Proximity Sensors

What is a Proximity Sensor

A proximity sensor is a self-contained device that senses the approach of a target object within a defined distance and changes the state of a semiconductor output, all without mechanical contact. Where an old-style limit switch needs a lever, a roller, and a spring that physically touch the moving part, a proximity sensor projects a field, light beam, or sound pulse and watches for a disturbance. Because nothing touches, there is nothing to wear: a quality inductive unit is rated for hundreds of millions of switching cycles, and its lifetime is limited by the electronics and the cable, not by contact erosion.

The industrial definition is anchored by IEC 60947-5-2, "Low-voltage switchgear and controlgear, Part 5-2: Control circuit devices and switching elements, Proximity switches." The 2019 fourth edition covers inductive and capacitive switches that sense metallic or non-metallic objects, ultrasonic switches that sense sound-reflecting objects, photoelectric switches that sense objects optically, and non-mechanical magnetic switches that sense a magnetic field. These devices are self-contained, use semiconductor switching elements, and are intended for circuits whose rated voltage does not exceed 250 V AC at 50/60 Hz or 300 V DC. In North America the same document is adopted as UL 60947-5-2, which keeps the test methods and sensing-distance definitions consistent across markets.

Structurally, every proximity switch has three functional blocks: (1) the active element that interacts with the target, a coil for inductive, a capacitor plate for capacitive, a transducer for ultrasonic, or a Hall element for magnetic; (2) an evaluation circuit, typically an oscillator with a demodulator and a trigger stage that decides when the field has been disturbed enough to switch; and (3) an output stage, usually a transistor that sources or sinks load current, often with short-circuit and reverse-polarity protection. The whole assembly is potted in resin inside a threaded barrel or rectangular block, which is what gives these sensors their characteristic immunity to vibration, dust, oil, and coolant.

The lineage runs from the mechanical limit switch and the magnetic reed switch of the early twentieth century to the solid-state inductive sensor, which appeared in the 1960s as semiconductor oscillators became cheap and reliable. Inductive sensing remains by far the highest-volume category in industrial automation because most machine targets, cams, flags, gear teeth, and pistons, are metal. Capacitive, ultrasonic, photoelectric, and magnetic types fill the gaps where the target is non-metallic, far away, transparent, or hidden behind a wall.

The reason a proximity sensor is treated as a switching device, not a measuring instrument, matters for selection. A standard inductive or capacitive proximity switch gives a binary present or absent output around a single setpoint with built-in hysteresis. When an application needs a continuous distance reading, the engineer moves to an analog proximity device defined by IEC 60947-5-7 (PDAO, proximity device with analog output), to an ultrasonic or laser distance sensor, or to a measuring inductive sensor, all of which output 0 to 10 V, 4 to 20 mA, or an IO-Link process value rather than a clean on/off edge.

The Four Sensing Families

Selection starts with the target, because each family senses a different physical property. Inductive sensors see only conductive metal; capacitive sensors see almost anything that changes the local dielectric; ultrasonic sensors see any surface that reflects sound, including transparent and dark objects; magnetic sensors see a magnet or a magnetized actuator, even through non-magnetic walls. Photoelectric (optical) sensing is a fifth family covered by the same IEC 60947-5-2 standard but is documented separately on SpecForge because its selection logic centers on light modes rather than near-field physics. The table below compares the four near-field families on the parameters that drive a first cut.

Inductive is the default for machine position sensing. It detects cam edges, gear teeth, piston flags, indexing slots, and metal parts on a conveyor. Because the field is generated and read entirely by the sensor, dust, oil mist, coolant, and non-metallic debris in the gap are ignored, which is why inductive sensors dominate harsh metal-cutting and stamping environments. The trade-off is that the target must be metal and the achievable range is short, on the order of a few millimeters for small barrels.

Capacitive extends sensing to non-metals: it triggers on plastic, glass, wood, paper, cardboard, water, oil, grain, powder, and pellets. The most common industrial use is non-contact level detection, where the sensor mounts outside a plastic or glass tank wall and detects the product rising behind it. The same broad sensitivity is also a liability: condensation, foam, dust films, and changes in humidity can shift the trigger point, so capacitive sensors carry a sensitivity adjustment and are usually set up on the actual application rather than trusted to a single catalog number.

