Stanford Artificial Intelligence Laboratory Records

Observation:

1) Errors increase rapidly as size of area read increase.

2) Misses appears to be somewhat corelated with system activity. Total [activity?] was 1 [bip?] program and 2 music programs. [Wast?] misses (last amount read) seemed to occur when someone typed.

3) Parity errors may be same ones we have seen for several days when programs seen in ampex [?] and not a result of using TV. There were no parity errors found by 167.

4) Error rate for size of rectangles read by edge follower appears to be around the limit we can tolerate but we can live with it if necessary. MICROSCAN users may be unhappy when reading large frames. It appeared, although I have no statistics, [like?] the misses were fairly evenly distributed through the read instead of banded together. I could test this if desired.

[Pingle?]

1/30/68 9:30

Readjusted format tape lights for Analex printer to try to stop it from skipping pages. It will not fail for me, but time will tell.

Tod [Pagliz?]

Feb. 2, 1968 0900

System died at 0900. Had a NONEX MEM AND A PARITY ERROR. Could NOT RESTART AT 140, but was able to restart at 143.

J. Gleason

Note: A Humidity meter has been installed nest to the air conditioning controls behind the disk. To read it, first subtract the reading of the wet-bulb thermometer from the dry-bult thermometer; Then set the red number on the top of the conversion dial to this temperature diffrence. Read the humidity on the dial across from the dry-bulb thermometer.

The temperature difference this morning is 72° = 60° = 12°. The humidity reading across from 72° is 49%.

J. Gleason

1020

TTY 7 exhibited the "bad guy" bug; i.e. it would respond to ↑C, but not to other characters. When this occurs the system usually has to be restarted at 140. However, it was empirically determined that ↑M restored control this time!

J. Gleason

1200 Scratch Tape 140 seems defective (or DTA 1 is N.[G?])

L. Smith

14:15 DTA1 DID NOT READ TAPE PROPERLY GIVING "BINARY OUTPUT ERROR" IN F[?] COMP. SWITCHED TO DTA3 and problem went away

D. [Prepis?]

0015 5 Feb 68

Tried 3.15/05. Found 533 bug [still?] strikes, about as frequently as before. I suspect that the adjustment of R/W current had no effect, but I can't be sure.

SRR

3.15/05 with 2 copies of 40K BLT ST runs with core 3 IF & 1H + 10 A & B margins from [10?] V low to 1 V low. fails occasionaly at zero. Leave the margins supply at 5 V

SRR

1000 - 5 Feb 68

We, Leland Smith & I, spent 15 minutes [?] trying to see what is wrong with audio-output only to find that the hardware log says 'it is down'. Would it be possible to have a prominent display of the things that are not working so that every one can know the state of the system with looking through lot of hardware log failures

Raj Reddy

Feb. 6, 1968

We apparently have two bad balun boards. One (dated 2/5/68) says "SECT 4 Loses." The other (dated 2/6/68) says "SECTION 13 loses." I checked all of the diodes, balun transofmers, and termination resistors on both boards. The board labeled "SECTION 4 LOSES" had an open eyelet on pin 1 (which connects to Sect. 4). This has been resold and tested with a meter. It still must be tested on the machine. I couldn't find anything wrong with the board labeled "SECTION 13 loses"; but I noticed that its contacts have not been erased. We'll have to find out what "SECTION 13 loses" means before we can look further for the trouble.

J. Gleason

1630 6 FEB 68

ALL DATAPHONES & TWX interaces don't work on REMOTE → COMPUTE TWX [originals?] get to output of 4504 but not into the 630 line scanner Local TTY's still work.

?

SRR

1700 6 Feb 68

Replaced [Y?] sink driver in AMP 1 Mag Module G JG. It had not been touched since Sat night. The final version of the bug was that no locations in MAG MOD G didn't [shook?] Deposit of all 1's gave back almost all zero's. Current [?]forms looked fine.

SRR

0130 7 Feb 68

TROUBLE WITH 630 not talking to data sets was a subtly intermittent patch cord. When patching in a data set like so: [diagram] it was open on the [sleve?] lead. When patched to the local teletypes - like so: [diagram] it was conducting.

