The 50 or so PLCs installed where I work are all wired using 120 VAC control circuits. Limit switches, manual switches, proximity switches and such all source 120 volts to their particular PLC inputs. When not closed, the wire to each input floats (actually, it is capacitively coupled to the PLCs' modules' common terminal - 120 V neutral, in our case - through the modules' input circuit). The PLCs are installed in remote cabinets, off the production floor.
The operator controls' environment is wet, so there have been occasions that a water short across a switch's terminals conducted enough current to turn on an input. This condition is made apparent by a dimly lit annunciator LED on the input module corresponding to the partially shorted switch. Lately, some inputs have been turning on without the tell-tale dim LED. With a 'scope, we traced that to intermittent electrical noise spikes on the input wire induced by a new style of frequency drive we began using.
It's clear there is a design flaw in the control circuits for the environment in which they operate. (My bad). I believe the solution is to add a contact block to each manual switch such that the input is grounded in the switch's open state. Mechanical limit switches of the type we use typically have DPDT contacts, and they can be rewired to ground the signal wire when not actuated. Electronic proximity switches are more problematic, as we usually use 2-wire AC switches. I plan to load the inputs connected to the proxes to reduce the chance of water short actuation.
I don't know of cheaper solution to the problem. Water in a food plant is a fact of life, whether it comes from high pressure hoses or from condensation during rapid and drastic temperature changes. We have used petroleum jelly (food grade acceptable) on the switch terminals with limited success. Turning a switch's contact block such that there is no horizontal plane between input and output doesn't work due to the bridging film of condensation that forms after each cleaning shift. Maintaining sealed, pressurized operator control boxes is prohibitively expensive and time consuming. I did toy with the idea of using magnetically actuated, hermetically sealed reed switches for manually inputs, but there are limitations in available contact arrangements. And that would be a lot of magnets. :)
I just thought I'd pass this on and invite comments.