Wednesday, October 4, 2017

HP 1000 system restoration - memory now working, moving on to tape drive


Today was Wednesday, when I spend much of the day at CHM working with the 1401 restoration team. We have a tape problem on one drive on the "German" system, where certain sequences of tape operations result in the drive spewing data throughout memory, overwriting any programs.

Last week it manifested by the tape running away continuously after the system Stop button was pushed. We set up the scope to watch the signal that should tell the tape adapter unit to disconnect and halt tape operation, but that prior runaway problem had ceased and the overwriting behavior manifested. We didn't get to the bottom of it this week. The other ("Connecticut") machine ran fine.

Once home I wired and inserted the rack power switch back into the HP 1000 cabinet and it worked just fine. The bulb inside that should illuminate is burned out, but I have a replacement on the way.

I then began pulling out the four memory boards, each implementing 64K words of high speed parity protected memory. First up I validated that the DIP switches on the boards had configured them for their proper address range. I then inserted just the lowest 64K board and tested the system to see if the alternating all-one-bit and all-zero-bit words would appear.

The system worked properly. I inserted the second board and tested again. All was good. Third board, same thing. Although the logical conclusion is that the fourth memory board is bad, I inserted it anyway and ran the tests. The problem went away! I have to assume that I had erratic contact of a board with the memory backplane and/or the front ribbon cable that bridges them to the memory controller.

In any case, with memory working right, I looked at the initial boot load sequence to verify that operation as far as possible without an actual working peripheral on the end with code to read. I set up the sequence to boot from my 7906 disk drive, using the controller card in socket o13.

In fact, I identified and verified all four ROM loader contents - doing an IBL with the slot number in bits 15 and 14 of the S register and the octal slot number of the controller card in the IO cage. I then stepped through locations o077700 to o077777.

Now, the error light (Overflow) didn't light up, meaning it set up the boot ROM in the upper words of accessible memory, checked that the controller card was live, and prepared to actually boot once the drive was ready.

Of course, the drive isn't ready, but I hit RUN anyway and saw the loader attempt to read from the drive. Later I will look at a listing of the ROM chip in socket 11 which is how I attempted the IBL, verifying that the contents are correct by stepping through the memory addresses.

I turned to my 7970B tape drive to restore it to operating condition. Yesterday, it initially did load and rewind but failed to spot the reflective beginning of tape (BOT) foil as it passed. Later yesterday, I heard a very faint sound and now the drive won't load, rewind or respond to the four top buttons at all.

I checked the fuses in the drive and found all of them good. Next up I should verify that all the power supply voltages are present, since that is the most likely cause of the failure to respond. If I can fix that then I can return to debug the failing BOT sensor.

The power levels were just fine, +5, +12, -12, +20, -20, +57 and -57, thus I have to look at the logic itself to see what is going wrong. I printed out the appropriate schematics and board layouts to begin probing.

One failure candidate signal I found right away is called "Control delay" which is a timer that should hold the control logic in reset for a brief period while power comes up. If that fails, I would have the symptoms I am seeing.

Alas, it isn't delaying anything, it has dropped to ground. I also hear the relay K1 click on and off when the Load button is pressed. This should trigger the reel motors to tension the tape then roll forward until the BOT spot is sensed.

Still, the lamp in the Reset button should light. I suspect that bulb is out, as is the bulb in the BOT/EOT sensor area. I ordered bulbs and will make sure I have all good bulbs in the unit. 

No comments:

Post a Comment