IBM 1130 Restoration at TNMOC in 2012

01 January 2012 Update from Stewart Baker

• Worked on IBM 1442 to get it to read cards. Adjustments made to the card skew at the Stacker Cornering Station appear to have been successful, no further jamming.
• Occasional failures to load first card cured by lowering the light level (lamp voltage). Read channels roughly set up.
• Repeatable errors when trying to read cards commanded from IBM 1131. Read station, Punch Station and Feed clutch errors. Also CPU Interrupt 4 which appears even before the first card is loaded. Further investigation needed.

05 January 2012 Update from Stewart Baker

• Console Printer: Further attempt to get to the bottom of the continuous CR/LF problem. Bowden cables and linkages for 3 front panel switches checked. Further removal of stale grease. No improvement. On hand moving the carriage away from the LHS found that Tab pivot assembly had shaken itself loose. Special washer found, but not the retaining nut.
• IBM 1442: Tried to run cards using SS/Load, still gives the 3 errors. Checked and adjusted light source to read emitter. Found that timing of Card Feed Clutch appears to be way out (100mS instead of 20ms).
• Need to clean and re-lubricate Card Feed Clutch assembly. Read Clutch timing seems OK.

07 January 2012 Update from Peter Vaughan

• Cleaned and re-lubricated Card Feed Clutch, Read Clutch and various cams and wheels in both the front and rear mechanisms. During this activity a 3 inch rod was found loose within the mechanism. There was no obvious place where this came from and the thought was it is one of the temp engineer aligning pins used to align the reader photo transistor unit which may have been dropped in the past
• The 1442 was powered on and tried. While the first card fed into the reader station OK, it would not move into the punch station, in fact the result was 2 cards fed into the read station and pressing NPRO did not feed the cards through. On closer inspection the card feed mechanism was not activating and eventually the location of where the rod was used was found (used to activate the card feed cam to move the card from reader to punch station), the rod re-fitted (very difficult) and on re-testing the card fed correctly into the punch station. The rod probably fell out during manual activation on the read feed cam mechanism.
• However, this has made no difference to the problem with the same int 4 being produced when 1442 first feeds the first card and the same errors when trying the read the first card. As I did not have a scope or feeler gauges I could not check any alignments or timings so one for Stewart when he is in next.
• Other observations:
  • Pressing the NPRO key after initially powering on the unit also generates int 4 so the error is pretty permanent.
  • Sometimes when pressing and holding NPRO, the punch error light flashes dimly. The problem may actually be in the punch mechanism.
  • Frequent clonking noise heard previously is getting louder so cause needs to be found. It's louder near the card stacker mechanism near the bottom - there is some movement of the whole mechanism when the clunk is heard.

12 January 2012 Update from Stewart Baker

• Clonking sound traced to main bearing on Joggler shaft in stacker assembly. Shaft and associated components removed from machine, cleaned and re-lubricated. Forcing new grease into bearing together with setting correct alignment/tensioning of belts cured the problem. To minimize the effects of the combined indentations in the stacker feed rollers each set of rollers has been repositioned with an incremental 90 deg rotation.
• Feed clutch assembly (except for helical spring) dissembled, cleaned, lubricated and reset as per SRP FEMM specifications. The clutch pawl clearance was too tight and had been rubbing. Greasing of Feed and Read clutch nipples needs to be done with a grease gun to force any remaining grease out and insert new. This may help with the helical spring performance
• Clearances of CB1-4 to magnet on timing disk checked out OK. There is however, a discrepancy between the timing information in the FEMM and the timing disk. In the manual it says that the magnet aligns with CB1 @ 15 deg whereas the disk shows nearly 20 deg. There is no adjustment for this
• Timing of Feed Clutch Select to CB1 is now down from 120mS to 60mS, but still outside the FEMM limit of 24mS. The 3 errors when reading a card remain.

14 January 2012 Update from Peter Vaughan

• Feed & Read clutch nipples re-greased with grease gun, forcing out old grease and excess cleaned. Also lubricated some of the other bearings in-place with grease this time where accessible. An initial NPRO feed of cards caused a misaligned feed in stacker and one card went too far and wrapped around the end sensor. Subsequent feeds OK so may need to keep an eye on this.
• Concentrated on the punch mechanism; cleaned and re-greased various cams, linkages and bearings in-place, front and back, as well as another grease nipple below top cam. Also greased all stacker bearings and drive belt pulleys and belt tensioners where accessible.
• Problem still exists after the above.
• While observing the punch mechanism I noticed one area that never moves - This is the punch read pressure rollers (SRP FEMM diag 4-24) - this is not a fixed mechanism as it has several bearings holding it in place. Pressing the lever on the end of the punch mechanism raises 2 smaller rollers which normally engage with the 2 large wheels on the read pressure roller. The front small roller did not move so re-lubricated to get working again. I suspect the rear smaller roller may also be in the same state but this is not accessible without taking the punch mechanism out and apart. There was also no obvious way the read pressure rollers were driven but the only place must be in the front incremental drive unit (SRP diag 4-50). There are 2 pawls with a red dot (just visible from the front of the punch mechanism) which engage either side of the ratchet wheel but no sign of the ratchet wheel moving. This needs further investigation as to what it actually does and whether it should actually move. This may need the removal and stripping of the punch mechanism to progress this but it would then need the mechanism to be re-aligned.

