Installing and Troubleshooting Sensors

Planning an installation

Most “sensor problems” turn out to be installation problems: a bracket that flexes, a cable that rubs, a connector that was never fully tightened, a lens that fogs. Time spent in planning pays back many times. Before mounting a sensor decide:

• What exactly must be detected and with what margin: the worst-case target (smallest, darkest, farthest, fastest, in the dirtiest condition).
• Where the target will actually be, allowing for tolerances and wear, not where the drawing puts it.
• Access: can a technician reach the sensor, see its LED and replace it without removing half the machine?
• What will hit it, spray it, heat it or shake it, and how the cable is routed away from those things.
• How it is wired and where it is tested: a spare connector or a test point saves hours later (see Sensor Outputs and Wiring).
• Spares and standardization: use as few types as you can; label each sensor with its tag, the same as on the drawing.

Mounting and alignment

• Rigid and adjustable. Use a stiff bracket (a thin flat bar vibrates and shifts the switching point) with slotted holes to adjust, and lock the position with a lock washer or thread-locking compound. Mark the final position with paint.
• Keep an operating margin. Design to get the reliable (assured) range, not the nominal one: if the target should be sensed at 4 mm with a sensor of S_n = 8 mm that is 50 percent of its nominal distance, leaving a comfortable reserve for temperature, wear and tolerance (see Proximity Sensors).
• Protect the sensing face. A sensor that is hit by a moving part is destroyed and may not be noticed: put a guard or a stop in front, or mount it so that the target passes beside it rather than towards it.
• Respect free zones and spacing in the data sheet (flush, non-flush, side by side), and keep metal parts away from non-flush and capacitive faces.
• Align optical sensors carefully: through-beam pairs on the same line (use the alignment indicator, or a laser pointer), retro-reflective sensors perpendicular to the reflector (but tilt a polarized sensor slightly for shiny objects), diffuse sensors at a slight angle to shiny objects (see Photoelectric and Vision Sensors).
• Avoid direct sunlight, welding flash and glare into the receiver; a hood or a different position helps.
• Cable strain relief and bend radius: a minimum bend of about 5 to 10 times the cable diameter (more for moving cables), a clamp near the sensor, and a drip loop so that water runs off the cable and not into the connector.
• Torque the connector and the nut to the data-sheet value: M12 connectors hand-tight plus a little; a loose one lets in water, an over-tight one cracks.
• Do not use the sensor as a mechanical part: no stops, no supports, no stepping on it.

Protection: IP ratings and the environment

IP code	Protection	Typical use
IP65	Dust-tight; low-pressure water jets	General indoor machinery, light washdown
IP67	Dust-tight; immersion to 1 m for 30 minutes	Most industrial sensors; temporary submersion
IP68	Dust-tight; continuous immersion to a stated depth	Submersible sensors, tank probes
IP69K	Dust-tight; high-pressure, high-temperature wash (80 °C, 80 to 100 bar)	Food, beverage, vehicle washdown

• Temperature: operate within the stated range. A sensor mounted near a motor, oven or in the sun can be well above the ambient air. High temperatures shorten the life of electronics and LEDs, and lower the sensing range of inductive sensors.
• Chemicals and cleaning agents: check the housing material (stainless steel, nickel-plated brass, PBT, PTFE) and the cable jacket (PVC, PUR, PTFE) against oils, coolants and disinfectants. Coolants and cutting oil can destroy a PVC cable in weeks.
• Weld spatter: use sensors with a PTFE coating or a spatter-resistant face, and weld-field-immune types near resistance welders.
• Vibration and shock: check the data-sheet rating (IEC 60068-2-6 and -27); use lock washers and potted sensors; avoid mounting on a thin sheet that resonates.
• Moisture and condensation: a connector on the underside, a drip loop and a breather or a sealed sensor prevent condensate in the housing.
• Explosive atmospheres: use certified sensors (ATEX, IECEx, Class I Division 1/2) with the right barriers, and follow the installation code.

Noise, shielding and grounding

A sensor signal is small and the plant is noisy: variable-frequency drives, contactors, relays and welders all radiate and conduct interference. Most noise problems are fixed by good cabling practice, not by changing the sensor (see also Cabling, Grounding and Noise).

Practice	Why
Separate signal cables from power cables (motor, drive and mains) in different trays or at least 30 cm apart; cross at 90°	Noise couples capacitively and inductively over parallel runs; distance and crossing at right angles reduce it
Use shielded, twisted-pair cable for analog and pulse signals	Twisting cancels magnetic pickup; the shield blocks electric fields
Ground the shield at one end (the controller) for analog low-frequency signals; at both ends (with a good bond) for high-frequency or very noisy installations	One end avoids a ground loop current that drives noise through the shield; both ends give better screening against radio-frequency noise
Terminate the shield 360° in a clamp, not as a “pigtail”	A pigtail adds inductance and loses the screening at high frequency
Suppress inductive loads (relay and solenoid coils, contactors) with a diode, RC snubber or varistor at the coil	The switching spike is the commonest source of false pulses (see Inductors, Transformers and Relays)
Put a ferrite or filter on a noisy cable; use a drive output filter and a shielded motor cable	Reduces common-mode currents
Use a good, short earth and bond the cabinet and the machine frame	Gives noise currents a safe path away from the signal ground
Prefer current (4-20 mA) or digital (IO-Link) signals over voltage for long cables	Far more immune to noise and cable resistance

