Are you sure that a watchdog solved the problem? Or was it a simple clock which periodically turned-off and turned-on the power supply?

A disadvantage of a watchdog that only provides a reset is, that it cannot recover circuitry from typical lock-up. Only a full power-down followed by a power-up can help then. So, a better watchdog would provide a power-down and power-up cycle, instead of a simple reset. This could be combined with a very well defined power-up Vcc slew rate, which is essential for many todays µC. They get problems if Vcc rises too slow or too fast. The power-down and power-up cycle could also guarantee that Vcc goes down to 0V, which is also essential for many todays µC.


These discussions always suffer from the different point-of-views of people and the somewhat different experience they have in this area. If you tell me, that you zapp with an ESD-gun to the µC's port lines and the part lock-ups all the time, then I would not recommend you the use of the internal watchdog, because then there's a high probability, that the watchdog locks-up as well. In such an application it would make way more sense to immunize the port lines by ESD filters, wouldn't it?

This does not mean, that a watchdog is always useless, of course. I do have it in my circuits too, because I HOPE that it might be useful in some instances. But if I have the chance to immunize the circuit by proper filtering, shielding and grounding, of course, I will go this way (and still have the watchdog...).

And yes, you are totally right, when you say, that filtering, shielding and grounding cannot solve all problems.

Kai Klaas