Help needed with 32K PET 2001 - makes a chessboard pattern on startup

[kiwwy]

I couldn't spot any visual damage around the CPU socket, so I replaced the socket.
With the new socket, it was very dificult to get the expansion board seated, it has thicker pins than usual. But finally managed to get it in without breaking something else. Should probably have done it with the mainboard out on my table instead of mounted in the case, very dangerous when it doesnt have any support in the middle.

But it works great now. Can even load and save with the Datasette :D

I kinda like the chicklet keyboard actually. Did anyone ever write a word processing program for PETs?
I noticed that the keyboard connector in my VIC-20 is identical to the one in my chicklet PET. They wouldnt be compatible, would they?

 

[Nama]

errr..well that was easy!

 

[carlsson]

On behalf of Nama, I looked up this data a couple of months ago:

Pin | C64 | P8296 | C64 Function
----+-----+-------+-----------------------------
 20 | PA7 | KIN 0 | col #7 / "1" -> paddle set B
 19 | PA1 | KIN 1 | col #1 / joy #2 down
 18 | PA2 | KIN 2 | col #2 / joy #2 left
 17 | PA3 | KIN 3 | col #3 / joy #2 right
 16 | PA4 | KIN 4 | col #4 / joy #2 fire
 15 | PA5 | KIN 5 | col #5
 14 | PA6 | KIN 6 | col #6 / "1" -> paddle set A
 13 | PA0 | KIN 7 | col #0 / joy #2 up
 12 | PB0 | DEC 0 | row #0 / joy #1 up
 11 | PB1 | DEC 1 | row #1 / joy #1 down
 10 | PB2 | DEC 2 | row #2 / joy #1 left
  9 | PB7 | DEC 3 | row #7
  8 | PB4 | DEC 4 | row #4 / joy #1 fire
  7 | PB5 | DEC 5 | row #5
  6 | PB6 | DEC 6 | row #6
  5 | PB3 | DEC 7 | row #3 / joy #1 right
  4 | +5V | DEC 8 | +5V
  3 | RST | DEC 9 | RESTORE key
  2 | -   | -     | N.C. (polarity mark)
  1 | GND | GND   | GND

Now, I'm not entirely sure the 8296 and 2001/3000/4000 internal keyboard connectors are identical but to a big part I think they are. As you can see, the PA (col) and PB (row) numbers are flipped around on the VIC/C64 pinout, as well as the keyboard expects pin 4 to deliver 5V and pin 3 (Restore key) to be wired directly to NMI.

However with a bit of crafty rewiring, I'm sure you could get a VIC-20 keyboard mostly working in a PET if you can live with not all keys are present.

 

[fachat]

For what it is worth, the checkerboard pattern is one graphic character as well. In your case I think you get 255 ($FF) all across screen that displays as a checkerboard pattern. If you get a solid screen of another character without the CPU + RAM board.. hm, yes maybe it makes some difference.

If the screen comes up with garbage, and then gets reset to the checkerboard pattern, that would make me assume that:
a) the CPU seems to be working in that it runs the reset routine, writing stuff to the screen (which means working ROM, some low 1k RAM)
b) working screen memory
c) some problem in the data path from the CPU to the screen memory, so that when the CPU writes _anything_ to the screen memory, the screen memory only sees "unconnected" signals, which is interprets as high (due to pullup resistors), so the screen memory is written with $ff

IIRC there are some '244 buffers in that path, you may want to check them if you have a two-channel scope with separate trigger. Trigger at the enable signal, one channel the input and the second channel the output. You should see the same signal on both sides if it works correctly.

André

Edit: Bummer - I was late to the party ;-) Congrats for fixing the PET!

