true, but coding errors aren't processor faults.

The ultimate misconception is that a watchdog will make your code 'better'.

Watchdog, and other means of detection, will react due to software design errors 75% of the time, ESD 15% of the time, hardware design errors 9% of the time and hardware faults 1% of the time (unless, of course you have a flying lead thrown together piece of crap).

So, the protective means are not active often as to detecting a bum transistor, but a whole lot detecting a bum deigner (and ESD faults).

This discussion has run off track by discussing protective means as protection against unforseeable event, when in truth they most often catch problems that should have been forseen.

Just take the "if an EDD event happens, the watch dog will activate" approach BALONEY, first, if an ESD event can penetrate and yout thingy has any importance, the design is bad and second, what guarantee do you have that an ESD even will trigger your 'protector'.

Yes, watchdogs DO serve a purpose, but if you ever consider the watchdog as something that WILL happen, your design is faulty.

Erik