[vcf-midatlantic] Workshop results

[David Gesswein]

Worked on straight 8. Reported problem was toggled in RIM loader wouldn't
load the BIN loader tape. Verified issue was still occurring. Since we knew
we had issues with connection to cards from last workshop cleaned all the 
processor cards edge fingers and backplane connector using paper/paper towel 
with contact cleaner.  Battled them during restoration of the machine was 9 
years ago due to storage conditions. 

Didn't change symptoms so single stepped through the loader.
Found the RTL (Rotate Twice Left) instruction didn't do anything. Traced it to
bad transistor in R603 pulse amplifier in PB19. With transistor replaced
and after figuring out I loaded the paper tape backwards the machine worked.

Processor sensitivity to motion seemed fixed and running finger down cards
also didn't cause crash. Memory side was still sensitive so cleaned those also.
Luck didn't hold so memory was non functional after cleaning. Tracing signals
with scope found timing pulses missing from W300 delay line board. Inspection
found cracked solder joints on both of the board. The delay lines are too tall
so get pulled on when inserting or removing boards. Think that cracked joints
were present and got worse when cards removed for cleaning. After resoldering
Still not working. More probing and found one signal for write was not being
generated. Inspected transistors on the W300 generating it and noticed one
transistor lead was all rusty. Poked it and it disintegrated. With
transistor replaced machine now worked.

Ran diagnostic set and reloaded demos and declared ready to go. Hopefully
the problems on the W300 was causing some of the occasional flaky behavior.

All board cleaned except sense amplifier boards that had additional connections.
Verified moving them didn't cause crashes. The boards plugged into the memory
box also weren't cleaned since they are gold to gold. Also didn't clean front
panel cable boards. If I was able to accurate count that high I cleaned
227 edge and backplane connectors. Each has 18 contacts.

We discussed enabling the 12 bit A/D that was installed in the machine when
we got it. My memory was that wire(s) needed to be removed from the backplane
to enable when I looked at it years ago. Maintenance manual says only cards 
need to be installed. Decision was to get it working removing wires if needed
but I finished work on machine at 4pm Sunday so decided not a good time to 
mess further.

A/D can convert with 6 to 12 bits accuracy taking 7.6 to 56.6 microseconds.
Input is 0 to -10V. We discussed some what demo to use that would be of 
interest to the public but not too gimmicky. Still thinking about 
it. Have Teletype where we can print results or text graph.