitch's root cause was noise in the 729's read pre-amp. The gain on all the read pre-amps were set much too high. Instead of 8Vp-p, the signals (measured at the 1401 to 729 interface) were 12 to 14Vp-p.

We previously set these levels to 8Vp-p, but I think that we made substantial 729 power supply adjustments after that. The adjustments must have seriously impacted the levels of these signals. I remember during our first the pre-amp level adjustment, the adjustment pots were set at their upper limits to obtain the 8Vp-p level. Now the pots are set much closer to center.

After these level adjustments, the above error still occurred, but much less frequently.

I then ran the tape diagnostic battery as a batch. Move Tape, Load Tape, and Tape VRC diagnostics ran without error. The Backspace/Skip diagnostic was last in the batch and the card reader started mis-behaving. After several read attempts I got this diagnostic loaded, but multiple errors were reported. I am not sure that the diagnostic program loaded correctly.

I did not try to run the Card to Tape diagnostic today.

Regards,
Bob Feretich

A Proposed Demo

The following is a Demo of a proposed demo of the 1401 system -
  a) Guests enter the 1401 area and are introduced to the 1401
          and card processing in general
  b) Guests are invited to key punch their name(s) onto an IBM
      card starting at column 1
  c) The guest's name card is inserted into the demo deck,
         which has already been set up with a date card
     - There is one card run per guest card,
         - The first name in the guest card (say "Steve")
           is printed in double sized BIG characters
         - The second name (say Wozniak) is printed further down
            the page in double sized BIG characters
         - A fixed greeting is printed in single sized BIG characters
         - The date card is printed in single sized BIG characters
         - The rest of the deck (pairs of 65 column data cards)
            are printed in normal print 130 characters wide.
           This text is by Robert Garner, introducing the 1401
  d) Each guest gets to keep the IBM card the guest punched,
           and the printout  :-))

    SSSSSSSS  TTTTTTTTTT  EEEEEEEEEE  VV      VV  EEEEEEEEEE
    SSSSSSSS  TTTTTTTTTT  EEEEEEEEEE  VV      VV  EEEEEEEEEE
  SS              TT      EE          VV      VV  EE
  SS              TT      EE          VV      VV  EE
  SS              TT      EE          VV      VV  EE
  SS              TT      EE          VV      VV  EE
    SSSSSS        TT      EEEEEEEE    VV      VV  EEEEEEEE
    SSSSSS        TT      EEEEEEEE    VV      VV  EEEEEEEE
          SS      TT      EE          VV      VV  EE
          SS      TT      EE          VV      VV  EE
          SS      TT      EE            VV  VV    EE
          SS      TT      EE            VV  VV    EE
  SSSSSSSS        TT      EEEEEEEEEE      VV      EEEEEEEEEE
  SSSSSSSS        TT      EEEEEEEEEE      VV      EEEEEEEEEE

                          WW      WW    OOOOOO    ZZZZZZZZZZ  NN      NN    IIIIII      AAAAAA    KK      KK
                          WW      WW    OOOOOO    ZZZZZZZZZZ  NN      NN    IIIIII      AAAAAA    KK      KK
                          WW      WW  OO      OO          ZZ  NN      NN      II      AA      AA  KK    KK
                          WW      WW  OO      OO          ZZ  NN      NN      II      AA      AA  KK    KK
                          WW      WW  OO      OO        ZZ    NNNN    NN      II      AA      AA  KK  KKKK
                          WW      WW  OO      OO        ZZ    NNNN    NN      II      AA      AA  KK  KKKK
                          WW  WW  WW  OO      OO      ZZ      NN  NN  NN      II      AA      AA  KKKK
                          WW  WW  WW  OO      OO      ZZ      NN  NN  NN      II      AA      AA  KKKK
                          WW  WW  WW  OO      OO    ZZ        NN    NNNN      II      AAAAAAAAAA  KK  KKKK
                          WW  WW  WW  OO      OO    ZZ        NN    NNNN      II      AAAAAAAAAA  KK  KKKK
                          WW  WW  WW  OO      OO  ZZ          NN      NN      II      AA      AA  KK    KK
                          WW  WW  WW  OO      OO  ZZ          NN      NN      II      AA      AA  KK    KK
                            WW  WW      OOOOOO    ZZZZZZZZZZ  NN      NN    IIIIII    AA      AA  KK      KK
                            WW  WW      OOOOOO    ZZZZZZZZZZ  NN      NN    IIIIII    AA      AA  KK      KK

