|
|
The Restoration of the PDP-8/I Minicomputer (PAGE 6)
Now that the front panel is complete (yes, I ended up using LEDs in the short term; but
I have a quantity of official DEC PDP-8/I lamps waiting to be plugged in) I noticed
that the LOAD ADDR and DEPOSIT switches don't quite work.
The extended memory works fine with LOAD ADDR -- all the bits on the front panel correspond
with the bits set in the switches.
However, the regular (12 bit) addressing has some stuck bits, and (now, for the mystery)
counts backwards when you hit DEPOSIT!
2003 07 05
I finally figured out what this was all about.
Due to an existing problem where bits in the Accumulator are gated in to the adder logic,
when the Accumulator had all ones in it, this would cause the "backwards counting".
A bug, but still neat :-)
| |
After the machine warms up for a while, it starts counting forwards.
(2003 07 05 -- Turns out that it had nothing to do with "warming up" but just luck as to the contents of the Accumulator.)
Another mystery is in the execute cycle.
If I do a LOAD ADDR with 001/001 for the DATA FIELD and INST FIELD, and
hit START, the RUN light comes on and it looks like it's trying to execute some
program.
However, using 000/000 for the two fields doesn't turn on the run light.
I'm suspecting a memory problem with the first 4k of core, just off the top of my head.
It's at this point that I decided to exhaustively test all of the cards in the PDP-8/I.
|
Cards Tested -- 2003 06 29
Ok, so now I've tested all the single-height cards that were simple -- and things are
working a little bit better.
However, I have diagnosed that one of the M220 cards is hosed.
The schematics on the web are for an M220A,
whereas mine is an M220B.
So, I'm busy drawing the schematics for the "B" version.
This is an exhausting process of getting all the traces on both sides of the card
into a text database, and then connecting all the wires in a schematic drawing package.
I cheated a little bit and scanned the card from both sides in my scanner, and then used a graphics
package to mirror the solder-side.
By printing that out, I was able to trace (with different colour pencils) the ground, +5V, and
signal lines.
Each line was entered into a text file (for example, "e1-15 e2-4 e3-7" to indicate that chip E1's pin 15 went to chip E2's pin 4 and chip E3's pin 7).
Then, I used the existing M220A schematic as a base,
and drew enough of the M220B schematic to get started testing.
These are the two pictures that I used to create the schematic.
Of course, the original was only in black and white.
Using a graphics editing package, I was able to "mask out" most of the colours, and all I was left with
was the grey shade that corresponded nicely with the traces and the pins.
Also, notice how the two pictures line up.
If you're detail oriented, you'll notice that every feed-through hole has a designation of the form "X" and a number.
This allowed me to construct the wiring listing without having to follow all the feed-through holes, I just treated them as if they were another component.
I used different colours to indicate +5V (red), Ground (green), and signal wiring (yellow initially, but when I couldn't really see it I switched to orange).
|
|
|
