Inductive Proxes
The catalogs sell you an M18 flush mount sensor saying it has a 8mm range. Good luck with that. You might get 3mm if the target is aluminum. People forget about reduction factors all the time. A maintenance guy will swap out a steel bracket for an aluminum piece because it was lying around the shop. Suddenly your sensor goes blind and nobody knows why because "nothing moved".
Also mount them tight.
Photo Eyes
I hate these things. We had a conveyor counting boxes with retro-reflectives. It ran perfect all day. But at 3 AM the line would just stop randomly. Took me a week to figure out that the temperature drop in the warehouse at night was causing micro condensation on the reflectors. The beam would scatter. I didnt buy a better sensor I just angled it down slightly so the moisture wouldnt pool up. Manuals dont teach you this stuff.
Capacitive Sensors
They measure the dielectric of whatever is in front of them. Guys love using them to check liquid levels through plastic tanks. It sounds great until the humidity spikes or the liquid splashes and leaves a thick film on the inside of the tank. Then the sensor just latches on and stays on forever. You try to dial the pot down and then it stops reading the liquid completely.
If I can avoid them I do. Just punch a hole in the top and bounce an ultrasonic down.
Ultrasonics
They use sound waves. So if you point one at a pile of soft powder like flour the sound just gets absorbed. No echo back to the sensor. You have to actually know what the product is made of.
NPN vs PNP Wiring
This drives me crazy. The absolute worst mistakes happen when someone replaces a dead sensor.
In the US everything is mostly PNP. Sensor sends 24V to the input card.
But you buy a used machine from Taiwan and suddenly you are dealing with NPN which switches ground.
A sensor gets smashed. The new guy grabs a PNP off the cart and wires brown to 24V and blue to 0V and lands the black wire on the input. The box goes by and the sensor LED lights up... but the PLC input stays dark.
The input card is waiting for a ground signal but the sensor is hitting it with 24V. Check the drawing. If no drawing meter the common wire on the input block. If common is 24V you need NPN.
Temperature
People use RTDs everywhere because someone told them they are accurate. RTDs have a fragile little resistor inside. Mount an RTD on a pipe that shakes hard when a valve closes and that element will snap in a month.
Just use a Type J or K thermocouple. Yeah you lose a tiny bit of accuracy but they are literally just two wires welded together. They survive almost anything.
Cables and Connectors
Turck and Brad Harrison M12 cords. Everyone assumes if the sensor light is on the cable is fine. Not true. I spent 8 hours tracking an intermittent fault on a gantry. The sensor was constantly moving and flexing the cable. The copper inside the PVC jacket eventually snapped. The ends were still touching enough to pass current but when the machine vibrated it would break connection for a millisecond. PLC catches it and halts the whole line. The cable looked brand new on the outside. Always bypass the cable to test if you are stuck.
IP Ratings
IP67 is a joke in food plants. It means it survives temporary water. But the sanitation guys come in with 1000 PSI hoses and 160 degree water mixed with caustic chemicals. They blast everything. That water blows right past the rubber O-ring on an IP67 connector. The pins corrode into green fuzz and short out. Buy IP69K rated stuff for washdown areas and pack the pins with dielectric grease. Or you will be replacing them every weekend.
Electrical Noise
You install a brand new 4-20mA pressure transmitter. It reads perfect. Then the 100HP extruder motor kicks on and your HMI reading jumps from 40 PSI to 90 PSI randomly.
Thats EMI. Someone ran your unshielded 24V sensor wire in the same tray as the 480V motor leads from a VFD. The VFD screams electrical noise and your sensor wire picks it up like an antenna. Pull the wire out. Run shielded cable. And only ground the drain wire at the panel side. Grounding both ends creates a ground loop which makes it worse.
Limit Switches
You think mechanical limit switches are fool proof but they arent. We use heavy duty limit switches on overhead crane rails. Over time the mechanical lever arm gets loose on the splined shaft. The crane hits the arm, the arm moves, but the internal cam doesnt rotate enough to click the contacts. So the crane just keeps driving until it hits the hard stop. You have to actually check the setscrews on the arms during PMs. Also water gets into them. A switch is rated NEMA 4 but after 5 years the gasket rots out and it fills with water. Then you get a ground fault that takes down the whole 120V control circuit. Finding a ground fault on a machine with 50 limit switches is a nightmare. You end up unlanding wires one by one for hours in the main panel.
Encoders
Rotary encoders on conveyor shafts to track product position. A guy will mount a solid shaft encoder directly to the conveyor roller using a rigid coupling. Huge mistake. The conveyor roller always has a little bit of runout or wobble. That wobble gets transferred directly into the encoder bearings through the rigid coupling. The bearings shatter in three months. You have to use a flexible coupling or a hollow shaft encoder with a tether arm to absorb it.
Also dont run encoder cables near servo motor cables. The high frequency noise will induce extra pulses on your A and B channels and your PLC will think the conveyor is moving way faster than it actually is.
Safety Light Curtains
Operators hate them because they slow them down. I caught an operator who taped a piece of cardboard over one receiver so it always saw a safe signal. If you do PMs you have to actually push your hand through the beam and listen for the safety contactor to drop out. Looking at the green light on the sensor means absolutely nothing.