  V   V  III   SSSS  III  TTTTT EEEEE DD          TTTTT H   H EEEEE        CCC   OOO  M   M PPPP  U   U TTTTT EEEEE RRRR
  V   V   I   S       I     T   E     D  D          T   H   H E           C   C O   O MM MM P   P U   U   T   E     R   R
  V   V   I   S       I     T   E     D   D         T   H   H E           C     O   O M M M P   P U   U   T   E     R   R
  V   V   I    SSS    I     T   EEEE  D   D         T   HHHHH EEEE        C     O   O M M M PPPP  U   U   T   EEEE  RRRR
  V   V   I       S   I     T   E     D   D         T   H   H E           C     O   O M   M P     U   U   T   E     R R
   V V    I       S   I     T   E     D  D          T   H   H E           C   C O   O M   M P     U   U   T   E     R  R
    V    III  SSSS   III    T   EEEEE DD            T   H   H EEEEE        CCC   OOO  M   M P      UUU    T   EEEEE R   R

              H   H  III   SSSS TTTTT  OOO  RRRR  Y   Y       M   M U   U  SSSS EEEEE U   U M   M
              H   H   I   S       T   O   O R   R Y   Y       MM MM U   U S     E     U   U MM MM
              H   H   I   S       T   O   O R   R Y   Y       M M M U   U S     E     U   U M M M
              HHHHH   I    SSS    T   O   O RRRR   Y Y        M M M U   U  SSS  EEEE  U   U M M M
              H   H   I       S   T   O   O R R     Y         M   M U   U     S E     U   U M   M
              H   H   I       S   T   O   O R  R    Y         M   M U   U     S E     U   U M   M
              H   H  III  SSSS    T    OOO  R   R   Y         M   M  UUU  SSSS  EEEEE  UUU  M   M

    JJJ  AAA  N   N U   U  AAA  RRRR  Y   Y       33333              222   000   000  77777
     J  A   A N   N U   U A   A R   R Y   Y          3              2   2 0   0 0   0     7
     J  A   A NN  N U   U A   A R   R Y   Y         3                   2 0  00 0  00    7
     J  A   A N N N U   U A   A RRRR   Y Y           3                 2  0 0 0 0 0 0   7
     J  AAAAA N  NN U   U AAAAA R R     Y             3  ,,            22 00  0 00  0  7
  J  J  A   A N   N U   U A   A R  R    Y             3   ,           2   0   0 0   0  7
   JJ   A   A N   N  UUU  A   A R   R   Y          333   ,          22222  000   000   7


                 TEXT BY ROBERT GARNER

                IBM 1401

THE IBM 1401 "DATA PROCESSING SYSTEM", INTRODUCED IN OCTOBER 1959, WAS THE "MODEL-T FORD OF THE COMPUTER INDUSTRY" AND "ONE OF THE
MOST IMPORTANT AND SUCCESSFUL PRODUCTS IBM HAD EVER ANNOUNCED." ABOUT 15,000 1401-FAMILY COMPUTERS WERE MANUFACTURED AND DELIVERED
WORLDWIDE IN THE 1960'S, FAR EXCEEDING INITIAL EXPECTATIONS. BY MARCH 1961, AFTER ONLY A SINGLE YEAR OF FULL PRODUCTION, 2,000
MACHINES HAD ALREADY BEEN DELIVERED TO CUSTOMERS. BY THE VERY NEXT YEAR, 1962, INCOME FROM THE IBM 1401 SURPASSED OLDER
CARD-PUNCH/UNIT-RECORD ACCOUNTING MACHINES FOR THE FIRST TIME. BY 1964, 40% OF ALL EXISTENT COMPUTERS WERE 1BM 1401-FAMILY
MACHINES. ITS FOLLOW-ON FAMILY COMPATABLE MACHINES, 1410, 1440 AND 1460, WERE SOLD UNTIL 1970. THE 1400 FAMILY WAS SO SUCCESSFUL
THAT IT LED TO THE INDUSTRY'S FIRST PROGRAM-COMPATIBLE COMPUTER CLONE, THE HONEYWELL H-200, OR "LIBERATOR."  A RUMORED 1480 WAS
DISCONTINUED IN FAVOR OF THE 360/25.

