Steen:
UPDATE!!I removed the #include <PinChangeInt.h> and all references to the lib ( the attachinterrupt() ) and now it's running smooth for about 7 hours, without port change.
Could it be that this lib somehow messes with ports 0 and 1, so the serial communication gets interrupted or something? i took a look at the lib, but i didn't really understand a lot of it.
i'll see if that lib really is the reason to the breakdown, by adding it tomorrow and see if it fails again. if so, is there another way to check the rising of any port by some kind of interrupt, or will a digitalRead() be needed?
If this is happening in Windows, something very weird is happening, even on physical reconnections windows will try to use the same com port number