neo is dead, what should i do?

[esw at alum.mit.edu]

> > It worked!!!
> > I putted the battery out for at least an hour and tested another battery
> > and my neo was totally dead. now, after ~13 hours without battery
> > inserted it is again alive and fully operational. :-)
> > How can this be?
> 
> This could be the most likey cause
> http://en.wikipedia.org/wiki/Memory_effect

I hope that's a joke.  The memory effect has to do with a NiCd
battery's ability to hold a charge after being regularly partially
discharged.  Nothing to do with this problem.  At all.

My guess is that the power management microcontroller has some dead
end states in its state machine and there's either an ultracap or just
a normal capacitor that can end up supplying power to that micro for
an awfully long time.  Either that microcontroller is also the same
chip that does the real time clock, or at least they're sharing
standby power... I think I remmeber noticing my neo keeps its hardware
clock time over short/medium long periods of time with no battery, but
then with the overnight drain the hardware clock got reset as well.

The failure mode reminds me of classes on state machines where they
told us to make sure that even states that are "impossible" to get in
to should have logic to make sure they are transitioned to a sane
state.  I'm guessing that some low level hardware is getting put in a
state it never expected to be put in, and the state transition logic
never expected to be in that state so theres no deliberate transition
out of it and it just gets stuck there.

-erik