Wednesday, July 5, 2017

Core problem on 1401, progress on disk tool and in wiring up relay tester


Our Connecticut system has a bad core location up in high memory. Consistent errors at one location thus it is almost certainly a single core issue. It will be fun to fix that one. Otherwise, both systems are working fine.

One of the 729 tape drives has had a problem with a coupler failing that connects the capstan to its drive motor. The rubber disintegrated. A different coupler was installed but it appears that the shafts are misaligned somehow which is putting the stress on the coupler. We will watch to see if this coupler will work long term, otherwise we must either correct the alignment or do some other more serious surgery. 

George was working on his SMS card tester, which has signal issues in the cabling between the digital logic section and the analog card or somewhere on that card. 

While working on that, we discovered that a consequence of using the laptop power bricks to supply the +/- 19V power for the unit is that the ground of the SMS tester is actually set 20V off from ground of the oscilloscope (thus 20V off from the AC nuetral). We will move over to a lab power supply, using it to provide the input voltage without the ground offset, and see where we get to. 


Finally today I beat my dummy data driver logic into shape and have a properly formatted string of data bits going into the disk tool, looped from some temporary output pins, to let the logic attempt to read a sector properly. With this, I can check and debug the ReadSector logic.
Now I can use the logic analyzer and track the operation of the ReadSector and ReadField state machines. First step in reading any field is to sync up by finding the first non-zero data bit, from then count off bits in each word as they are deserialized and stored in RAM. Right away I can see that the first non-zero data bit came in but the logic did NOT work properly from that point.

The trace shows the sync state being properly recognized from an initial '1' data bit yet the checksum validation failed. I began reconstructing the signals to watch the extracted words from the deserializer. It should tell me whether I am reading the stream on the right boundaries, when I resume testing tomorrow.


Today I found some clamps used with the relay sockets as they are installed on 1950s era IBM equipment. There is a clamp that latches over one end of the socket, providing round holes for screws to mount the socket properly, but the other end of the relay does not fit these. It seems that IBM used long common bars that a group of relays would slip under, anchoring them individually with the small clips I found.

The clips come in two widths, for the red 4-circuit and the green 6-circuit relay types. I altered one of each clip type, cutting off enough so that it could install upside down on the far end of the relay socket, still giving me holes for mounting screws.

I used these to firmly mount the sockets into my box and could do the final wiring of the relay sockets. Next up I have to mount two boards inside the box - the Arduino Mega 2560 and the dual relay module - plus a barrier strip.

Top with relay sockets mounted
Additional steps are routing of two power brick cables into the box to a short barrier strip, cutting an opening to insert the USB cable through the box onto the Arduino and final wiring of the relay socket wires to the components inside the box.

Arduino and relay module to be mounted (also a barrier strip)
Wires and resistors installed on sockets, ready to insulate

No comments:

Post a Comment