 

[sjgray]

On behalf of Nama, I looked up this data a couple of months ago:

Pin | C64 | P8296 | C64 Function
----+-----+-------+-----------------------------
 20 | PA7 | KIN 0 | col #7 / "1" -> paddle set B
 19 | PA1 | KIN 1 | col #1 / joy #2 down
 18 | PA2 | KIN 2 | col #2 / joy #2 left
 17 | PA3 | KIN 3 | col #3 / joy #2 right
 16 | PA4 | KIN 4 | col #4 / joy #2 fire
 15 | PA5 | KIN 5 | col #5
 14 | PA6 | KIN 6 | col #6 / "1" -> paddle set A
 13 | PA0 | KIN 7 | col #0 / joy #2 up
 12 | PB0 | DEC 0 | row #0 / joy #1 up
 11 | PB1 | DEC 1 | row #1 / joy #1 down
 10 | PB2 | DEC 2 | row #2 / joy #1 left
  9 | PB7 | DEC 3 | row #7
  8 | PB4 | DEC 4 | row #4 / joy #1 fire
  7 | PB5 | DEC 5 | row #5
  6 | PB6 | DEC 6 | row #6
  5 | PB3 | DEC 7 | row #3 / joy #1 right
  4 | +5V | DEC 8 | +5V
  3 | RST | DEC 9 | RESTORE key
  2 | -   | -     | N.C. (polarity mark)
  1 | GND | GND   | GND

Now, I'm not entirely sure the 8296 and 2001/3000/4000 internal keyboard connectors are identical but to a big part I think they are. As you can see, the PA (col) and PB (row) numbers are flipped around on the VIC/C64 pinout, as well as the keyboard expects pin 4 to deliver 5V and pin 3 (Restore key) to be wired directly to NMI.

However with a bit of crafty rewiring, I'm sure you could get a VIC-20 keyboard mostly working in a PET if you can live with not all keys are present.

The PET has an 8x10 matrix and the VIC/64 have 8x8. You can match up the row and column connections but it won't help as the actual keys are in different positions within the matrix. You would have to edit the PET's keyboard matrix table (in the Editor ROM) to get the keys to work. That would be fine if all programs used the kernal routines for keyboard input but would fail if you ran a program that scanned the keyboard itself and expected a certain key at a specific row/col...

Steve

 

[carlsson]

By the way, this place in San Francisco that even after the second round of grab party had plenty of PETs, shouldn't they also have had a lot of spare or broken parts unless that was cleaned out on beforehand? The place I used to go to had a fair deal both of complete (working and non-working) machines and loose keyboards, motherboards etc.

 

[Eudimorphodon]

By the way, this place in San Francisco that even after the second round of grab party had plenty of PETs, shouldn't they also have had a lot of spare or broken parts unless that was cleaned out on beforehand? The place I used to go to had a fair deal both of complete (working and non-working) machines and loose keyboards, motherboards etc.

There actually wasn't much in the way of loose PET parts lying around. If you wanted parts you took more PETs. (Some of which *were* missing parts. It seemed like about half of the non-N 2001s were missing motherboards. Perhaps the motherboards were all heaped in a box that wasn't uncovered before I left.)

 

[frank_DK]

Take a look at this link, and start at #13. I think this could help you.
http://www.vintage-computer.com/vcforum/showthread.php?24898-Frank-s-Danish-PET/page2

 

[kiwwy]

The PET has an 8x10 matrix and the VIC/64 have 8x8. You can match up the row and column connections but it won't help as the actual keys are in different positions within the matrix. You would have to edit the PET's keyboard matrix table (in the Editor ROM) to get the keys to work. That would be fine if all programs used the kernal routines for keyboard input but would fail if you ran a program that scanned the keyboard itself and expected a certain key at a specific row/col...

Yeah, I suspected it wouldnt be so easy. Just was suprised to see that it used the same connector. But I guess they recycled what they could when they designed the VIC-20.

 

[kiwwy]

There actually wasn't much in the way of loose PET parts lying around. If you wanted parts you took more PETs. (Some of which *were* missing parts. It seemed like about half of the non-N 2001s were missing motherboards. Perhaps the motherboards were all heaped in a box that wasn't uncovered before I left.)

How rare are PET machines actually? The only one I've ever seen in real life is the one I have. I could imagine that later models are more rare, if they didnt sell so well.

 

[carlsson]

Haha, PET was one of the biggest business machines in the late 1970's. However due to their relative weight and size, I'm sure a large number got scrapped as they stopped working. Five years ago, they were considered semi-uncommon, but as I mentioned elsewhere, well over 200-300 machines have surfaced in the last few years, machines no collector could expect to exist. If something, the original PET 2001 with chicklet keyboard and built-in tape recorder is the rare one, next to SuperPET 9000 and some of the rounded off SK models late in production. Most of them however cost more to ship than what the market value is.