THE GOAL OF THE 1401 DESIGN WAS TO OFFER A FLEXIBLE BUSINESS COMPUTER OF AT LEAST THREE TIMES THE SPEED AND AT LOWER COST THAN THE
 UBIQUITOUS, CARD/UNIT-RECORD, COLLATE/MERGE/PRINT, PLUG-BOARD-BASED ACCOUNTING MACHINES OF THE 1950'S (SUCH AS THE IBM 407 AND
604). ALTHOUGH OFFERING A GENERAL-PURPOSE COMPUTER WITH PROGRAMMABLE MAGNETIC-CORE MEMORY WAS A COST RISK, BY CLEVER ARCHITECTURE
DESIGN AND USE OF MODERATE-SPEED, ECONOMIC TRANSISTOR CIRCUITS, THE GOAL WAS ACHIEVED. BY FIRST READING CARD/UNIT-RECORD
INFORMATION INTO THE MAIN MEMORY OF THE 1401 BEFORE MANIPULATING IT, COLLATE/MERGE/PRINT OPERATIONS WERE MUCH FASTER THAN ON THE
OLD CARD MACHINES. AND THE DATA COULD BE QUICKLY ACCESSED TO/FROM MAGNETIC TAPE FOR STORAGE  - MUCH FANTER THAN HANDLING DECKS OF
 CARDS (WHICH ALSO WORE OUT AND HAD TO BE REJUVENATED VIA DUPLICATION).

THE 1401 ARCHITECTURE WAS DESIGNED FOR EFFICIENTLY HANDLING BUSINESS DATA-PROCESSING APPLICATIONS. ITS CENTRAL PROCESSING UNIT
(CPU) CAN EFFICIENTLY MANIPULATE VARIABLE-LENGTH CHARACTER STRINGS AND PERFORM VARIABLE-LENGTH DECIMAL ARITHMETIC (UNLIKE TODAY'S
FIXED-WIDTH BINARY ARITHMETIC). FOR INSTANCE, THE CPU CAN OPERATE ON TWO NUMBERS WHERE EACH IS OF AN ARBITRARY LENGTH OR NUMBER OF
CHARACTERS. A 1401 CHARACTER IS ENCODED IN 6 BITS, AN END-OF-WORD "WORD MAKE" FLAG BIT, AND A PARITY CHECK BIT. NOTE THAT EVEN
THOUGH A CHARACTER WAS 8 BITS WIDE, THE TERM "BYTE" HAD NOT BEEN INVENTED YET.

1401 SYSTEMS COULD BE CONFIGURED WITH 1,400 TO 16,000 CHARACTERS OF MAGNETIC CORE MEMORY (4,000 MIN TYPICAL). MEMORY IS ADDRESSED
VIA DECIMAL CHARACTER STRINGS, NOT BINARY DIGITS AS IN NEARLY ALL CONTEMPORARY COMPUTERS. MAGNETIC CORES RETAIN EITHER A "1" OF
"0" DEPENDING ON THE CLOCKWISE OR COUNTERCLOCKWISE DIRECTION OF THE STORED MAGNETIC FIELD.

