Category: Wiring & Installation

Most proximity sensor “failures” are not failed sensors. In the field the usual causes are, in order: the gap has drifted beyond the assured sensing distance, the target material is not what the sensor was sized for, the wiring polarity does not suit the PLC input, or electrical noise from a nearby drive is corrupting the signal. Work through them in that order before replacing anything.

The twelve faults, and how to identify each

1. Sensor never operates — no LED, no output

Measure supply voltage at the sensor connector, with the machine running. A long cable run, a corroded terminal or an overloaded 24 V supply can drop several volts under load. Below the sensor’s minimum operating voltage the oscillator will not start. If the supply is good, check polarity: brown to +, blue to 0 V. Most sensors survive reversal, but they do not operate.

2. LED lights but the PLC never sees it

The sensor is working; the interface is not. Check whether you have a PNP sensor on a sourcing input card, or an NPN sensor on a sinking one — the two must be opposites. This is the single most common commissioning fault, and our guide to PNP versus NPN wiring explains how to confirm which you have.

3. PLC input reads permanently ON

With a two-wire sensor, this is almost always residual leakage current holding a high-impedance input above its OFF threshold. Fit a bleeder resistor across the input or move to a three-wire sensor. With a three-wire sensor, suspect a short between the output core and the supply, or an unshielded non-flush sensor detecting its own mounting bracket.

4. Intermittent — works cold, fails hot

This is the classic symptom of designing to the catalogue sensing distance instead of the assured distance. Rated distance S_n is a nominal figure; guaranteed detection only exists inside 0.81 × S_n, and that must be further multiplied by the reduction factor for the target material. Measure the actual gap and recalculate. Nine times in ten the gap is too large.

5. Detection distance is far shorter than the catalogue says

Check the target material first. Aluminium at a reduction factor of about 0.4 gives you well under half the rated range; copper less still. Then check target size — a target smaller than the reference square loses range roughly in proportion to area. Then check thickness and any plating.

6. Unshielded sensor latches on permanently

It is detecting its own mounting. A non-flush sensor needs a metal-free zone: a clear diameter of at least three times the barrel diameter, with the head standing proud of surrounding metal by at least twice S_n. Either move it out or fit a shielded sensor — see shielded versus unshielded mounting.

7. Two adjacent sensors chatter randomly

Mutual interference. Two inductive sensors mounted too close beat against each other and both outputs become unstable, typically worsening as they warm. Minimum spacing is twice the barrel diameter for shielded units and three times for unshielded. If the layout cannot be changed, specify units with different oscillator frequencies.

8. Random false triggers when a drive or contactor operates

Electrical noise. Confirm by operating the drive with the machine mechanically stationary — if the sensor still trips, it is not seeing a target. Route sensor cables away from VFD motor cables, never in the same trunking or parallel for long runs. Use screened cable with the screen earthed at the panel end only. Add a suppressor across any inductive load switched near the sensor.

9. Output is on but the load will not operate

Check load current against the sensor’s rated output current — typically 100–200 mA for a DC three-wire unit. Contactor coils and lamps often exceed it. Also check voltage drop, particularly with two-wire sensors in series, where the drops add and can starve the load. Our note on two, three and four-wire wiring covers the arithmetic.

10. Works, then dies after a few weeks

Look for mechanical and environmental causes. Vibration loosens locknuts and fatigues cable at the strain relief — flex the cable at the entry and see if the LED flickers. Coolant or washdown ingress past a degraded seal is next. Check the sensor’s IP rating against the actual environment, remembering that the connector is part of the seal.

11. Missing targets at speed

Every sensor has a maximum switching frequency, and the target must remain in the field long enough to be recognised. Calculate the dwell time: target length divided by line speed. If it approaches the sensor’s response time, you need a faster sensor, a longer target flag, or a different sensing position. See our note on proximity switches in high-speed counting.

12. Capacitive sensor stays on after the product has gone

Build-up on the sensing face or product clinging to the inside of a tank wall. The sensor is behaving correctly — high-dielectric material is still in its field. Reduce sensitivity so bulk product is needed rather than a film, and relocate to a point with flow across it. Details in our guide to capacitive sensors for level detection.

