Do you mean the shields are connected to the machine and the controller? If they are connected at both sides this will be the cause. This is where optoisolation or electrical isolation between your computer and machine would have been bypassed. Also make sure your limit switch wiring is not bundled together with power or stepper wiring. This can lead to false triggering also, from induction I think is the term. The debounce interval in the general configuration page can be lengthened to say 1000 as a test. If that helps start lower it to 500 and less. The way I understand that option is that it only triggers if the signal is constant for the specified period of time, if it is fluctuating during that period it would be disregarded as noise. I know the debounce circuit was disabled for something and I cant remember what it was. Someone will chime in if so. Anyway try lengthening the debounce interval dont worry about the spindle debounce as it is for rpm feedback.