18 January 2012 Update from Stewart Baker

• Drive belt removed to isolate Card Feed assembly. from the rest of the drive chain. Card Feed Main Shaft can now be rotated easily with minimum drag. Attention shifted to the remainder of drive chain. Although there is quite a lot of friction due to transport belts the rest of the drive chain seems OK.
• Punch unit removed for inspection on the bench. When the unit drive pulley is manually rotated there is quite a noticeable notching feel, however this is caused mainly by the Restore, Interposer, and Punch Cam actions. No anomalies such as sticking bearings could be found, so only lubrication of previously unreachable parts was undertaken.
• The Punch unit was replaced and all the belts checked for correct tensioning. The Picker Knife belt was adjusted to track correctly on it's idler pulley. The unit still gives the same 3 errors with an occasional Feed Clutch error using NPRO feed.
• The timing between the Clutch Feed and CB1 is still around 60-70mS. It seems that no further improvement is to be had from mechanical cleaning, lubrication and adjustments.
• Triggering on the Clutch Feed signal I looked at the waveform on the Feed Clutch Solenoid. It takes quite a time ~20mS for the current to build. I would have expected to see a steeper drop as the driver turned on. The suppression diode across the solenoid checked out OK. Looking at the +12v supply on the other side of the solenoid showed that the voltage available dropped quite rapidly when the Feed Clutch signal was applied. It finishes up at about +6V when the unit is feeding continuously. The +12v is supplied from TB2-8 on the PSU. The circuit shows R4 (10ohms) which serves to prevent the Clutch solenoid from overheating and C4 (4000uF 15v) which acts as a reservoir to stiffen the output when short bursts of current are required.
• On my initial inspection of the 1442 I had found that C4 was open circuit and so, having no computer grade capacitor to hand I bridged the open circuit capacitor a new 4,700uF 63V one from Maplin. Maybe this capacitor has either a) a low capacitance or b) a highish internal resistance. If the capacitor cannot deliver the surge of current needed when the the Feed Clutch is requested then the operation of the solenoid may be slowed down.
• A way to establish if this is the problem is to bridge the existing capacitor with a larger value computer grade capacitor. I spotted a number of these in H24 while I was searching for the original replacement.

15 January 2012 Update from Peter Vaughan

• Further research shows the read pressure rollers & ratchet is used during punch operation so would not move during a normal read or NPRO operation.

26 January 2012 Update from Stewart Baker

• Bridged existing capacitor with computer grade 4,700uF 63V capacitor. Although +12V on TB2-8 remains higher there was no observable effect on the Feed Clutch response time.
• Drive motor slip ring/governor assembly cleaned and lubricated. No longer makes rubbing sound at motor start up. Motor contactor cleaned.
• Raw and shaped signals from CB1, CB2, CB3 and CB4 observed on the scope. All sensor Raw signals have the same amplitude except the first pulse after the Feed Clutch select from CB1 which is about 50% lower in amplitude. However, even though this pulse is quite a bit lower than the others it is still enough to generate a shaped pulse.
• The shaped sensor pulse widths are about 2 to 2.2mS which is within specification with the exception of the first CB1 pulse which is about 3mS wide. This is in line with poor acceleration of the clutch feed assembly. The total cycle time is 190mS for 360 deg.
• It maybe that the motor does not have enough time to get up to speed before the Feed Clutch select arrives, or we still have too much drag on the Feed Clutch assembly. It might be a good idea to investigate the helical spring in that assembly.

01 February 2012 Update from Stewart Baker

• Different tack. Investigated why 1442 motor does not time out. Unit appears to be busy. Signal tracing showed -SRP Busy and -Punch SRP signals active. After much signal tracing gate on SLT card A-B1H3 (type 3794) was found allowing Incur signals through. This was proved by swapping with another SLT card type 3794 in A-B1H4.
• 1442 now loads/feeds cards correctly, does not give INT 4, permits Reset and Program Load, and times the motor out after 15 seconds of inactivity.
• Cards need to be swapped back and the defective one in A-B1H3 replaced with the spare we have. The unit should be then ready for alignment of the read channels etc.
• 1442 open circuit capacitor C4 removed and replaced by computer grade 4,700uF 63V part.

04 February 2012 Update from Stewart Baker

• 1131 previously swapped cards swapped back and fitted replacement type 3794 card in A-B1H3 and found to work OK. Defective spare labeled with fault symptoms.
• 1442 now able to feed cards continuously using FEMM Single Step mode & Program load switch but cloning back. Jogger shaft in stacker assembly removed and examined, NFF. After reassembly and adjustment noise gone. Need to monitor.
• Started alignment of read light source. As per FEMM instructions Hopper switch bypassed to permit continuous feeding of cards using the NPRO switch.
• Occasional card jams, mainly in the Stacker Cornering Station but also in the hopper feed. Cleaned transport & hopper feed rollers. Card jams getting less frequent with exercising the unit. At end of day, 1442 was able to feed multiple stacks of cards with occasional error. Things may need to settle down.
• Feed clutch errors still occurring at times with no obvious reasons so need to investigate why.

10 February 2012 Update from Stewart Baker

• Card jams in all parts of transport path. Before feeding any cards 1442 must be allowed to warm up for a reasonable period of time, especially in cold weather. It should then be exercised with NPRO card feeds.
• Further cleaning and lubrication carried out, particularly in the stacker transport/hopper areas.
• Signal levels of read channels examined. Channels 0 and 1 are the lowest being about 1.25V below the others. Read light source voltage adjusted as per FEMM. The lamp terminal voltage is 3.6V.
• It maybe beneficial to carry out mechanical skew alignment of the read assembly . This involves the use of 2 locating pins which we should be able to fabricate (see FEMM).

