[vcf-midatlantic] Tales from the Repair Workshop 6/12 & 6/13, 2021
djg at pdp8online.com
Tue Jun 15 02:12:17 UTC 2021
Straight 8 work
Report before I arrived was that the paper tape text punch program
was no longer running. It was unclear if it needed to be reloaded or
machine had more problems. Found memory was not operating properly.
2 diodes replaced fixing one memory issue and hopefully fixing another.
Failure of accumulator card late in workshop prevented verification of memory
fix. Machine not useable until I fix and return the accumulator card.
Hopefully I can fix at home and send it back with someone at the System
Source event next month.
Initial testing showed the Computer would not reliably read or write memory.
Checking signals on memory state card MD16 the flip flops were not always set
back to idle state at the end of the memory cycle. Traced to the end of cycle
signal from delay line in MC/MD18 not always present. The signal amplitude
was gettting reduced as it went down the delay line taps. By the time it
got to CP output it was marginal on triggering the output gate. With testing
and swapping with card in MC/MD17 it did not look like the delay line card was
the issue. Looking further the input pulse to the delay line was only 25 ns
wide coming from W607 in MC19. This seemed short for the PDP logic. Checking
the W607 datasheet its output should be a 70ns pulse. The 25ns is too short
for the bandwidth of the delay line to pass without significant loss.
Checking card for the second time I found one of the diodes had .7V drop
measured with my meter instead of typical .6. Replacing diode got the pulse
to proper 70ns and fixed the memory operation. Diode replaced was D5 on
W607 rev D schematic.
The front panel had been getting flakey with examine and deposit double
incrementing and other behaviors. I had added a cap to increase the debounce
time but only helped some. I went through my contact cleaners and treatments
on the momemtary switches. They seemed to work well after so removed the extra
capacitor and they still worked well. Entered the moving light test program
without problems and it ran fine.
Long ago I had put a temporary cap in on the processor backplane to replace
a failed one. I replaced it with what I though was a match for the original
cap and also a similar cap in case it was going bad. Caps were supposed to be
exact match but found the one I removed was TE-1304 10uF 50V and the one
I had with me was TE-1507 10uF 150V. Not going to matter just annoying
that I got things mixed up.
Ran the 5 processor diagnostics without error. The memory diagnostics failed
intermitently. Checked the G603 memory selection matrix cards since they
caused similar problem previously. Found and replaced one bad diode
on second from left card. Thought everything would be good now but
the machine decided to toy with me. Had a stuck at 1 fault for memory
data bit 10. Traced to G007 card which input for core looked good but
output was bad. Nothing found wrong on card with DVM. Put it back in and worked
fine. Then machine still not running properly. When testing I noticed AC8
wasn't 1 when MB8 was 1 examining memory. MB and AC should both get the value
from memory. The problem moved when I swapped R210 AC8 and AC9 card. After
the machine was off for a while then turned back on the front panel light
would light but would go out after a few seconds when reading memory. The
fault also dependent on the value read from memory. If memory has 0014 AC
gets 4. 0034 loads correctly and 0010 loads correctly.
R210 is a complex card and little time was left so I packed up and brought
the card home to hopefully fix. It has tape on it indicating I had problems
with it previously. Haven't checked to see if it will reproduce in my
More information about the vcf-midatlantic