AN EXALTED FEATURE OF THE 1401 WAS ITS RELIABLE AND ROBUST INPUT/OUTPUT PERIPHERALS - A LONG IBM TRADITION INCLUDING AN 800-CARD
PER MINUTE CARD READER, A 250-CARD-PER-MINUTE CARD PUNCH, A 600-LINE-PER-MINUTE LINE "CHAIN" PRINTER, UP TO SIX REEL-TO-REEL TAPE
DRIVES, AND A 20-MILLION-CHARACTER MAGNETIC DISK STORAGE UNIT. THE SYSTEM AND PERIPHERALS CONTAINED CONSIDERABLE ERROR CHECKING
LOGIC.

THE 1401 MAY HAVE BEEN EQUIBALENT TO THE MODEL-T IN POPULARITY FOR IBM, BUT WAS FAR MORE COMPLICATED TO MANUFACTURE.  WHEREAS THE
MODEL-T HAS ABOUT XX COMPONENTS, A LARGE 1401 SYSTEM COMPRISES ABOUT 20,000 MECHANICAL PARTS AND ABOUT 50,000 ELECTRICAL
COMPONENTS (10,000 TRANSISTORS AND 14,000 DIODES ON 3,000 CARDS).

A TYPICAL, 4K-CHARACTER 1401 SYSTEM RENTED FOR ABOUT $7,000 PER MONTH IN THE 1960'S, EQUIVALENT TO $42,000 PER MONTH IN TODAY'S
DOLLARS (6X DUE TO INFLATION). THIS RENTAL FEE INCLUDED MAINTENANCE SERVICE AND OPTIONS FEATURES. A TYPICAL 1401 SYSTEM WOULD HAVE
COST ABOUT $370,000 IF PURCHASED OUTRIGHT IN 1961 OR $2,240,000 IN TODAY'S DOLLARS. MOST SYSTEMS WERE RENTED, A "CASH COW" FOR
IBM.

MAIN MEMORY ITSELF WAS VERY EXPENSIVE IN THE 1960'S: ONE 8-BIT CHARACTER OF 1400 CORE MEMORY IN THE 1960'S COST ABOUT FIVE DOLLARS
(OR $30 TODAY). THIS IS 300 MILLION TIMES MORE EXPENSIVE THAT TODAY'S COST OF MEMORY (ABOUT 0.1 MICRO-DOLLARS PER BYTE).