12 February 2012 Update from Peter Vaughan

• Initially warmed up 1442 by feeding increasing number of cards until at least 100 would run through without issue so all seems fine. Then set-up to check reader photo-transistors to compare with Stewart's finding during the week. Other than one which was higher for me the rest were the same voltage (my findings left with your card Stewart).
• Then decided to feed blank cards through while watching channel 0 on the scope but I had a card jam in the hopper caused by the right-angle stacker card feed getting jammed. After removing several cards I repeated the loading but the cards got jammed in the right-angle feed mechanism again. On doing a manual card feed sequence using the wheel, it looked like the right-angle stacker pushers were not pushing the cards far enough to get grabbed by the feed wheels. I cleaned the first set of feed wheels and got a lot of ink off them, I then checked the pusher alignment and this was correct and could not be moved any further forward. I then looked to see if the feed rollers could be moved towards the pushers but again no adjustment could be made. I tried running cards through normally and things appeared to be working fine again, then realised when the pushers were operating at normal speed they would have a lot more momentum so would push the cards further... so this was not a problem.
• So, while cards seem to be feeding OK again, the stacker collection bin is not. I don't know what I disturbed but the stacker bin is no longer allowing cards to drop properly - maybe the 2 wires got moved when I was removing cards originally, but this needs to be resolved before cards can be fed through again.

23 February 2012 Update from Stewart Baker

• Pushers on Stacker Cornering Station too far forward. Assembly removed and inspected. Adjustment made to pusher rod. Clearance on Punch exit roller too tight, adjusted. Card skew adjusted. Found cards intermittently binding on Stacker Cornering Station plastic cover. Adjusted clearance.
• Occasional card jam. Believe this is caused by feeding damaged cards. Loose cards inspected and damaged ones removed. This problem may occur if we keep running the same cards through the 1442.
• Fitted Hopper Override Switch. This allows the hopper micro switch to be bypassed permitting NPRO card feeding for adjustment/alignment etc.
• Internal clean of upper section of 1442 including wiring and TB's at front of machine.
• Following a suggestion by Adam, removal of sound proofing material residue using IPA was tried. With one application of IPA a fair amount of residue was removed from the inside of the lower front cover. However, it is a very messy process and uses lots of IPA.

16 February 2012 Update from Stewart Baker

• Found that the shank of 3mm twist drill was just the right size to use as a pin for the read assembly skew alignment.
• Restraining Springs in collection bins cleaned and adjusted as per FEMM.
• Although cards hang up in the collection bin the problem starts earlier on at the Stacker Cornering Station. The cards are travelling skewed after leaving the station, and are entering the collection bin edge on.
• Manual card feeding into the Stacker Cornering Station shows that they are frequently riding up on the top of the first pusher. This is causing the card to skew when entering the transport path, and in some cases jamming the following cards.
• As it is likely this has been causing the 'train wreck' type of card jam we have experienced, I believe we should concentrate on this area, paying special attention to mechanical adjustments as detailed in the FEMM.

27 February 2012 Update from Peter Vaughan

• Cornering station still causing problems. Various adjustments made to card pushers and plastic cover. While this has improved things we are still getting feeding problems. First set of feed rollers may be having problems picking up the card due to not enough friction or excessive wear so will need further investigation.
• Stacker wire 'catchers' adjusted but still having problems with cards not falling properly into stacker resulting in not being able to feed many cards sequentially before the stacker gets congested. Need to figure out why this is happening as it was OK a few weeks ago. May still be related to misaligned cards entering stacker at the wrong angle due to initial pick-up by rollers. The 1442 is not reliable enough to try any multi-card diagnostics card decks yet which will limit what we can do in testing the rest of the system. This also means we can't try any card copy/punching either which limits what we can do.
• Decided to leave the feeding problems for another time and attempted to read our own copy of the one card diagnostics to test that part of the card reader and 1131 CPU - these cards are read in one at a time so the feeding issues don't affect them. To our surprise, each card was read in and ran successfully with no reading errors. We were able to read and run 6 out of the 7 one card diagnostic we have (card 5 could not be copied previously). This has proved that the cards can be read successfully by the card reader, the data transferred to the CPU core store and run successfully.
• Attempted to copy the one remaining diag card (card 5) on the 029 card punch, but as before this kept failing with the newly punched card often getting stuck in the punch. Attempted to lubricate the punch pins but this proved difficult and no obvious improvement when trying to copy cards after the cleaning. May need to look at removing the punch mechanism to clean properly but this will involve removing a lot of inter-linked mechanisms so not an easy job.

29 February 2012 Update from Stewart Baker

• Following success in removing some of the sound proofing residue from the 1442 cover I tried the same approach on other covers. Unfortunately this did not work anywhere near as well. Whether it is the composition of the grey paint, or more adhesive, it was unsuccessful. Need to try other solvents.
• Temporary installation of step-down autotransformer to provide 220v AC for the 1131 mains input. After a days running of the 1131 and 1442 NPRO'ing the transformer was only slightly warm. The transformer now needs to be switched, suitably fused and enclosed in a ventilated metal box.
• Investigation into the occasionally skewing of cards leaving the Stacker Cornering Station. I found that the pressures exerted by the two First Stacker Feed Rollers were unequal. The LHS roller (looking from the 1442 front was about 50% of the RHS roller. The LHS first lower stacker roll shaft bracket was adjusted to give a pressure equal to that of the RHS roller. They were then checked against the FEMM (4-30) limits.
• After this and slight adjustments to the Stacker wire 'catchers I was able to NPRO feed card stacks of 100 or so cards many times without a single jam...

