Looking for tips debugging DEC TC01 controller

[Josh Dersch]

Hey all --

I picked up a TC01 in trade (for a TC08) a couple of weeks back and this
past weekend I got it hooked up and powered up with my PDP-8/I + TU55
transport.  I've been debugging it and have solved a couple of issues but
the current one has me stumped and I'm looking for advice, hoping I
overlooked something obvious.

At this point the PDP-8/I can talk to the TC01 and use it to control the
TU55 without issue.  I've mostly been using MAINDEC-08-D3BB to exercise the
hardware (see:
http://svn.so-much-stuff.com/svn/trunk/pdp8/src/dec/maindec-08-d3b/maindec-08-d3bb-d.pdf)
and issuing FWD/BACK commands via the "Basic Motion" routine the tape runs
from one end of the tape to the other, stopping at the endzones correctly.
The tape comes up to speed properly, the COUNTER register increments, and
in general the various status lights on the panel do what I expect them to
do.

The root symptom I'm seeing now is that DECtape operations (like SEARCH)
don't complete, and I've traced this back to the MK BLK MK signal not
getting asserted.  This signal is generated by a large AND gate (see p. 112
of
http://bitsavers.trailing-edge.com/pdf/dec/dectape/tc01/DEC-08-I2AB-D_TC01_Jul69.pdf).
This gate ANDs together a pile of signals from the WINDOW register and the
counter, looking for a specific bit pattern on the mark tracks; this AND
gate is never satisfied because the WINDOW register's MSB (W1) is stuck
low.  On the scope it's a completely flat line, no glitches or anything
noticeable.

Despite my best efforts I cannot figure out why this is.  Here's what I've
looked at and what I've discovered thus far:

1) The inputs to the W1 flip-flop (pins U (clock) and V (data) look to be
correct -- U is identical to the TP1 signal being passed to all the other
flip-flops in the chain, and V is identical to the data coming out of W2 --
levels all look fine.

2) The outputs of W1 (pins T and S) are not being pulled up or down by
anything external. I've gone so far as to completely disconnect T and S
from the backplane (with some tape over the card fingers) and the outputs
are still a flat line.

3) The output of pin T goes to -3V when 0->WINDOW is de-asserted (when tape
is in motion), returns to 0V when 0->WINDOW is asserted after the tape
stops.  (S is the inverse of this, as expected.)  (As an aside, the
schematic drawing suggests that 0->WINDOW ought to be a pulse given the
arrow symbol; this does not appear to be the case.  Other flip flops in the
chain seem to behave properly, regardless...)

4) The backplane connection to the R203 flip-chip that the W1 flip-flop is
on is fine.  I've beeped this out and there's no significant resistance
between the backplane pins and the connection to the card in the slot.

5) W1 is not affected by the state of the two other flip flops on the R203
(also tested by disconnecting all pins other than power/gnd and pins
R/S/T/U/V with tape).

6) Swapping in a different, working R203 shows no change in behavior.

6) Sometimes, randomly, W1 starts working.  Last night I noticed that the
solder side of R203 was rubbing against the top of the "flip chip" packages
on the R107 next to it and even though those shouldn't be conductive, I
stuck a piece of cardstock between the two.  W1 started working!  I tested
things for about a half hour in this state (interrupts were still not
occurring, much to my annoyance) and this morning I powered it up and it
was still working.  Pulled the cardstock out and W1 stopped working; this
would seem to causation.  Put the cardstock back in... still won't work.
Backplane connections all test out fine (see (4) above).  Put the R203 on
an extender (to completely avoid interference with neighboring cards) and
still no go.  Possibly the insertion of the cardstock flexed the board or
the connection with the backplane slot ever so slightly to make things work
but if so I've been unable to replicate it.

All of this still sounds like a bad connection somewhere (backplane, cold
solder joint, broken trace) but the evidence seems to be against it
(checked, double-checked, triple-checked backplane, swapped R203s around,
etc...)

Anyone have suggestions?  I'm going to start looking at things on the board
itself (which is going to be a pain given how things are currently arranged
in the chassis) but curious if I've missed anything here...

- Josh