Commissioning and setting up

• Check before power: wiring against the drawing, polarity, the supply voltage at the sensor, the connector seating, and the sensor’s type (NPN or PNP, NO or NC) against the input.
• Power up and look at the indicator: most sensors have an LED for power and for the output. Check that the output follows the target.
• Set the sensitivity (teach). For photoelectric and capacitive sensors set the threshold midway between the signal with and without the target (a teach button does this), using the worst-case targets (the darkest, smallest, with the dirtiest background).
• Verify the margin: check detection at the edges of the allowed position and the worst case; check that it releases properly.
• Check the PLC side: the input lights up in the program, the address and scaling are right, and the filter time suits the speed (see Sensor Outputs and Wiring).
• Test the failure mode: disconnect the sensor and a wire and confirm that the machine goes to a safe state and shows an alarm.
• Record the settings and the sensor’s position (photograph and note), and save an IO-Link or fieldbus device’s parameters.

Maintenance

• Clean optical lenses and reflectors on a schedule (the dirtiest places weekly): most photoelectric faults are dirt. Use a soft cloth and a mild cleaner, not abrasives that scratch the lens.
• Check the mounting and the connectors at regular intervals: tight, dry and undamaged. Replace cracked cables and worn glands.
• Test the sensor during planned stops: put a target in front and see the output; for safety devices follow the required test schedule.
• Track the failures (which sensor, how often, what environment) and fix the cause: a bracket, a cable, a cooling problem, a wrong sensor type.
• Keep spares of the critical sensors, and the connectors and cordsets that fail most.
• Recalibrate analog sensors at the interval in your quality system, and compare the sensor with a reference (see Sensor Basics and Selection).

A fault-finding flowchart

Symptom table

Symptom	Likely cause	Fix
Works for a while, then fails	Temperature, a bad connection that heats up, a dirty lens building up	Check the temperature at the sensor; clean; re-terminate; allow warm-up
Works at the bench, not on the machine	Vibration, noise, ambient light, a different supply	Bracket, shielding, a hood, a clean supply
Random false triggers	Noise from a nearby coil or drive, a ground loop, sunlight	Suppress the coil, separate the cables, ground the shield at one end, shield the sensor
Misses fast targets	The sensor is too slow, or the input filter or scan time is too long	A faster sensor, a high-speed input, a shorter filter time
Detection point drifts	Temperature, a loose bracket, dirt build-up	Rigid bracket, cleaning, a margin in the setting
Dead after a washdown	Water in the connector or housing, the wrong IP rating	IP69K sensors, sealed connectors, a drip loop
Sensor damaged	Hit by a part, a short circuit, over-voltage, chemical attack	A guard, a fuse, protection, a different material
A new sensor behaves differently	A different type, NO and NC swapped, a different range	Compare the data sheets; set the same parameters

Working safely

• Lock out and de-energize before replacing a sensor on moving equipment; a sensor failing in the wrong state can start a machine.
• Do not bypass a safety sensor to “test” anything: a jumper on an interlock switch has killed people. Safety devices are covered in Safety Sensors and Devices.
• Be careful testing live: a meter on a 24 V DC supply is usually safe, but a cabinet also carries mains. Use the right category of meter and probes.
• Check laser and UV sensors and their class before looking into the beam.
• Remove pressure, temperature and flow before taking out a process sensor.

Related

• Sensors and Field Devices: the overview of this guide.
• Sensor Outputs and Wiring: types of output and the PLC connection.
• Sensor Basics and Selection: error budgets and calibration.
• PLC Troubleshooting, Safety and Documentation and Troubleshooting Industrial Networks.
• Test Tools and Fault Finding: the multimeter and oscilloscope for the electrical side.

FAQ

What is the quickest way to find out whether the sensor or the PLC is at fault?

Measure the output of the sensor at the sensor with a meter while the target moves. If the voltage changes correctly the sensor is fine; follow the cable to the input terminal, then look at the PLC program.

How do I choose between grounding the shield at one end or both?

For analog and low-frequency signals ground at one end (the controller). For high-frequency noise and in installations with an equipotential bonding system, grounding at both ends is better. When in doubt follow the manufacturer’s instructions and the plant’s grounding plan.

Should I use an air purge?

In dusty, steamy or misty conditions a gentle stream of clean air across a lens or a ultrasonic face keeps it clean and cuts maintenance greatly. Use clean, dry, oil-free air.

How often should I replace sensors?

There is no fixed interval for solid-state switching sensors: replace on failure, and review the failure pattern. Replace analog sensors by drift or the calibration schedule, and safety devices strictly by the maker’s mission time.

References

• IEC 60529, Degrees of protection provided by enclosures (IP code); IEC 60947-5-2 (proximity switches).
• IEC 61000-5-2, Earthing and cabling (EMC installation guidelines); Siemens and Rockwell, Wiring and grounding guidelines.
• Manufacturer installation guides (Pepperl+Fuchs, Sick, Balluff, Banner, Omron) for mounting, spacing and cleaning.

Last reviewed: October 2026. Distances and torques are typical; follow the manufacturer’s installation instructions.