03 March 2012 Update from Stewart Baker

• A number of card stacks were NPRO'd through the 1442 without any jams occurring.
• About 2 hours after the IBM equipment was powered on a burning smell was noticed in the Large System Gallery. This was traced to the 1130 which was immediately powered down. The smell was coming from the 1131 B-Gate.
• Investigation showed that the middle fan of the 1131 B-Gate was not running, although the fan could be easily turned by hand.
• Peter and I removed the fan tray and ran the three fans from the normal mains supply. All fans ran OK. The fan tray was reconnected to the 1130 and the supply voltage to the tray measured. It was around 190V. Although the middle fan was now running, as the fans are 230v ones it is possible that particular fan had stalled and overheated due to the low voltage applied.
• The supply to the system from the autotransformer was measured as 205V instead of the desired 220V . The mains input voltage to the autotransformer was measured as 226V. So the system was being starved of volts. Measurements on a weekday showed 240V input and 217V output.
• The fan tray was replaced in the system (not an easy job). The system was just about to be powered up using the Variac when a phase went in the museum supply, and "Rain stopped Play" !
• It would appear that the museum mains supply can vary considerably depending on the load applied. For the time being the only safe way to power the 1130 is via a Variac where the voltage supplied to the system can be set independent of the mains voltage.

07 March 2012 Update from Stewart Baker

• Further to stalled fan incident examined area B-B2. No obvious signs of any heat stress, However card in J2 found to be seated proud of others. On investigation found jammed/broken socket in board connector. Board type 4619 fitted from our spares.
• Started measuring and recording mains supply (off load) at end of the orange extension lead. Most of today (Wednesday) it was around 243V. It will be interesting to see what it is on Saturday.
• 1442 now feeds cards fine. Many packs of blanks NPRO'd without any jams.
• Tried REPRO card copying process. Cards could be read in one at a time by pressing start, but no punching action took place. Compared memory contents with the card, and found a number of data errors. I suspect that there are pattern issues with the memory. We should run previous diagnostics to check that no new hardware problems have occurred as a consequence of the fan issue, and then investigate the memory.

11 March 2012 Update from Stewart Baker

• The mains voltage measured at the extension lead (off load) was... 241V Saturday AM, 232V PM.
• The 1130 was run most of the day with CPU Gates open. Good airflow, with no signs of overheating taking place.
• Reviewed REPRO card suspect memory errors. Determined that they were false, as many of the IBM single cards RUN after being read and overwrite lower memory. Setting the machine to DISP when reading in a card stops it from self running.
• The complete set of One Card Diagnostics run without any errors. Then tried to run memory and CPU tests. The loader that was used in front of each card deck stopped every time with errors.
• The cause of these errors has yet to be determined, but failure of the 1442 to complete an operation is current favourite.
• There is some confusion over the type of loader decks the we have, primarily as the number of cards do not agree with the program listings.
• Progress made on the standalone Zone Temperature Monitor/Alarm project. http://wiki.tnmoc.org/index.php/IBM_1130_projects

15 March 2012 Update from Stewart Baker

• Attempt made to identify the punched card loaders we have. Listings of Basic Diagnostic Loader and Relocatable Loader compared. The first card in each case is identical, with the exception of locations x003C and x003E.
• It would appear that the loader which we have (and it's copy), although being a Relocatable Loader is incorrectly labelled as 03A3 whereas it should be labelled 03AA.
• I was unable to find a Basic Diagnostic Loader card.
• Re-ran the loader we have, and confirmed the error halt. This needs signal tracing with a scope. The signals are on XR281 and XR371 and include +End Operation Response and - SRP DSW Bit 4.

18 March 2012 Update from Stewart Baker

• Investigation into loader failure hampered by Read Reg errors on 1442.
• Increasing the lamp voltage slightly got rid of the errors, however the unit needs accurate setting up as per FEMM before trying any more diagnostic decks.
• Zone temperature sensors fitted to 1130 CPU gates. Tested OK.

25 March 2012 Update from Stewart Baker

• Stack of all channel column 1 cards punched in preparation for 1442 adjustments.
• Read clutch backstop too tight. Adjusted as per FEMM.
• Temperature limits of alarm exceeded in zone 6 (28 deg). High museum ambient temperature contributed to this. Checked IBM installation limits for permitted ambient temperatures. Maximum of 90 deg F (32 deg C).
• The system should not be run if the ambient temperature is at or above 30 deg C.
• I will install another sensor (zone 7) with the alarm limit set at 30 deg to monitor the ambient temperature.

01 April 2012 Update from Stewart Baker

• Ambient temperature sensor (zone 7) reads high when the Sun shines on it... DUH ! Need to reinstall, probably on the front LHS under the keyboard.
• Removed rubber pressure block and roller assemblies from in-front of the corner stacker. Found inner bearing dragging due to old grease. Cleaned bearings and repacked. Readjusted roller pressure.
• It is important not to try and run any cards that are bowed or bent through the machine. Putting rubber bands round a card stack is a No-No unless they are protected by thicker cards on the outside.

04 April 2012 Update from Stewart Baker

• Adjusted fibre optic bundle for optimum position as per FEMM page 4-21.
• Adjusted lamp voltage as per FEMM fig 4-9 to ensure all read channels give correct pulse output length.
• Lamp voltage now 3.7v.
• Read emitter timing and duration checked OK as per FEMM page 4-2.
• Skew across all read channels checked <100uS OK.
• Creasing noticed in cards that had been read a number of times. Suspect accumulated residue on some transport feed rollers. Suspect rollers cleaned.
• Now ready to resume diagnostics of 1131.

09 April 2012 Update from Stewart Baker

• Inspected IBM 2315 cartridge found by Peter in D-Block. Disk surfaces in nearly new condition.
• Cartridge needs to be securely packaged and labelled prior to storage in the boiler room.

06 April 2012 Update from Stewart Baker

• Plastic covers at Read and Cornering Stations adjusted as per FEMM (stack of 4 IBM punched cards).
• Found Cornering Station Platen .06" lower than Stacker Platen (should be level). Rubber isolation mount under Cornering Station had become compressed over time. Shim added to bring the surfaces level.

