The Restoration of the PDP-8/I Minicomputer (PAGE 7)
At this point, I got enough of the M220B schematic done that I could begin testing.
Since DEC was "kind" enough to leave the 7474's PRESET and CLEAR pins floating (no, not tied HIGH through a resistor),
I was able to mount the M220B's in my testjig and manually tickle the PRESET and CLEAR lines while observing
the Q and bar-Q outputs of the 7474.
Bingo! One dead 7474 (1969 vintage) -- the PRESET had no effect on it.
So, I replaced that chip (with a 1983 vintage chip -- hey, at least it's an old chip :-)).
At this point, I was able to manually load each and every bit from the front panel switches into the program counter.
That's progress.
However, that soon stopped working.
I'm suspecting another dead chip somewhere along the chain is putting bogus inputs into the adder,
or perhaps the adder itself is dead -- looks like there's a shorted bit from bit 11 to bit 0 -- just the kind of
thing you'd expect if the gating was hosed on the SHIFT RIGHT or SHIFT LEFT logic.
Well, the M220B board is now fixed.
E18 (7453) was the culprit.
This one bothered me a fair bit -- I went to a lot of trouble to find it.
After drawing the schematic, there were a number of places where the problem could be.
The symptom was that a LOAD ADDR worked just fine for each and every bit, but as soon as I hit
DEPOSIT, one of the PC bits would light up -- even when it wasn't its turn to.
If you have the schematic handy, you'll notice that E18 is the gate between the PC output bit and
the adder.
S
M
L
XL
This is the test setup I used to diagnose the M220B card.
Boy those extender cards sure come in handy!
I used the scope more or less as a fancy voltmeter, as I was mainly interested in probing the
static levels on the various gates on the M220B card.
|
Turns out, E18 was broken, and would assert the bit.
This one was difficult to diagnose, because just measuring the voltages with a scope didn't
show the bit as being stuck.
It must be a combination of factors that yields the stuck bit.
(I managed to get a 7453 with a datecode of 7314 as the replacement part, how's that for vintage!)
In any event, now all of my M220 cards work (as far as I can tell :-)).
Obviously, not quite -- I just replaced E8 (another 7474) on another M220 card; the symptom was
that the Memory Buffer (MB) light for that bit would not light up.
Setting the flip flop by grounding the PRESET pin would cause the light to light up,
but then when ground was removed from the PRESET pin the light would go out again.
The replacement was a chip from 1981.
|