A bench test that settles most arguments

When a sensor is suspected, remove it from the machine and test it on a bench with nothing else connected:

1. Apply rated supply — brown to +24 V, blue to 0 V.
2. Leave the output (black) unconnected.
3. Meter between black and blue for a PNP unit, or between brown and black for an NPN unit.
4. Present a mild steel target, 1 mm thick, at least as wide as the sensing face.
5. Move it in slowly and note the distance at which the output switches, then move it out and note where it releases. The difference is the hysteresis, normally 5–20% of the sensing distance.

If the switching distance is close to the catalogue figure on mild steel, the sensor is good and the problem is in the installation, the target, or the interface. That single test eliminates most of the guesswork — and it is worth doing before ordering a replacement.

Quick symptom index

Symptom	Check first	Then
No LED, no output	Supply voltage at the sensor	Polarity, cable continuity
LED on, PLC sees nothing	PNP/NPN versus input card common	Output core continuity
PLC input always on	Two-wire leakage current	Sensor detecting its mounting
Intermittent, worse when hot	Working gap versus 0.81 × Sn	Supply voltage under load
Range much shorter than rated	Target material and size	Thickness, plating, coating
Random trips with drive running	Cable routing and screening	Earthing, load suppression
Misses fast targets	Dwell time versus response time	Target flag length
Fails after weeks in service	Cable at the strain relief	Seal, IP rating, vibration

Frequently asked questions

How do I test a proximity sensor with a multimeter?

Power it at rated voltage with the output disconnected. For a PNP unit, measure between the black output and blue 0 V — it should read close to supply voltage when a target is present. For an NPN unit, measure between brown and black — the voltage should collapse when it operates. If the sensor switches correctly on a mild steel target at roughly the catalogue distance, the fault is elsewhere.

Why does my sensor work at commissioning but fail in summer?

Because the gap was set to the rated distance rather than the assured distance. Sensing range falls at elevated temperature and low supply voltage — that variation is exactly what the assured operating distance of 0.81 × S_n exists to cover. Reduce the gap or fit a larger sensor.

Can a proximity sensor be damaged by wiring it wrongly?

Most quality sensors have reverse-polarity and short-circuit protection and survive common mistakes, simply not operating until corrected. What does destroy them is connecting the output directly across the supply without a load, applying AC to a DC unit, or exceeding the rated supply voltage. Check the label before applying power to an unfamiliar sensor.

What causes a proximity sensor to trigger with nothing in front of it?

Three main causes: an unshielded sensor detecting its own mounting bracket, electrical noise coupled from a nearby drive or contactor, or — on capacitive units — build-up on the sensing face. Test by removing the sensor from the bracket and observing it in free air; if it releases, the mounting is the problem.

How long should an industrial proximity sensor last?

There is no wearing contact, so electronic life is long — many run for a decade or more. Practical life is set by the environment: cable flexing at the entry, seal degradation under washdown, abrasion in dusty plant, and thermal cycling. Sensors that fail early almost always fail mechanically or through ingress, not electronically.

A two-wire sensor sits in series with the load and must draw a small current even when off — that residual current is what causes false-on PLC inputs. A three-wire sensor has its own supply and a clean switching output. A four-wire sensor adds a second, complementary output. For new DC designs, three-wire is almost always the right answer.

Two-wire: elegant wiring, awkward electronics

A two-wire sensor is wired exactly like a mechanical limit switch: one cable, in series with the load. That is genuinely convenient, particularly when retrofitting into existing conduit where pulling a third core is expensive.

The difficulty is that the sensor’s own electronics have to be powered from the same two wires. When the output is off, no current should flow — but the oscillator still needs to run, otherwise the sensor cannot detect the next target. So a two-wire sensor always draws a small residual current, typically between 0.5 and 1.7 mA, through the load.

When the output is on, the sensor is not a perfect switch either. It drops a residual voltage across itself, commonly 3 V and up to 8 V on some units, which is subtracted from what reaches the load.