The card deck to run the above demo follows

,008015,022026,030037,044,049,053053N000000N00001026                   0001     
L068116,105106,110117B101/I9I#071029C029056B026/B001/0991,001/001117I0?0002     
,008015,022029,036040,047054,061068,072/061039              ,00100110400003     
000                                    L003089,040040,040040,04004010400004     
000                                    L003094,040040,040040,04004010400005     
000                                    L003099,040040,040040,04004010400006     
11111                                  L005205)201201,040040,04004010400007     
000N/0801MJ29/82MJ32337MU74U24         L030364,338339,343344,35135810400008     
M337089MJ32094M0'1S05MS05U!0AS51089    L035399,372379,386393,04004010400009     
AS51094CJ35094B431SCJ36S05B379/M089337 L038437,407414,419426,43104010400010     
B605MU74U24M337089MJ39094M0'1S05MS05U!0L039476,442449,456463,47004010400011     
AS51089AS51094CJ35094B515SCJ36S05B463/ L038514,484491,498503,51004010400012     
M089337B605MJ40/82MU99U24B605MV24U24   L036550,522526,533540,54404010400013     
B6051M025U24B605/332/,201              L025575,555556,563567,57157210400014     
,2661M0652651M0653302B580              L025600,580581,588589,59659710400015     
.601H/80MS47/98MS47/90MJ39S39/332/     L034634,605609,616623,63063410400016     
M/90089MU'0S05AS51/90MJ43S14CJ44S05    L035669,642649,656663,04004010400017     
B686/MJ45S13B870CJ46S05B709/MJ29S13B870L039708,675682,686693,69870510400018     
CJ47S05B732/MJ48S13B870DS05S14CS05S48  L037745,716721,728732,73904010400019     
B769/MJ49S13MS48S14B870CJ50S05B870U    L035780,751758,765769,77604010400020     
CJ51S05B800TAS52S13CJ52S05B819TAS52S13 L038818,788793,800807,81204010400021     
CJ53S05B838TAS52S13CCJ50S05B870SB870T  L037855,826831,838839,84685110400022     
MS47S14MJ36S05NMS14S31AS14S31AS14S31   L036891,863870,871878,88504010400023     
AS14S31AS14S31AS14S31AS14S31AS31S31    L035926,899906,913920,04004010400024     
MS31S14A/98S14A/98S14AS55S14MS39099    L035961,934941,948955,04004010400025     
MS14089MVS4S22MJ32094CSP1S22B'27T      L033994,969976,983990,04004010400026     
MS05S06SSP1S22DS22S69MS69S22B'34MJ36S06L039'33,'02'09,'16'23,'2704010400027     
AJ39094MS062?1AJ56099CJ40/82B'81S      L033'66,'41'48,'55'62,04004010400028     
MS062?1AJ56099CJ59094B983TAJ56099      L033'99,'74'81,'88'93,04004010400029     
CJ40/82B/19SAJ56099M099S39CJ62099B635T2L039/38,/07/12,/19/26,/33/3810400030     
CJ40/82B/52S2AS51/98CJ65/98B616T/332   L036/74,/46/51,/52/59,/66/7110400031     
/2B00000WCOL:000                       L016/90,/76/77,/81/82,/83/8810400032     
WROW:000PCHAR:00INDX:111               L024S14,/96/99,S05S06,S07S1210400033     
MASK2:00WORK3:000PCOL:000ZERO3         L030S44,S21S23,S29S32,S37S4010400034     
0000001100200516                       L016S60,S48S49,S52S53,S56S5910400035     
0804020101608                          L013S73,S63S65,S67S69,S70S7210400036     
04020100                               L008S81,S76S78,S80040,04004010400037     
0000000000000000000000000              L025U24,040040,040040,04004010400038     
                                       L025U74,040040,040040,04004010400039     
VISITED THE COMPUTER                   L025U99,040040,040040,04004010400040     
  HISTORY MUSEUM         00000000000000L039V38,040040,040040,04004010400041     
0412040404041414170102030431           L028V66)V39V39,040040,04004010400042     
3102040201011402061018310202           L028V94)V67V67,040040,04004010400043     
3116300101171406081630171714           L028W22)V95V95,040040,04004010400044     
3101020408080814171714171714           L028W50)W23W23,040040,04004010400045     
1417171501021200000000001212           L028W78)W51W51,040040,04004010400046     
1417171731171730171730171730           L028X06)W79W79,040040,04004010400047     
1417161616171424181717171824           L028X34)X07X07,040040,04004010400048     
3116163016163131161630161616           L028X62)X35X35,040040,04004010400049     
1417162317171517171731171717           L028X90)X63X63,040040,04004010400050     
