The error is invariably as follows in any of the failed configurations. It appears that the programmer is failing to handshake with the on board AVR.

avrdude: Using SCK period of 10 usec
CMD: [ac 53 00 00] [00 00 00 00]
CMD: [ac 53 00 00] [00 00 00 00]
avrdude: initialization failed, rc=-1
Double check connections and try again, or use -F to override
this check.

Well, here's one thing that you can try: Put extra LEDs in the relevant holes-- loosely, not soldered down-- and see if that makes a difference.
(I cannot see how it could, but....)

Tried that. Loaded the good board up with a boatload of extra LEDs across the rows that flash during programming and other random rows, too. Couldn't get it to fail.

And we have both been over the board that won't program a bunch of times. All connections look solid and there doesn't seem to be any random semi-shorts.

Chips have been re-seated across both boards (swapped, too). Batteries have been swapped. Tried a bench supply.

And, ultimately, the most totally whacked out bit is that Ben's programmer wouldn't work on either board with the 10 pin to 10 pin cable attached -- except for one brief period where it did work. The only difference between Ben's programmer and mine is that his 7400 series buffer chip is soldered directly and mine is in a socket.

Some voodoo noise issue??!?!?!

Baffled. At this point, I have a viable solution -- one board does work for in circuit programming. However, I'm terribly curious as to what the hell is going on!!!

Quote by: bbum

Baffled. At this point, I have a viable solution -- one board does work for in circuit programming. However, I'm terribly curious as to what the hell is going on!!!

Yup, viable, but sucks.

The 10-pin issue does lead me to suspect some sort of voodoo noise issue, but it seems very unlikely to me that the noise properties of the two boards are so very different. One thing to try on that front: lower the bit rate by using the option -B 32 in avrdude (see http://www.ladyada.net/make/usbtinyisp/avrdude.html ).

I really do suspect that there's a bad connection of some sort on the one board. Something hard to find, apparently. With the chip *out* and power off, measure the resistance between each of the following lines and (1) ground and (2) VCC: SCK, MISO, MOSI, RESET.

You should see: RESET- GND: no connection, RESET - VCC: ~ 5 k.
MISO, MOSI, SCK to ground or VCC: Each > 1 k.