On 7/15/06, Roberto Bazzano <r.bazzano at ulm.it> wrote:
I'll try to replace 2114 video ram: on three pet I
repaired they were
defective.
While 2114s are notorious, it doesn't sound to me like the behavior he
is describing has anything to do with video RAM. "Garbage characters"
on power-up is a normal state. What's also normal is that less than 1
second later, as the tube is warming up, the CPU clears video RAM.
That's the step his machine is stuck at.
The "problem" is that there's no one cause for having a frozen PET
with garbage characters on the screen... Anything that interferes with
the CPU accessing RAM or ROM will do that. As other posters have
mentioned, different failures will cause different types of crashes at
different points, but the visual effect is the same.
If I were scoping a problem at that level, I'd pull out my Fluke 9010A
and the 6502 pod and go to town; not everyone has one of those, but if
you do, it's a really handy tool at this level of debugging - built-in
RAM tests, ROM checksums (once you define where ROM is and what the
expected checksums are), bus tests (including shorted bus-lines), etc.
If you had or could build one of the modernish ROM/RAM replacements,
it would go a long way to eliminating swaths of circuitry. I have one
at home made by one of the cbm-hacker members... it's a daughter card
that sits in the CPU socket. You pull the RAM and ROM from the
machine, move the CPU to the daughter card, install a compatible ROM
image to the single EPROM/EEPROM, set the DIP switches accordingly,
and boot it all up. If it were not to boot, then the problem is
likely some sort of stuck bus transceiver or total I/O selection
failure. If it does, then you could re-install the ROM or RAM first
and at least get an idea where the problem is via the time-tested
divide-and-conquer method.
-ethan