14 April 2012 Update from Stewart Baker

• Return of 'clonking' sound from stacker after shim was added. Tension on Stacker Isolation belts adjusted. Position of Joggler shaft adjusted. Noise gone (for now).

18 April 2012 Update from Stewart Baker

• Fitted 12v DC PSU to Temperature/Alarm unit.
• Cleaned air filters in 1131 as PM.

19 April 2012 Update from Stewart Baker

• IBM 29:
  • Fitted metal guides from spares to replace cracked plastic ones.
  • Removed metal base plate to gain access to the friction clutch and gears.
  • Cleaned all old grease from gears, clutch and casing.
  • Replaced assembly that had worn helical gear and wheel with one tooth missing.
  • Clutch repacked with correct (silicon) grease. Other gears greased.
  • Runs much smoother, but still has intermittent 'Clear' overruns.
• IBM 1442
  • Cleaned rollers and card path at pre-punch station to try and remove card soiling.
  • Cleaned transport path for alternate card hopper.
  • Successfully ran cards through transport, and into alternate card hopper.

28 April 2012 Update from Peter Vaughan

• Manually punched 5 card basic loader set but attempts to read it on 1442 failed with read check errors on each card. Also manually punched one card diagnostics card 5 but this also failed to load. Also manually punched some replacement cards for relocating loader but again cards would not read. Either the manual punch is out of spec (looks fairly good when checking the cards on an 80 column card guide template) or we need to adjust the 1442 reader some more.
• With nothing else to do... I tried the REPO card (a single cold start card used to make copies of card decks) and after a few false starts the system actually managed to read a set of cards to copy and then, to my surprise, proceeded to punch copies of the cards... the first time it has done this in over 25 years!
• Confirmed that the punch was working reliably by manually comparing original and those punched by the 1442. I also confirmed that the 1442 could read the cards it had punched. I then made copies of the relocating loader (03AA) and one card diagnostic sets plus a few more 1442 test cards with all holes punched in card column 40.

29 April 2012 Update from Stewart Baker

• Confirmed that 5 card basic loader set gives consistent Read Reg errors, and that cards checked on 80 column card guide template look good.
• Examination of 1131 Logic Vol 4 XR291 shows grouping of read channels for error checking. Groups are (12,11,0,1) (2,3,4,5) and (6,7,8,9).
• Hand punched cards for 80 col in single channels and combinations.
• Found (12,11,0) gave consistent Read Reg error, so suspect we may have a pattern sensitive logic fault in the 1131, rather than an 1442 problem.

05 May 2012 Update from Stewart Baker

• Peter and I spent a very frustrating day trying to find the solution to a reproducible problem !
• It appears to be a logic fault in the 1130 but tracing through the logic diagrams, checking signals and swapping / replacing a few logic modules along the way has not got us any further forward in identifying the cause.

07 May 2012 Update from Stewart Baker

• Decided to recheck card path for possible skew causing Read Reg errors as they seem to occur toward the last columns of the card.
• Mechanical skew alignment checked on the Read Head.
• Read Pressure Roller clearance to Read Feed Wheel inadequate. Adjusted backstop for 2 thou clearance.
• Read Nudge Pressure Roller clearance twice what it should be. Adjusted Pressure Roll Lever for 25 thou clearance. As the card was being held by this roller when the card was being contacted by the Card Pusher an adjustment was made to the position of the Cam Clamped Hub.
• Because of this adjustment the cards are now able to move more freely into the Pre Punch Bed and are sometimes overrunning the Punch Card Lever slot giving Read Sta errors. The Punch Pusher needs to be adjusted to correct this.
• Outputs of all read channels viewed on a scope together with the Read Emitter signal. Nothing untoward found. However, it was only possible to check the relationship between column 1 data and the Read Emitter.
• The light source suddenly failed. The bulb was removed and found to be OK. Following reassembly the light source worked. Need to check connections in the bulb power circuit. Possible intermittent track on bulb voltage adjustment potentiometer.

09 May 2012 Update from Stewart Baker

• Checked the light source circuitry for bad connections. Exchanged connections on HI/LO switch for unused side of DPDT switch. Cleaned voltage adjustment potentiometer track. No re-occurrence of fault today.
• Readjusted Punch Pusher to place end of card 70 thou from edge of fibre bundle. While doing this I found a small bearing, part of the drag brake was seized. Freed bearing and lubricated it. Also found that pivot adjustment for Punch Pusher was very stiff, lubricated that as well.
• Checked engagement points of all drive bearings/wheels up to the punch against FEMI timing diagram. All OK except the Read Station bearing which stops driving at 210 degs (should be 235 degs).
• Although cards run through manually consistently position correctly in the Pre-Punch station, they do not when powered. The card positions in that station are all over the place resulting in frequent READ STA and PUNCH STA errors. Ran out of time (and ideas)...

16 May 2012 Update from Stewart Baker

• Rechecked Pre-Punch station mechanical timing against diagram in FEMI.
• Found that the Read Nudge Pressure Roller was releasing the card too late, and that the angle over which the roller was in contact with the card was too long.
• Increasing the Read Nudge Pressure Roller clearance slightly together with a readjustment of the cam decreased the time the roller is in contact with the card.
• Went over the Pre-Punch station adjustments again, and this seems to have cured the READ STA and PUNCH STA errors. The correct adjustment of this area of the machine seems extremely critical.
• I suspect that IBM CE's compensated for wear in this area by increasing the roller clearance beyond the FEMM specification.

