Trust never any signal from outside until you checked it is trustable !

At first check, how long was the way to catch noise until it reaches the µC.
If there was to big noise, try to reduce it e.g. by lower impedance (lower pull up resistor value), opto insulation, differential wiring and so on.


Then check the signal duration, signal interruption and signal reaction time.

E.g. on a human, which pressed a button, a signal duration and interruption not faster as 100ms seems trusty. Furthermore a reaction not faster as 500ms looks fast on a human view.

Then a debouncing intervall of 100ms, which check 4 identical readings every 25ms, was trusty enough in this case.

Also every signal, which was not given from a human, but connected over long lines, must be checked in a simliar way to be trusty.


Peter