AVR-Chat

> I fixed the second problem, it turned out to be a problem with the oscillator - it was not set to "full swing" and an LED power supply near the processor was causing the oscillator to stop and do unpredictable things. Very strange symptoms though.


From a better console:  I've seen the low power oscillator causing
problems since it's introduction on the mega128 (which had it as the
default mode!)    The problems generally appear as flaky bugs that you
can't track down,  Timers or baud rates may be slightly slower than
they should be.   It apparently will work if you carefully select for
crystals that require low loading capacitance values.


I've seen the debugger doing things like:

ldi, R16,0
(now R16 has some non-zero content)


If you're sure that you're running the rail-rail oscillator, and that
VCC etc is in spec, then I'd say you have a bad part most likely.
Definitely I've never seen a case where a register bit was "stuck", in
any AVR, in the last decade +.