19 May 2012 Update from Stewart Baker

• 1442 READ STA and PUNCH STA errors still present. Checked Punch Dark signal timing with manual card feed and this was OK.
• Cards are still being randomly mispositioned in Pre-Punch station. This seems to be due to drag on the front of the card, generated in the Punch Station.
• Found eccentric adjustment to Punch Roller loose, however after tightening, problems still remain. It maybe that following punching one or more of the bearings in the punch unit have seized.
• Next action is to remove Punch Unit, inspect and fix any defective bearings and readjust the eccentric coupling.

23 May 2012 Update from Stewart Baker

• Rather than remove the Punch Unit I was just able to see the front and rear bearings by looking down from the top of the punch. Phew !
• Pressing the Punch Roller release lever enabled me to check that both bearings rotate freely. They do, so I lubricated them.
• Removed the eccentric adjustment and fitted a spilt washer under the nut to stop it from moving. Adjusted the eccentric and checked that the Punch Rollers are activated correctly.
• Unit still gives READ STA and PUNCH STA errors due to erratic card positioning in the Pre-Punch station even if the Punch Roller release arm is held down to release the Punch Rollers.

30 May 2012 update by Stewart Baker

• Continued investigation into erratic card positioning in the Pre-Punch station.
• Found that grease was present on the gears associated with the Read Nudge Roller. These should be dry. Grease removed from the gear teeth
• Adjusted Read Nudge Roll Idler Assembly to give increased clearance as gears have high spots caused by wear which may have been binding. This may have caused the Read Clutch drive to give inconsistent drive to the rollers.
• All adjustments in the Pre-Punch station rechecked against the FEMM.
• A number of card decks were read into the 1130 without READ STA, PUNCH STA or READ REG errors.

04 June 2012 Update from Peter Vaughan

• I ran various sizes of card packs through the 1442 to check Stewart's previous adjustments were still OK. On the whole the cards read through fine with no READ STA or PUNCH STA errors reported. I did see occasional READ REG error on cards that have read through OK. I did try to read some of the cards that failed previously and they still fail with READ REG errors so we still have a problem there.
• Had a look at the 1053 console printer which is currently unplugged due to constant returns occurring. Discovered why the golfball head could not be removed; the release mechanism on the top has broken and fails to separate the two rods either side of the core locating pin. I managed to separate the wires manually and remove the head. This may be difficult to repair but we do have a spare that came with the system, but it's condition is unknown or if it is the correct type. We do have some other Selectric print-heads with what appears to be a better release mechanism so we may be able to use a release mechanism from one of those if our spare is not suitable.
• Also had a look for the missing the nut on 1053 tab linkage too see if it may have got lost in the mechanism, but found nothing. The bolt cannot be removed without taking the mechanism apart so will have to try and find and fit a replacement in-situ.

04 June 2012 Update from Stewart Baker

• IBM 29
• In an effort to cure intermittent 'Clear' overruns
• Removed large amounts of old grease from gears in the top half of the machine.
• Re-greased the gears.
• Removed unit relays one at a time, inspected and cleaned each one with IPA.
• Found solenoid latch for card picker was loose, this could be the cause of the problem. Tightened.

10 June 2012 Update from Stewart Baker

• 1053 console printer swapped with spare from D-Block. As this printer is incomplete some items such as the platen needed to be transferred from the original printer. The printer had been re-furbished in 1985, however because of the long storage in poor conditions it was decided not to power it on for the present.
• The original console printer from the 1130 was put on a workshop bench for testing. After a period of trying to rectify the underlying CR/LF fault Peter and I decided that this was beyond our knowledge and capabilities.
• The printer has been boxed up with notes as to what the problems are, together with some small parts that need refitting.
• Investigations started into the possibility of repair/overhaul of the printer by an outside agency.
• The 1442 is still giving indeterminate READ REG errors. It is still undetermined whether these are being caused by mechanical or electronic issues or a combination of both.
• Systematic checking of the 1442 mechanical drive path carried out to ensure that parts are moving with the minimum amount of effort. This involved cleaning and re-greasing a number of bearings, together with adjusting belt tensions.
• As we are still getting Feed Clutch Overruns this problem is the target for further investigation.

13 June 2012 Update from Stewart Baker

• Investigation into Feed Clutch Overruns.
• All parts associated with Card Feed Clutch assembly removed, cleaned, re-assembled and adjusted. Result - no more Feed Clutch Overruns !
• Pre-read station Cover clearance adjusted. Card Pusher adjusted to give correct card positioning at the 40th column.
• Timing of the Read Emitter Pulse found to be marginal WRT Column 1 Data Channels.
• Adjusted Read Emitter disc to centralise the Read Emitter Pulse in the Data Pulse.
• Read Basic Diagnostic Loader and Relocatable Loader into 1130 without any READ REG Errors !

17 June 2012 Update from Stewart Baker

• Found that 40th column alignment had drifted out by 10thou. Card pusher readjusted.
• Re-checked Data Channels vs Read Emitter Pulse alignment. Alignment still correct.
• Card Feed and Read Clutches pressure greased. Alignment still correct.
• Getting occasional Hopper Errors and first cards not being loaded.
• A number of diagnostic card decks were read and copied successfully.
• The Basic Loader gives read checksum errors. This is probably incorrectly punched. Peter will investigate.
• Using the Relocating Loader the following diagnostics were run without any errors:-
• Meter Timing Test - identified the 1131 as having a 3.6uS clock.
• Low Core Memory Test
• High Core Memory Test - identified the 1131 as having 8k core memory.
• Information found while looking through card storage identifies the system as having been used at:-
• Liverpool University Department of Nuclear Physics.