1404040404041400000031000000           L028Y18)X91X91,040040,04004010400051     
0702020202181217182224221817           L028Y46)Y19Y19,040040,04004010400052     
1616161616163117272121171717           L028Y74)Y47Y47,040040,04004010400053     
1717252119171714171717171714           L028Z02)Y75Y75,040040,04004010400054     
3017173016161614171717212213           L028Z30)Z03Z03,040040,04004010400055     
3017173020181700000000120408           L028Z58)Z31Z31,040040,04004010400056     
0001020304160015161614010130           L028Z86)Z59Z59,040040,04004010400057     
3104040404040417171717171714           L028!14)Z87Z87,040040,04004010400058     
1717171717100417171721212110           L028!42)!15!15,040040,04004010400059     
1717100410171717171710040404           L028!70)!43!43,040040,04004010400060     
3101020408163114171921251714           L028!98)!71!71,040040,04004010400061     
1234567890123456789012345678901000012  L038J36,J29J30,J33J36,04004010400062     
00200  ,3.-                            L011J47,J40J41,J44J45,J46J4710400063     
24ZSJA001                              L009J56,J49J50,J51J52,J53J5410400064     
009132007                              L009J65,J60J63,040040,04004010400065     
                                       /338080                         0066     
STEVE WOZNIAK  
JANUARY 3, 2007
                                                                    1
                                                                    2
                 TEXT BY ROBERT GARNER                              1
                                                                    2
                                                                    1
                                                                    2
                IBM 1401                                            1
                                                                    2
                                                                    1
                                                                    2
THE IBM 1401 "DATA PROCESSING SYSTEM", INTRODUCED IN OCTOBER 1959   1
, WAS THE "MODEL-T FORD OF THE COMPUTER INDUSTRY" AND "ONE OF THE   2
MOST IMPORTANT AND SUCCESSFUL PRODUCTS IBM HAD EVER ANNOUNCED." A   1
BOUT 15,000 1401-FAMILY COMPUTERS WERE MANUFACTURED AND DELIVERED   2
WORLDWIDE IN THE 1960'S, FAR EXCEEDING INITIAL EXPECTATIONS. BY M   1
ARCH 1961, AFTER ONLY A SINGLE YEAR OF FULL PRODUCTION, 2,000       2
MACHINES HAD ALREADY BEEN DELIVERED TO CUSTOMERS. BY THE VERY NEX   1
T YEAR, 1962, INCOME FROM THE IBM 1401 SURPASSED OLDER              2
CARD-PUNCH/UNIT-RECORD ACCOUNTING MACHINES FOR THE FIRST TIME. BY   1
 1964, 40% OF ALL EXISTENT COMPUTERS WERE 1BM 1401-FAMILY           2
MACHINES. ITS FOLLOW-ON FAMILY COMPATABLE MACHINES, 1410, 1440 AN   1
D 1460, WERE SOLD UNTIL 1970. THE 1400 FAMILY WAS SO SUCCESSFUL     2
THAT IT LED TO THE INDUSTRY'S FIRST PROGRAM-COMPATIBLE COMPUTER C   1
LONE, THE HONEYWELL H-200, OR "LIBERATOR."  A RUMORED 1480 WAS      2
DISCONTINUED IN FAVOR OF THE 360/25.                                1
                                                                    2
                                                                    1
                                                                    2
THE GOAL OF THE 1401 DESIGN WAS TO OFFER A FLEXIBLE BUSINESS COMP   1
UTER OF AT LEAST THREE TIMES THE SPEED AND AT LOWER COST THAN THE   2
UBIQUITOUS, CARD/UNIT-RECORD, COLLATE/MERGE/PRINT, PLUG-BOARD-BAS   1
ED ACCOUNTING MACHINES OF THE 1950'S (SUCH AS THE IBM 407 AND       2
604). ALTHOUGH OFFERING A GENERAL-PURPOSE COMPUTER WITH PROGRAMMA   1
BLE MAGNETIC-CORE MEMORY WAS A COST RISK, BY CLEVER ARCHITECTURE    2
DESIGN AND USE OF MODERATE-SPEED, ECONOMIC TRANSISTOR CIRCUITS, T   1
HE GOAL WAS ACHIEVED. BY FIRST READING CARD/UNIT-RECORD             2
INFORMATION INTO THE MAIN MEMORY OF THE 1401 BEFORE MANIPULATING    1
IT, COLLATE/MERGE/PRINT OPERATIONS WERE MUCH FASTER THAN ON THE     2
OLD CARD MACHINES. AND THE DATA COULD BE QUICKLY ACCESSED TO/FROM   1
 MAGNETIC TAPE FOR STORAGE  - MUCH FANTER THAN HANDLING DECKS OF    2