Both of these are unavoidable consequences of the topology, not manufacturing shortcomings. They only become problems when the load is a modern, high-impedance PLC input rather than a contactor coil.

The false-ON problem, and how to fix it

A 24 V DC digital input on a typical PLC has an input impedance in the region of 3–5 kΩ and guarantees an OFF state below roughly 1.5 mA. Feed it from a two-wire sensor leaking 1.5 mA and the input sits right on the boundary. It may read correctly on the bench at 20 °C and read permanently ON in a hot panel, because leakage rises with temperature.

The standard remedy is a bleeder resistor wired in parallel with the PLC input. It provides an alternative path for the leakage current so that the voltage across the input falls below its OFF threshold.

Choose R so that: I_leakage × (R ∥ Z_input) < V_{off-threshold}

In practice, for a 24 V DC input, a resistor in the range of a few kΩ to around 30 kΩ solves it. Start around 5 kΩ and verify with a meter across the input in the off state. Check the power rating — at 24 V a 5 kΩ resistor dissipates about 115 mW continuously when the sensor is on, so a 0.5 W part is sensible.

The resistor is a workaround, not a design. It burns power continuously, it is one more thing to fall off a terminal rail, and the next engineer will not know why it is there. If you are designing rather than repairing, use a three-wire sensor.

Three-wire: the sensible default

A three-wire sensor takes brown (+), blue (0 V) and black (output). Its electronics run from the supply, not through the load, so:

• Leakage current in the off state is negligible — microamps, not milliamps.
• Voltage drop in the on state is small, typically well under 2 V.
• The output can be optimised purely for switching, which allows better short-circuit and reverse-polarity protection.
• Switching frequency is generally higher, because the oscillator is not starved for supply.

The output polarity — PNP sourcing or NPN sinking — becomes a separate choice, covered in our guide to PNP versus NPN outputs.

Four-wire: when you need both states

A four-wire sensor adds a white core. It is normally one of two things:

1. Complementary outputs — one NO and one NC of the same polarity. Useful for safety and diagnostic logic, because the control system can check that exactly one of the two is active. If both are on, or both off, something has failed.
2. Antivalent or dual-polarity outputs on some ranges, offering a PNP and an NPN output from the same head. Less common, and usually more expensive than simply ordering the right sensor.

Complementary outputs are worth the extra core wherever a false reading has real consequences. They are a poor man’s line-monitoring — genuine fault detection at the sensor level requires a NAMUR sensor and isolating amplifier.

Side-by-side comparison

	2-wire	3-wire	4-wire
Cores required	2	3	4
Residual current when off	0.5 – 1.7 mA	Negligible	Negligible
Voltage drop when on	3 – 8 V	< 2 V	< 2 V
Works on AC and DC	Yes — AC/DC universal types available	DC only	DC only
Suits high-impedance PLC inputs	Often needs a bleeder resistor	Yes	Yes
Polarity choice	Not applicable	PNP or NPN	PNP or NPN, both states
Best for	Retrofits, AC loads, limit-switch replacement	Almost all new DC machine building	Safety and diagnostic logic

Series and parallel connection

Connecting multiple sensors together compounds every effect above, and it is where two-wire installations most often come unstuck.

In series (logical AND), the voltage drops add. Four two-wire sensors each dropping 5 V consume 20 V of a 24 V supply, leaving the load 4 V — it will not operate. Three-wire sensors in series drop far less, but you still need to check the total against the load’s minimum operating voltage.

In parallel (logical OR), the leakage currents add. Four two-wire sensors each leaking 1.5 mA present 6 mA to the input, which will read permanently on regardless of any bleeder resistor you fit.

There are also limits on how many three-wire sensors can be paralleled before the combined off-state leakage matters, and rules about protecting each output. Our note on series and parallel connection of proximity switches covers the arithmetic in detail.

