The Restoration of the PDP-8/I Minicomputer (PAGE 11)
I finally found the problem with the M220 card and the rotates.
Turns out that there's a transistor on the M220 version B card that I didn't
put into my schematic, because it wasn't on the version A card, nor was it connected to any pins that were documented on the version A card.
Therefore, it was missed.
This transistor is connected into the inputs of a bunch of the AND/OR/INVERT gates on the M220B card.
It was dead.
The part number was "f3009" which I assumed is a 2N3009 -- not having any in stock, I replaced it with a 2N2222 and it works just fine!
I found this particular problem transistor because a good card had a different logic level coming out of the transistor than the bad card.
From then it was a simple matter of tracing out the circuit and determining that the transistor wasn't ... er ... "transisting" :-)
The symptoms of the failure were that the two AC bits that the M220 card is responsible for would correctly rotate a zero in,
but the zero would remain stuck. Putting it into tabular form:
|Original Value ||New Value ||Actual
|0 ||0 ||0 (correct)
|0 ||1 ||0 (bad)
|1 ||0 ||0 (correct)
|1 ||1 ||1 (correct)
Once the rotate problem was fixed, I returned back to the memory section of the project because
I wanted to ensure that I had a solid base from which to work.
So I created a webpage of tiny little test programs, available here.
Using an M113 card, a buddy and I built a
W076 TTY interface card, and made it compatible with RS-232 level signals.
Unfortunately, we didn't invert the RS-232 signals, so I need to add a few more components to get the levels right :-)
I have completed repairs on the front panel; all of the bulbs are now working, and the LEDs have been
replaced with genuine light bulbs. There were two bad transistors on the front panel board as well,
and these have been replaced with 2N2222s.
I successfully got the TTY output to work today!
There's a test program that writes data from the SR and another one
that writes all character values.
The funny thing was that the baud rate was 90 baud instead of 110 baud -- I'll have to poke around on the M452 to see why
it's not generating the correct baud rate.
20031124 -- adjusted pot on M452 and now have 150 baud. Tried to crank it to 300 baud but the oscillator circuit fails at
the higher clock rates; the nice square waves being output at J2 on the M452 turn sloppy. So, I can live with 150 baud
And there it was, just running along, everything working nicely, when all of a sudden, the AND light comes
on and the MB goes blank. Turns out that it no longer lets me deposit or examine core! All locations return
I guess I'll need to take a look at the power supply levels, the drivers, X and Y selects, and all that good stuff
that makes the core work. And it had all been going so well! :-(
The scary thing about this failure is that it's the same failure mode as one or two of the other core planes; in that
case, however, I was able to swap core planes. Now, however, I'm fresh out of spare (working) core planes. I really
really really hope that it's not a dead core plane...
And now, a few hours later, it all works again! The core plane is not dead, just some circuit that "works and plays with"
the core memory is flaky.
M720 perhaps? The symptom was that the RUN light came on, and the PC was stuck at 0001.