Ultrasonic emits a burst of high-frequency sound and times the echo, so it reads distance to almost any surface regardless of color, transparency, or shine, including clear film, black rubber, and liquid surfaces that defeat optical sensors. Ranges reach several meters, far beyond inductive or capacitive. Every ultrasonic sensor has a blind zone, a minimum distance below which the transducer is still ringing and cannot hear an echo, and performance degrades on sound-absorbing foam, steeply angled surfaces, and in strong air turbulence or large temperature gradients.

Magnetic proximity sensors, using a reed switch or a solid-state Hall element, respond to a permanent magnet rather than to the bare metal of a part. Their signature application is detecting the piston inside a pneumatic or hydraulic cylinder: the piston carries a ring magnet, the sensor clips into a slot on the aluminum cylinder body, and it senses the magnet straight through the non-magnetic aluminum wall, which no inductive sensor could do. Hall-effect versions have no moving parts and longer life than reed types and can report field direction as well as presence.

How Each Principle Works

Understanding the physics prevents the most expensive selection mistakes, because each principle has a hard boundary that no datasheet wording can overcome. The table below summarizes the active mechanism, the governing variable, and the boundary condition for each of the four families, after which each is described in detail.

Inductive (high-frequency oscillation). A ferrite-core coil at the active face is driven as the inductive element of an LC oscillator, radiating a high-frequency alternating field. When a conductive metal target enters that field, the field induces circulating eddy currents in the target surface. Those eddy currents draw energy from the oscillator, reducing its amplitude; a demodulator and trigger stage detect the amplitude drop and switch the output. Ferromagnetic targets such as mild steel also concentrate the field and damp the strongest, which is why steel gives the longest range and is used as the standard reference target. Because the sensor both creates and reads the field, non-metallic contaminants in the gap are invisible to it.

Capacitive. The active face acts as one plate of a capacitor, with the target and surrounding environment completing the dielectric path. As any material with a dielectric constant higher than air approaches, the effective capacitance of the sensing circuit rises. That capacitance change pulls an internal oscillator from quiescent into oscillation (or shifts its amplitude), and the trigger stage switches. The higher the dielectric constant of the target, the longer the range: water and metals trigger easily, while dry plastics and powders trigger at shorter distance and benefit from the sensitivity adjustment. This is also why a capacitive sensor can see product through a thin plastic or glass wall, the wall and the product together change the dielectric.

Ultrasonic. A piezoelectric transducer emits a short burst of ultrasound, typically in the tens to low hundreds of kilohertz, then switches to listening mode and measures the time of flight until the echo returns. Distance is half the round-trip time multiplied by the speed of sound in air, about 343 m/s at 20 degrees Celsius. Because it relies on a returning echo, the transducer cannot listen while it is still ringing from the transmit pulse, which creates the blind zone close to the face. The speed of sound varies with air temperature, so precision models include temperature compensation. Output can be a switching setpoint, an analog 0 to 10 V or 4 to 20 mA proportional to distance, or an IO-Link process value.

Magnetic (Hall effect and reed). A reed switch is a pair of ferromagnetic contacts sealed in a glass capsule that close mechanically when an external magnetic field exceeds a threshold. A Hall-effect sensor is a solid-state transducer whose output voltage varies with the perpendicular magnetic flux density; a comparator then switches the output once flux passes a setpoint. Hall devices have no moving parts, switch faster, last longer, and can resolve field amplitude and polarity, whereas reed switches are simple, need no supply current, and tolerate wide voltage. Both detect a magnet through non-magnetic barriers such as aluminum, stainless steel, plastic, and air, which is the basis of pneumatic-cylinder position sensing.

