🚀 go-pugleaf

a funny thing happened to me today... i rebuilt my 5.25" drive interface
thing (previously 3 ICs hanging off the end of a ribbon cable, supported
only by tack-soldered wire wrap wire) to live on a perfboard, complete with
connectors for everthing (power supply filtering, even!).  anyways, after
it got working, i decided to take a look at the +12 and +5 supplies going to
the disk drive.  of course, the point i chose to look at had both supplies
right next to each other on a header connector - slip, ZAP! (got a nice spark).
the amiga drives gronked (sounded like a recal), but the machine seemed
otherwise was unfazed.  i rebooted just in case, and things seemed ok.
after playing with MOUNT DF2: and pcutils for a while and happy that my
power supplies weren't overly noisy, i decided to call a few local bbses.
for that, i use Access! - i use Diga! for usenet cuz it knows vt100 pretty
well.  anyways.  access! comes up and after the screen is up, it goes to
df0: (presumably for serial.device).  task held.... the guru number varies,
but it's either all-zeros with a 3 on the end, or a B on the end (address
error or line-F opcode).  puzzle a bit, try again.  same thing.  try diga!.
no problem.  after diga!, try access again.  works.  reboot.  try access.
fails.  reboot.  try dir >ser:.  fails.  reboot.  try amigabasic terminal
demo.  fails.  reboot.  try online!.  works.  reboot.  you get the picture.
only Diga! and online! seem to like the serial device now, but the others
are happy once diga! or online! have blazed a path to the device.  i ran
my insider ram test, it was happy (though i don't really trust it - testing
testing 1 meg of ram should take more than 30 seconds).
i have run lots of programs, tried 1.2 (all above was 1.3 with FACCII),
tried FACClessness.  same conditions.  tried rebuilding my 1.3 disk and
access! disk, but no change.  i have run lots of programs that eat memory
and do alot of work (games, compiles, large text edits, etc), and none
of them fail.  the condition seems to be to try to get the serial.device
when the moon is in the right phase.
anyone have any ideas?  if i can find someone willing to play the chip swap
game until i find it, where is a good place to begin?  as i remember, the
actual uart is inside one of the custom chips and the non-tx/rx lines go
to one of the 8520's.  (oh, and this is a 1000 - forgot!).
  --Scotty