CARDS (WHICH ALSO WORE OUT AND HAD TO BE REJUVENATED VIA DUPLICAT   1
ION).                                                              2
                                                                    1
                                                                    2
THE 1401 ARCHITECTURE WAS DESIGNED FOR EFFICIENTLY HANDLING BUSIN   1
ESS DATA-PROCESSING APPLICATIONS. ITS CENTRAL PROCESSING UNIT       2
(CPU) CAN EFFICIENTLY MANIPULATE VARIABLE-LENGTH CHARACTER STRING   1
S AND PERFORM VARIABLE-LENGTH DECIMAL ARITHMETIC (UNLIKE TODAY'S    2
FIXED-WIDTH BINARY ARITHMETIC). FOR INSTANCE, THE CPU CAN OPERATE   1
 ON TWO NUMBERS WHERE EACH IS OF AN ARBITRARY LENGTH OR NUMBER OF   2
CHARACTERS. A 1401 CHARACTER IS ENCODED IN 6 BITS, AN END-OF-WORD   1
 "WORD MAKE" FLAG BIT, AND A PARITY CHECK BIT. NOTE THAT EVEN       2
THOUGH A CHARACTER WAS 8 BITS WIDE, THE TERM "BYTE" HAD NOT BEEN    1
INVENTED YET.                                                       2
                                                                    1
                                                                    2
1401 SYSTEMS COULD BE CONFIGURED WITH 1,400 TO 16,000 CHARACTERS    1
OF MAGNETIC CORE MEMORY (4,000 MIN TYPICAL). MEMORY IS ADDRESSED    2
VIA DECIMAL CHARACTER STRINGS, NOT BINARY DIGITS AS IN NEARLY ALL   1
 CONTEMPORARY COMPUTERS. MAGNETIC CORES RETAIN EITHER A "1" OF      2
"0" DEPENDING ON THE CLOCKWISE OR COUNTERCLOCKWISE DIRECTION OF T   1
HE STORED MAGNETIC FIELD.                                           2
                                                                    1
                                                                    2
AN EXALTED FEATURE OF THE 1401 WAS ITS RELIABLE AND ROBUST INPUT/   1
OUTPUT PERIPHERALS - A LONG IBM TRADITION INCLUDING AN 800-CARD     2
PER MINUTE CARD READER, A 250-CARD-PER-MINUTE CARD PUNCH, A 600-L   1
INE-PER-MINUTE LINE "CHAIN" PRINTER, UP TO SIX REEL-TO-REEL TAPE    2
DRIVES, AND A 20-MILLION-CHARACTER MAGNETIC DISK STORAGE UNIT. TH   1
E SYSTEM AND PERIPHERALS CONTAINED CONSIDERABLE ERROR CHECKING      2
LOGIC.                                                              1
                                                                    2
                                                                    1
                                                                    2
THE 1401 MAY HAVE BEEN EQUIBALENT TO THE MODEL-T IN POPULARITY FO   1
R IBM, BUT WAS FAR MORE COMPLICATED TO MANUFACTURE.  WHEREAS THE    2
MODEL-T HAS ABOUT XX COMPONENTS, A LARGE 1401 SYSTEM COMPRISES AB   1
OUT 20,000 MECHANICAL PARTS AND ABOUT 50,000 ELECTRICAL             2
COMPONENTS (10,000 TRANSISTORS AND 14,000 DIODES ON 3,000 CARDS).   1
                                                                    2
                                                                    1
                                                                    2
A TYPICAL, 4K-CHARACTER 1401 SYSTEM RENTED FOR ABOUT $7,000 PER M   1
ONTH IN THE 1960'S, EQUIVALENT TO $42,000 PER MONTH IN TODAY'S      2
DOLLARS (6X DUE TO INFLATION). THIS RENTAL FEE INCLUDED MAINTENAN   1
CE SERVICE AND OPTIONS FEATURES. A TYPICAL 1401 SYSTEM WOULD HAVE   2
COST ABOUT $370,000 IF PURCHASED OUTRIGHT IN 1961 OR $2,240,000 I   1
N TODAY'S DOLLARS. MOST SYSTEMS WERE RENTED, A "CASH COW" FOR       2
IBM.                                                                1
                                                                    2
                                                                    1
                                                                    2
MAIN MEMORY ITSELF WAS VERY EXPENSIVE IN THE 1960'S: ONE 8-BIT CH   1
ARACTER OF 1400 CORE MEMORY IN THE 1960'S COST ABOUT FIVE DOLLARS   2
(OR $30 TODAY). THIS IS 300 MILLION TIMES MORE EXPENSIVE THAT TOD   1
AY'S COST OF MEMORY (ABOUT 0.1 MICRO-DOLLARS PER BYTE).             2
                                                                    1
                                                                    2