Sensing Distance and Target Correction

Sensing distance is the parameter most often misread, because the single number printed on the barrel is a defined reference value, not the distance you will get in your machine. IEC 60947-5-2 builds a strict hierarchy of distances so that catalog figures from different vendors can be compared on equal footing. The nominal sensing distance Sn is the headline value, measured against a defined standard target and deliberately excluding manufacturing tolerance, supply-voltage variation, and temperature effects. For inductive sensors the standard target is a square plate of mild steel Fe 360, 1 mm thick, with a side dimension equal to the diameter of the active face or three times Sn, whichever is larger.

From Sn the standard derives the values that describe a real device. The real sensing distance Sr is measured on an individual sensor at rated voltage and rated ambient temperature and must fall between 0.9 Sn and 1.1 Sn. The usable sensing distance Su is measured across the whole permitted supply-voltage and ambient-temperature window and must fall between 0.81 Sn and 1.21 Sn. The assured operating distance Sa, from 0 up to 0.81 Sn, is the conservative working limit within which the manufacturer guarantees detection under all rated conditions; engineers set the actual target gap inside Sa, not at Sn. The table below shows representative figures for common DC three-wire inductive barrels.

The flush versus non-flush distinction is mechanical, not electronic. A flush, or shielded, sensor carries a metal collar around the coil that focuses the field forward, allowing the body to be mounted flush in a steel fixture without the surrounding metal triggering it; the price is a shorter Sn. A non-flush, or unshielded, sensor has no collar, throws its field sideways as well as ahead, and reaches a longer Sn, but it must sit in a metal-free zone, typically a clear annulus around the face and a free distance ahead equal to several times Sn, or the surrounding metal will hold it permanently on. Installing the wrong style is one of the most common field faults found at machine commissioning.

For inductive sensors the second correction is the target material. Sn is defined only for the standard mild-steel plate; any other metal damps the oscillator differently, so the real range equals Sn multiplied by a reduction (correction) factor below 1. The values below are representative of conventional inductive sensors and are the reason an aluminum target can trigger at less than half the catalog distance. A target smaller than the standard plate, or thinner than 1 mm, reduces range further. Factor 1 sensors use a redesigned coil and evaluation circuit to hold roughly the same range on all metals.

Capacitive, ultrasonic, and magnetic families have their own distance physics. Capacitive range depends on the dielectric constant of the target, so a water column triggers much farther than dry plastic granulate, and the catalog figure is usually quoted for an earthed metal plate or for water, with a potentiometer to tune the rest. Ultrasonic distance is set by transducer frequency and power and is bounded below by the blind zone and above by echo strength, with typical industrial windows running from roughly 20 mm out to several meters. Magnetic range depends on the strength of the target magnet and the presence of any steel that diverts the field, so cylinder sensors are matched to the magnet the cylinder maker fitted.

Key Specification Parameters

Beyond sensing distance, a proximity sensor datasheet lists a dozen parameters, but only a handful change a selection decision: repeat accuracy, hysteresis, switching frequency, output type and logic, supply voltage and load current, ingress protection, and operating temperature. Each is explained below in the language of the IEC 60947-5-2 datasheet.

Repeat accuracy (repeatability) is the spread of the operating point across many approaches under constant conditions, expressed as a percentage of Sr, commonly within 1 to 5 percent for inductive sensors. It governs how tightly a part can be positioned by a sensor. Hysteresis (differential travel) is the gap between the switch-on point as the target approaches and the switch-off point as it retreats, typically 1 to 15 percent of Sr. Hysteresis is deliberate: it stops the output chattering when a target sits exactly at the limit or vibrates, but excessive hysteresis blurs the true position.

Switching frequency is the maximum number of detect-release cycles per second, set by the standard target test in IEC 60947-5-2. Small inductive barrels reach several hundred hertz to a few kilohertz; large M30 units are slower, often tens to low hundreds of hertz; capacitive and ultrasonic types are slower still. A fast indexing or counting application must check this number against the line speed. Response and release times are the matching delays in milliseconds, and the larger of the two limits real throughput.

Output type and logic is where wiring compatibility is decided, and it has two independent axes. The transistor polarity is PNP (sourcing, switches the positive supply to the load) or NPN (sinking, switches the load to 0 V); the contact function is NO (conducts when a target is present) or NC (conducts when absent). PNP versus NPN says nothing about NO versus NC, a frequent source of confusion. Wiring count follows: two-wire DC (or AC/DC) units splice into the load line and pass a small residual current when off; three-wire DC units have separate supply and signal and are the modern default; four-wire units provide both NO and NC outputs at once. The table below maps the common configurations.