23 June 2012 Update from Peter Vaughan

• Manually punched basic diagnostic loader (03A3) cards checked manually and found to be punched correctly.
• Once the 1442 had warmed up by feeding cards via NPRO, I checked the 40th column alignment in the read station. This was still correct so no adjustments needed.
• Hopper first card feed pick-up issues still occurred throughout the day so this needs to be investigated further. Once the first card is picked up no problems with feeding of following cards.
• Also on the first few occasions there was a delay between pressing the start button and the first card actually loading into the read station - same issue as we had last week. However after a few loads the problem did not occur again during the day.
• Loaded and ran the low core (03B1) and high core (03B0) diagnostic tests for 10 minutes with no errors.
• Investigated why the manually punched basic loader was not working. Last week we found the first 5 cards in cpu function test deck appeared to be the same as the basic loader. I checked this again and apart from a few columns on one of the cards, the first 5 cards did match the basic loader so assumed it actually was it.
• Loaded the CPU function (03A1) test deck but this time did not add a basic loader before it (because the first 5 cards were the basic loader). This time the whole deck apart from the last card (see later) loaded and the program started running as documented. I then let it run, it ran for 2 minutes and completed successfully with no errors reported. I repeated the test with the same result. I then let it run continuously for over an hour with no errors reported. This has now proved all but a few of the CPU instructions are working and that we have a fully functional CPU. One thing that appeared strange was the last card in the deck, which was not blank, was not actually read, the loader stopped after reading the card just before it. This did not stop the test running so not sure what that cards contents do.
• Spent most of the afternoon making copies of the diagnostic card decks. Most of the card decks have had at least 2 copies made but as most of them cannot be run yet it is not known if there are any punch errors. Note the first couple of cards punched produced an error in position 2 on column 2 (same as last week - hole not punched) but this problem did not occur (as far as I can tell) with subsequent punching.
• I looked at the manually punched basic diagnostic loader deck again. Knowing the CPU function test read and ran fine, I replaced the existing 5 cards with the manually punched basic loader and attempted to read in the CPU function test deck. It failed with a checksum error part way through the deck. I repeated the load and it stopped at the same place. I repeated with the original 5 cards and it loaded fine. I have concluded the manually punched basic loader is not correct so copies of the one used on the CPU function test deck have been copied as the basic diagnostic loader. This copy has not been proven on another diagnostic as none of the others using it can be run yet.
• Noticed a different smell during the running of the CPU function test after it had been running for a while. Something may have been getting warmer than usual or we were using something new which may have had dust on it. All fans checked out OK and temps were not too high so something to watch out for.

27 June 2012 Update from Stewart Baker

• Investigation into intermittent first card feed and Hopper miss-feed problems.
• Found corroded contacts on Clutch Solenoid connection strip. Cleaned with contact cleaner.
• Found Hopper Throat Knife not parallel to Throat Roller. Adjusted for 10thou clearance.
• No problems observed during a day of testing.

06 July 2012 Update from Stewart Baker

• Hopper miss-feeds may be caused by card decks being left in the hopper with the Throat Rollers distorting the first card. Recommend that card decks are removed from the 1442 when not in use.
• Work started on the 2310 disk drive.
• All accessible surfaces on upper half of the drive cleaned.
• Drive mechanics cleaned and lubricated as per FEMM.
• Checked drive belt condition and tension.
• Found "Even" Detent Plunger rubbing against the Detent Rack. Adjusted to give clearance.
• Disconnected data connector A2 from drive electronics to enable CE mode. '''NOTE - Warnings about switch positions in FEMM.'''
• Exercised heads across disk surface using CE switches. As this was tried using original disk with surface defects the time testing was deliberately kept to a minimum. We need more new disks - CHM ???
• Air filter at rear of the drive is missing. A replacement needs to be found.
• The Air Blower bearings need lubricating. This unit should be removable with the drive in-situ.

12 July 2012 Update from Stewart Baker

• No Hopper miss-feeds since card decks left out of Hopper between runs.
• 2310 Air Blower removed and cleaned out. Vast amount of 'crud' on fan blades and inside blower casing.
• Bearings lubricated.

23 July 2012 Update from Stewart Baker

• Visit to TNMoC by Leny, a friend of Olli. Leny had been an engineer working for IBM on Selectric typewriters.
• Within an hour of "laying on of hands" on our defective console printer he had it rising from the dead.
• Agreed that, when Leny has cleared sufficient space in his spare room we will take the printer to him for refurbishment. This will be in September, and the exercise should take approximately 2 months.

02 August 2012 Update from Stewart Baker

• Work started on the 1132 printer.
• Inter-unit wiring cleaned.
• Clogged air filter under electronics cage cleaned.
• Corrosion present on the electronic backplane due to poor gold coverage and high humidity. Treated with Deoxit D5 applied with a toothbrush.
• "Earth Safety Test" label dated 24/11/78 found on the PSU top. This may give a clue as to how long the system was in use i.e. 10/11 years.

06 August 2012 Update from Stewart Baker

• Partial disassembly of print head to determine condition.
• Removal of Magnet Assembly gave good access to the majority of the print mechanisms.
• Very pleased to see that none of these assemblies have been affected by rust...
• Although initial stiff to move all operating mechanisms responded well to WD40 then Meths followed by lubrication. Note: we need to find a spray-on lubricant equivalent to IBM #6.
• The next action is to remove all internal metal panels to gain further access around the print head. The timing disk should then receive attention.
• Cover removed from the Main PSU, prior to checking/reforming capacitors, followed by in-situ testing.