SRR

0910

I tried to start a job over the 1799 TWX line and was unable to do anything except "handshake" with the 1799 Dataphone. I was able to TAIK to the TWX TTY, but the 630 did not receive characters from it.

J Gleason

930 7 Feb 68

The TYPE 30 turns itsself off (the display not power) itermitantly. There have been many entries made about it for over a week. It needs to be fixed as soon as possible

Pingle

1045 Feb. 7, 1968

Turned off machine for ~2 min to switch off Ampex CORE 0 so that Phil McKay could work on it. Restarted at 143.

Note: This morning we have had 4 Core Parity Errors - all of which were in the Ampex Core. Locations 375735, 202073, 221565, & 340657.

J. Gleason

2:00 7 Feb 68

Fixed bias supply to A-D speech input.

Phil McKay of Ampex installed [?] pair on AMPEX #1.

SRR

0033 8 Feb 68

While chasing disk bugs we found that running chess in a certain region of AMpex core seemed to drop bit 35 and get parity errors.

To get we did following

R FAILSAFE R CHESS * PS

got parity errors all with data = 777777777776 at PMA = 63506, 63514, 63502, 63572 63636, 63630, 63506, 63556, 63504, 63506 63514, 63572, 12766, 63506, 63514, 63572 63572, 63622. (FULL INTERLEAVE]

Replaced B28 + B3, B205, R181 for low order 2 [bits?] of MA for 167 fixed reread problems with 167 in AMpex

Traded 6 with 7

Bit 6 (2 De 11 was microphonic, Traded it with Bit 7 (2 DE 12 ). Machine worked.

Run margins up and down Banged on boards, etc. bugs didn't Found that wiping core 3 if 485 boards hard caused bits 3 & 4 to be c p ok split cycles. Couldn't make it go away, Checked sockets, changed cards, pinched connectors, no trouble formed troubles not there, Probably not related to " 033 bug " not related to " 033 bug ". [SRP] 26 Jan . Type 30 goes away randomly when viewing it. [--?] will come back again when [reinitalized?]. can't tell if Hardwood or softwood [Pingle?]

12 00 1/26/67 [D+H?] Drives 6 and 7 now have guides with no avoiding on the surface. After instalation ran Dectape Test with a static tape. No [Erross]

Fred Pa------?]

(Pg.58) 48

Traded 6 with 7.

Bit 6 (2DE 11) was microphonic traded it with Bit 7 (2 DE 12). Machine worked

Run margins up and down Banged on board, etc. bugs didn't return.

Found that wiping core 3 if 4 & 5 boards hard caused Arts 3-4 to be dropped when running disk & having C P do split cycles. Couldn't make it go away. Checked sockets, changed cards, pinched connectors. no trouble found trouble not there. Probably not related to " 033 bug ". machine back at 0030 SRP 26 Jan.

Type 30 goes away randomly when running it. [--?] will come back again when [reinitaslized?] can't tell if hardwere is softwere Pinole 12:00 1/26/67 [D + h ?] Drives 6 and 7 now have guides with

no anodizing on the surface. After instalation

ran DECTAPE Test with a static tape. No Errors.

[Ted Pa------?]

the first hole in the vacuum column I don't think it used to do this. In fact my remembrance is that it doesn't have vacuum on in high speed rewind. SRR

1200 Feb. 14, 1968

Data input errors on mag tape. Magtst claims everything is fine. Suspect that current magtape unit is not compatible with the one that existed 2 days ago. (may be running magtst for 1 hr knocked it out of alignment) B.P.

1200 14 Feb 68

(1 = *1 Type 30 is sick again. RPG

The +10V reg supply (1562 module) was putting out +11.5V and not regulating. After adjusting it for +10.0V, the trouble R.P.G. described disappeared.

1000 Feb 28

System put down and kluge wires in Ampex interface replaced with wire-wrap. After one wiring error and one bad wrap formed, system put up at 12:45

[Jack Pa?]

2400 29 Feb 68

2.15 has been [elubriting?] a bug not seen with either DECTAPE TEST or 2.8. This was determined to be the 4113 that and [?] & flags in the 551 at 1C15 wasn't going all the way to ground. It also turned out that 2 of our spare 4113s had problems. 1 was mysterious and I had the same problem in a different section. Also replaced PI encoder [it?] 1C8, but it seemed it was OK after all. SRR