Supply voltage and load current. The industrial DC standard is a 10 to 30 V DC window centered on 24 V, with continuous load currents commonly 100 to 200 mA and short-circuit, overload, and reverse-polarity protection on quality units. Two-wire units list an off-state residual (leakage) current and a minimum holding current that must be respected so the device powers itself; ignoring these is a classic cause of a two-wire sensor that will not turn fully off. AC and AC/DC versions also exist for direct contactor switching but are less common on new builds.

Ingress protection and temperature. Ratings come from IEC 60529: IP67 (dust tight, temporary immersion to 1 m) covers most factory duty; IP68 adds continuous immersion; IP69K, per ISO 20653, adds close-range 80 degrees Celsius, 80 to 100 bar steam-jet washdown for food and pharma. Standard operating temperature is roughly minus 25 to plus 70 degrees Celsius, with extended-range and high-temperature variants available. Also note the housing material (nickel-plated brass or V4A / 316L stainless), the connection style (M8 or M12 connector versus fixed cable), and any short-circuit and EMC ratings to IEC 61000.

Selection Decision Factors

To turn the preceding chapters into a specific part number, work the decision sequence below in order. Most selection errors come not from one wrong value but from deciding range or output before the target material and mounting are fixed. These steps double as a fixed RFQ template.

1. Target material first: Metal only points to inductive; non-metals, granules, or level behind a wall point to capacitive; long range, transparent, or dark objects point to ultrasonic; a piston or a magnet target points to magnetic. This single choice eliminates most of the catalog.
2. Required sensing distance and approach: Set the working gap inside the assured distance Sa (up to 0.81 Sn), not at Sn, and account for mechanical tolerance. For inductive sensors apply the target-material reduction factor (about 0.85 stainless, 0.40 aluminum, 0.30 copper) before choosing a barrel size, or specify a Factor 1 sensor.
3. Mounting style: Flush (shielded) if the body must sit embedded in metal, accepting the shorter range; non-flush (unshielded) for longer range with a guaranteed metal-free zone around and ahead of the face. Confirm the free-zone dimensions on the datasheet.
4. Form factor and connection: Threaded barrel (M8, M12, M18, M30) or rectangular block; pre-wired cable versus M8 or M12 connector for quick replacement. Match thread and length to the existing fixture or bracket.
5. Output type and logic: Match the transistor to the PLC input common, PNP for sourcing inputs and NPN for sinking inputs, then choose NO or NC, preferring NC where a broken cable should read the same as a removed guard. Consider IO-Link if the controller has masters and you want auto-parameterization or diagnostics.
6. Switching frequency and response time: Verify the maximum switching frequency and response/release times exceed the fastest event the application must catch, with margin for line-speed increases.
7. Environment and protection: Choose the IP rating for the splash, immersion, or washdown reality (IP67, IP68, or IP69K), the operating temperature band, the housing alloy for the chemistry, and EMC and short-circuit ratings. Add ATEX or IECEx versions for hazardous areas.
8. Total cost of ownership: Weigh unit price against replacement labor, downtime cost, and spares availability. A connectorized, IO-Link, stocked sensor costs more upfront but is swapped and re-parameterized in minutes, which on a high-uptime line outweighs the price of a cheaper hard-wired part.

One last dimension that is easy to overlook is manufacturer serviceability: regional stock for fast replacement, IODD files registered for IO-Link devices, documented reduction factors and free-zone drawings, and consistent UL, CE, ATEX, or IECEx certification across the family. Omron, Pepperl+Fuchs, ifm electronic, Balluff, SICK, Baumer, Turck, Carlo Gavazzi, Schneider Electric, Eaton, and Rockwell Automation all publish complete IEC 60947-5-2 data and maintain regional distribution, which is what determines repair response years into a production line's life. These factors seem irrelevant at the quoting stage but dominate the real cost of keeping a machine running.