F#&*ing EEPROM again???

[Byron G. Jacquot]

>The whole system began to fail 2 days ago. Yesterday it just refused to
>boot up. Or at least it refused to get past it's initial LED test.
>Thinking that the problem was something to do with the front panel, or even
>the PSU which is still blowing close to the wind, I pulled the whole thing
>apart.

Quick PSU related question:

Do you have a brownout detector?  I've had troubles where power transitions
do funny things to EEPROMS.  If the system isn't held by having reset
asserted during the power transitions, it can write gibberish into the
EEPROM.  As the power rails ramp up or down, the program counter can wind up
places it shouldn't otherwise be, with invalid data.

AVRs are notoriously bad abut this, if that's what you're using.  I'll bet
others have similar troubles.

>It would appear that as soon as the micro accesses the EEPROM for the first
>time, the whole system falls over.

Are you validating the data in any way?  CRC or checksum?  At least spitting
out invalid values?

This might get you started.