Chip died: Have you had this happen before?

[Hans Wedemeyer]

I have two prototypes, one is constantly connected to the debugger (ICD-U40), I call this the debug unit, and the other gets programmed every so often to do comparison testing.

Today the debug unit works if connected to the IDC-U40 but will not run stand alone.
Oscillator is working at correct frequency and amplitude.
It is a 3.3V system.

I concluded the chip must be bad. I changed it (no easy task 64 pin TQFP) and after that everything is back to normal.

I have noticed the unit can take several 10's of milliamps sometimes when the ICD-40 is doing stuff.

I don't accept it is just a bad chip, and with humidity around 70-90% here in Houston, it's hard to generate any static, added to which the work stattion is fully protected.

Any idea what could have happened ?

[Mark]

Well I have had all sorts of fits with the 18F6720's. At least the early ones. They would work in debug but not in stand alone. They wouldn't function at all or would just reset. Sometimes they appeared to work until you executed different parts of the code and strange things would happen. I could change a line or two of code and the problem went away. Put the same code in another unit and it worked great. The problem was with the earlier chips. I haven't seen it in the newer ones. If the date code ends in 04 then I haven't seen the problem, only the early 03's were the ones that I found to be bad. I have changed many of those 64 pin TQFP's. It is a lot easier with a surface mount work station, a little paste, some flux, and a microscope