08 August 2012 Update from Stewart Baker

• Some internal panels removed to gain access.
• Timing disk carefully cleaned with fine grade wet & dry to remove some of the rust. Electrolytic treatment not used as the heat generated could warp the disk. Disk covered with a thin layer of oil for protection.
• Photocell assembly cleaned and the lamp checked OK.
• Printer ribbon assemblies removed, cleaned and lubricated.
• Gears cleaned with WD40 and Meths. These will be lubricated latter.
• Print mechanism turned by hand. At first very stiff with a number of Print Clutches not working. WD40, Meths and lubrication applied. All columns are now working OK.
• '''Note:''' The print mechanism should only be turned CCW by hand.

12 August 2012 Update from Stewart Baker

• Individually checked each print mechanism by manually tripping Print Magnets. Found some sticking. Cleaned and lubricated mechanism. Now OK.
• Checked 4 off 11000uF 65V electrolytic capacitors in PSU for ESR. All OK.
• Capacitors need reforming.

16 August 2012 Update from Stewart Baker

• All major gears lightly lubricated with grease.
• Reformed 4 off PSU capacitors to 60V.
• Broken panel lamp assemblies need attention.

23 August 2012 Update from Stewart Baker

• Broken panel lamp assemblies repaired and new 12v 0.1A bulbs fitted.
• Printer mechanism lubricated with Gun oil.
• All PSU fuses checked for value and continuity.
• All PSU lines on GTB checked for short circuits.
• Relay and contactors cleaned.
• Close to powering up printer.

26 August 2012 Update from Peter Vaughan

• Replacement Variac found that will supply the power we need for all 3 units to be powered up at the same time. Initial cleaning done, it still needs testing under load.
• Cleaned fan under 1132 logic box.
• System ran CPU test all day without error
• Lubricated disk pack flap on both packs so they no longer stick and can be inserted into drive when flap is fully closed (drive not powered on).
• Had an ex IBM engineer visit who used to maintain 1130s (lives too far away to be able to help us). I did get some useful tidbits of info during our brief chat:
• Had very few logic board faults during the life of many of the systems he maintained.
• Some common faults on the 1132 were the solenoids driving the hammers and the ribbon getting chewed up if the gap to the paper was wrong. Also it uses a metal eye at the end of the ribbon to reverse the ribbon direction and this sometimes did not work resulting in the ribbon being shredded.
• I asked if he still had any info on the 1130 as we were very interested in getting docs for the machine. He said he might so gave him my email address - however I forgot to note his name. If he does contact me I'll ask if it's ok to contact him with any questions we have.

01 September 2012 Update from Peter Vaughan

• A 2315 disk pack has been kindly donated by Henk Stegeman who runs a collection of IBM System/3 systems in Holland - He has had it in his house for years but was not compatible with any of his systems so he was happy to donate it. This is a special (and rare) CE pack which was used for head alignment and diagnostics.

26 August 2012 Update from Stewart Baker

• Internal panels replaced.
• Gear and bearing assemblies in paper transport mechanism cleaned of old grease.
• Gear and bearing assemblies in paper transport mechanism re-lubricated.

17 September 2012 Update from Stewart Baker

• Replacement Variac reassembled and tested on load. Output smoothly variable up to 266V. Needs to have the output voltage continuously monitored when powering the IBM 1130 system.
• Cleaned and adjusted throat components of the 1442 hopper to try and improve first card handling.

08 September 2012 Update from Stewart Baker

• Cleaned badly corroded 1132 interface connectors.
• 1442 disconnected from 1131 prior to 1132 power up.
• Systematic power up of 1132 carried out.
• 48v rail measured as 50v with correct capacitor discharge on power down.
• 1132 successfully powered up with the only obvious problems - rubbing of pulley against a cover and the drive motor making excessive noise.
• Exercised the printer off-line as far as possible. Desperately need a working console printer to run the diagnostics...
• All of the 1130 system components connected to the heavy duty Variac.


• The complete 1130 system:- 1442, 1131 and 1132 are now running together for the first time in 30+ years !!!

21 September 2012 Update from Stewart Baker

• Cleaned, then re greased and oiled the paper advance mechanisms.
• Intermittent forms loop overrun cured.
• Cleaned, lubricated and adjusted ribbon transport.
• Printer drive motor and pulley idler bearings cleaned with WD40 prior to re greasing.

28 September 2012 Update from Stewart Baker

• Pulley assembly removed from printer, disassembled, bearings packed with grease and the unit refitted.
• Drive belts inspected. They are in good condition.
• Drive belt tensions set to correct value.
• "Rumbling" noise from drive train almost gone.

9 November 2012 Update from Peter Vaughan

• With Stewart away from the museum and the console printer still not working no progress has been made recently. A decision on how best to fix the console printer will be made when Stewart returns.
• When running, the reader and processor are still working fine. The printer is powered on for short periods and seems to run OK but we are unable to progress testing until the console printer is fixed.

02 December 2012 Update from Peter Vaughan

• Console printer now with Leny in Leeds for service / repair. He will also be looking at the interface to the 1130 and how to interface to the printer. This will enable us to design a test box to test the printer stand-a-lone. It will also enable us to design an interface to replicate the printer allowing output to a PC screen / terminal for times when the console printer is not working.
• Had problems reading cards on the 1442 today with several card jams when running through scratch cards. When observing the green demo desk it was clear that cards were getting stuck at the corner station and on many occasions several cards ended up together before they eventually continued to the stacker - on two occasions the cards got jammed and damaged. Further testing on other green cards saw the same with lots of cards hesitating at the cornering station. I cleaned the area and rollers with meths but the problem did not go away. Subsequent testing with pink and white cards did not show the same problem so we may have a problem with using the green cards or something has gone out of adjustment. I re-punched the demo decks with white cards and both sets passed through the reader without any problems. Maybe time to check and re-lube that area.