Sat Jan 20 - general

Wed Jan 24 - general

Thu Jan 25 - Tape Team

TAU Debug Status - 1/25/07 (Grant and I)

Today we continued to troubleshoot intermittent errors in tape 
operations. Again we erased an entire tape and read it via TAU CE Panel 
operations. The most common error was still phantom bits occurring on 
the Read Bus A-bit during the read operations. The TAU indicators were:
Error - on
A-Reg VRC Error - on
R/W Error - on
        CBA8421
A-Reg   0000000
B-Reg   0000000
R/W-Reg 0010000
The noise bursts were similar to that found last week, but errors were 
less frequent. The noise was traced back to the skew adjustment Lumped 
Delay Line Card (BJ__) in the Read pre-amp circuit. The TAU samples the 
Read Bus only during the "Read Cond" phase of tape operations, but 
random noise bursts were visible in the 729 even when no tape operations 
were in progress.

For this bit, the delay was jumpered to its minimum. That is, the output 
jumper wire was connected directly to the card's input pin. Since, other 
than some circuit loading, the card should not have been participating 
in the circuit, I removed it. The noise disappeared.

We replaced the card with one from the other 729-V drive. No noise.

The BJ card contains no active components. Grant gave the card a cursory 
examination. There were no obvious failed components, but the noise 
bursts resumed when the card was plugged back into the tape drive. The 
card is now in the "Bad Cards" box.

With the replacement card, we were able to read the entire blank tape 
without error. We also wrote data records to the entire tape without any 
read-back errors. (Three bit patterns were used.) When read back, two 
errors occurred. These were probably at the points where I flipped 
switches to change the bit pattern. These results are very encouraging.
 
We were unable to run tape system diagnostics due to the broken card reader.

Other Tape Drive Problems:
The take-up capstan-in switch needs adjustment. High-speed rewind 
operations require that someone walk over to the drive and tickle the 
switch to make the drive unload the head. It seems that the switch 
magnet is too far away and frequently does not detect the capstan being in.

Repeated rewinds eventually pop the mercury switch out of its clip on 
the take-up reel shaft. Other drives have a non-slip rubber tube pressed 
over the switches glass bulb to prevent it from sliding out of the metal 
clip. When the mercury switch is out of its clip, rewind operations hang 
on the transition of high-speed to low-speed.

I did not log the above 729 problems, because the logbook for the 
functional 729 is missing. Allen, do you have it?

Other Problems:
About eight ceiling fluorescent tubes over the CPU and tape drives are 
burnt out. It would be a lot easier to see what pins we are probing if 
they were working. Can we get them replaced? I reported them in December 
and again today, but follow-up will probably be needed.

Regards,
Bob

---

Could be a bad inductor or capacitor. Whatever it is,
the noise bursts were intermitent. Bursts occured
about every 5 minutes. Maybe the card was tuned into
some other noise source in the room.

Regards,
Bob

--- Robert B Garner  wrote:

> > These results are very encouraging.
> 
> Yes!
> 
> > The BJ card contains no active components.
> 
> What the heck does it do then?  (Like "junk DNA?" 
> ;-)
> Perhaps problem is a bad cap?
> (We've also had faulty inductors...)
> 
> - Robert

Comment by Ed Thelen (Ron Williams does not do e-mail)

How interesting !!! The misreads by the 1402 are also bursty, like non-random noise,
see above.