Thanks Everyone,

The problem was indeed with my pullups.

I spent some time recently improving my prototype setup. I now have the ability to put solder mask on my circuit boards, and more importantly I can now make my own paste stencils. In the past I was able to put boards together without these, but lately I've found myself spending more time repairing the damage I did than assembling the boards. So now I assemble them by the pick and place method followed by reflow in an oven.

These were the first boards I made with this new process.


So what was the problem? There were four pull up resistors, two pairs, that were shorted together on the pin side (two separate short circuits). The net result was that with either of these pairs, if either of them was set to zero, they both went to zero. They were sort of ANDed together. The only way to get a one on either was to have a one on both.

And of course neither of these shorts was visible to the naked eye. They happened because I had to try and place 0603 chips with tweezers, and in two of eight cases I was just enough off to bridge over to a via (not covered by the solder mask).

Anyway, it's all fixed now and all is well once again. Thanks everybody.