Today one of the two 1401 systems began malfunctioning, with the first 4000 positions of memory failing to respond at all. We spent most of the day checking signals but hadn't yet determined what is going wrong.
Because all 8 bits (BA8421, parity C and wordmark) of all 4,000 characters are malfunctioning, it can't be an individual sense amplifier or driver. It has to be something that will affect all of the memory in the 4K section of memory.
In the 1401, the core stack for the first 4000 characters is inside the processor frame, while the remaining 12000 characters is housed in the external 1406 box (as three additional core stacks). Oddly, some of the addressing logic for the base 4000 character positions is housed in the 1406 then runs back to the 1401 to drive the core stack inside the main frame.
Card reading, punching and line printing all make use of additional core planes in the first 4K core stack, thus it is not identical to the other three stacks out in the 1406. Other than these differences, the four 4K stacks are pretty much the same. Logically, each 4K stack is treated as two 2,000 character stacks, but that makes no essential difference.
There are quite a few pages of logic diagrams covering the circuitry to select a particular 4K segment, drive reads and writes, and route the data from sense amps into the processor busses. We also swapped cards with like card types in other slots in the area most likely to cause our problem, the selection logic. As I said, we have more work to do before we can find the problem and repair it.
I gained new knowledge from well written content of this blog. It is showing some different kind of strategy to keep work better and improve with every new assignment. Gracefully written blogIBM TS2250 Tape Drive Model H5S
ReplyDelete