IBM 1130 Reparation Blog, continuation
The beginning you'll find here IBM-Blog, start
7.6.2014: The card 7432 we are looking for is only used in systems with a memory expansion. These mainframes are much rarer than the smaller IBM 1130 systems without the additional left enclosure. Therefore both, the card missing in out machine as well as its schematics have apparently not been preserved. Accordingly, we reverse-engineered the circuit diagram based on measurements on an existing card and rebuilt the card. Now the core memory of our IBM 1130 is working like a charm which is a significant milestone.
So it looks while repairing. Invisible: Measuring Park and many simultaneously opened Manuals.
Enlarge: click on picture!
The next test involved the console printer: A working console printer is necessary to communicate with the machine - unfortunately, this test showed that the printer is still faulty. Currently, we suspect resinified oil as the culprit.
31.08.2014: We would like to thank the people from the "Verein fuer ein Museum der Lochkartentechnik", cosecans.ch, who donated a set of plug in cards for the IBM 1131 processor. Without these spares our repair attempts would be next to impossible due to the SLT technology.
Further tests showed that memory and CPU test run flawlessly but the keyboard/printer-test still fails.
07.09.2014: We experienced some set-backs during our initial restoration attempts. This was not that unexpected since the machines have been sitting idle for several decades. We have at least two different problems in the machine. Without diagnostic programs, any repair attempt would be in vain. Just the keyboard/printer test shows an error:
"E0001 INVLD" denoting an erroneous address in the switch register for that particular test. Unfortunately the address set on the switches is correct so another problem is still lurking in the machine.
13.09.2014: The next step involved testing the card punch 1442 by means of the diagnostic tests. The test program punches cards with certain patterns in a first step. These cards are then read back in a second step and compared with the expected patterns to reveal any differences. Our punch only punches the first 40 columns of a card which is obviously an error. Fortunately, the console printer works fine despite the error message generated by the console-test program (07.09.2014).
Enlarge: click on picture!
30.09.2014: The printer test finally works but there were still problems in the printout: Characters were overprinted every 16th column. This could be traced back to the buffer memory of the printer. Bit 6 of the buffer, an SLT-flip-flop, was stuck. How an SLT circuit can be repaired is described in more detail here: Deeper insights.
3.12.2014: As already noted on September 13th, the IBM 1442 model 6 card punch creates erroneous hole patterns on the cards. Fortunately, we have a second IBM 1442 which we connected to the machine after adjusting all necessary parts. This is a model 7 and thus faster than the model 6 it replaced. With a working card punch it is now clear that the processor is working correctly, and the fault must be on the side of the original 1142.
We were also able to repair the defective 6th bit in the printer buffer (cf. 30.09.2014) by replacing the faulty SLT module (the two defective transistors could not be replaced).
25.1.2015: After cleaning all moving parts in the 1132 printer as well as in the console printer and adjusting both of them, the overall system is now fully working as a punched card computer.
The next project will be the restauration of the 2310 disk drive. Unfortunately we have no spare hard sectored removeable disks with eight sectors for this drive. All we have are hard sectored disks with 16 sectors. Accordingly we will try to convert one of these to eight sectors by masking every second index slot.
10.07.2016: Finally we found time to repair the 2310 disk drive. After replacing a 16 bit buffer register it works like a charm. We already compiled some FORTRAN programs which ran flawlessly. This completes the restoration of this unique machine which is now fully operational again.
29.01.2017: We managed to repair the paper tape readers for the 1130. By now we can load the ASCII picture "SNOOPY" from paper tape and print it. The machine can also punch arbitrary data to tape with by means of a program written in FORTRAN and assembly language. This program occupies 392 punch cards. Now only the plotter needs to be connected to the machine.