Adventures in board building

My part of yesterdays meeting was to stuff one of the new
printed-circuit AID controller boards, and, with about two
exceptions, it went pretty well.

Holes in the board were undersized for the 2x5 pins that
select Rx source, and for the 7805. Drilling them one size
bigger with the Dremel fixed that.

I drilled the two holes that stake down the DB-9 socket
to 1/8". This worked, but slightly smaller holes might work
better.

My innovation was to drill 1/16" holes next to the pads
for the 9v battery connector, being careful not to break
copper traces. This let me snake the connector wires up
from under the board through these holes, and solder the
connector wires through the board from the top as usual.
This protects the wire from too much flexion where it's
soldered through the board. The 9v connector might not
break off so soon.

When I powered the board up, there was no smoke, but the
PIC processor was stone dead. Logic probe showed voltages
OK, but the clock was not clocking. The oscilloscope
confirmed that. On close inspection, it seemed that one
solder joint to the crystal might not be very good..
touching up the joint made the PIC work.

Connected to AIDTester.pyw, the board displayed "hello",
but sent no other output. After many people were consulted,
we decided that Rx input wasn't reaching the PIC, and it
turned out that one of the pins of the 40-pin DIP socket had
missed its hole, and folded under the socket. Pulling out
that pin and replacing it with a pin taken from a dud socket
fixed that.

From that point, the board worked with AIDTester just
fine.

It was lucky that the bad pin was the Rx pin. Anything
else would have been a lot harder to diagnose.

It was lucky that I had a logic probe and the scope to
confirm my clock problem.

It was lucky that Rob and Stefan were there to confirm or
deny my hunches about parts placement on the new board
versi