Choosing for a real installation

1. Is there a 24 V DC rail available? If yes, use three-wire. This decides most cases immediately.
2. Are you replacing a mechanical limit switch in existing two-core conduit? A two-wire AC/DC sensor is the pragmatic choice — just check the load’s minimum operating current and voltage.
3. Is the load a contactor coil or a PLC input? Coils tolerate leakage happily. PLC inputs do not.
4. Do you need to distinguish “no target” from “sensor failed”? Four-wire complementary outputs give you a partial answer, NAMUR gives you a full one.
5. Will sensors be combined? Do the voltage-drop and leakage arithmetic before ordering, not during commissioning.

Frequently asked questions

Why does my PLC input stay on with a two-wire sensor connected?

The sensor’s residual current — typically 0.5 to 1.7 mA — is enough to hold a high-impedance digital input above its OFF threshold. Fit a bleeder resistor of a few kΩ across the input to divert that current, or replace the sensor with a three-wire type. The problem often appears only once the panel warms up, because leakage increases with temperature.

What value bleeder resistor should I use?

Size it so that leakage current times the parallel combination of the resistor and the input impedance stays below the input’s OFF-state voltage. For a 24 V DC input, values from a few kΩ up to about 30 kΩ are typical; around 5 kΩ is a reasonable starting point. Verify with a meter across the input, and check the resistor’s power rating for continuous operation.

Can I use a two-wire sensor on 230 V AC?

Yes — two-wire AC and AC/DC universal sensors exist precisely for this, and they are the standard replacement for mechanical limit switches on AC circuits. Confirm the minimum load current the sensor needs to operate correctly; a very light load such as a small indicator lamp may not draw enough.

Does a three-wire sensor have any leakage current at all?

A small amount, but it is measured in microamps rather than milliamps, because the output transistor’s leakage is not carrying the electronics’ supply current. In practice it is far below any PLC input threshold and can be ignored for normal design purposes.

Is a four-wire sensor worth the extra core?

Where a wrong reading has consequences — interlocks, position confirmation on a press, end-of-travel on something heavy — yes. Being able to check that exactly one of the complementary outputs is active detects a large class of sensor and wiring faults. For ordinary presence detection it is an unnecessary expense.

A PNP sensor sources current: when it detects a target it connects its output to +24 V. An NPN sensor sinks current: it connects its output to 0 V. The sensing element is identical in both. Only the output transistor differs, and that difference decides which PLC input card you can wire it to.

What actually differs inside the sensor

Every DC three-wire proximity switch has the same three stages: an oscillator and demodulator that detect the target, a trigger stage that makes the on/off decision, and an output stage that drives the load. Only the third stage differs between PNP and NPN.

In a PNP output, the switching transistor sits between the positive rail and the output terminal. When the sensor operates, that transistor turns on and pushes +24 V out of the black wire. Current flows out of the sensor, through your load, and back to 0 V. This is why PNP is called a sourcing output.

In an NPN output, the switching transistor sits between the output terminal and the negative rail. When the sensor operates, the black wire is pulled down to 0 V. Your load must already be connected to +24 V, and current flows into the sensor. This is a sinking output.

The confusion that costs commissioning engineers hours is that the naming refers to the sensor, while PLC input cards are named for what they do. A sourcing sensor must feed a sinking input. They are always opposites.

PNP vs NPN at a glance

	PNP (sourcing)	NPN (sinking)
Output when target detected	Switches to +24 V	Switches to 0 V
Load connected between	Output and 0 V	Output and +24 V
Current direction	Out of the sensor	Into the sensor
Required PLC input card	Sinking (common tied to 0 V)	Sourcing (common tied to +24 V)
Common in	Europe, India, most modern OEM machinery	Japan, Korea, legacy Far-East equipment
Short-to-earth behaviour	A chafed output cable can energise the load	A chafed output cable simply pulls the line low
Preferred for new builds	Yes — safer failure mode with earthed 0 V systems	Only when matching existing equipment

Why PNP is the safer default

In the overwhelming majority of industrial panels the 0 V rail is bonded to earth. Consider what happens when the signal cable is crushed in a cable tray and the output conductor touches the machine frame.

With an NPN sensor the load is already sitting at +24 V and is waiting to be pulled down. An earth fault on the signal wire pulls it down exactly as the sensor would — the load energises. A conveyor can index, a cylinder can extend, and nothing in the control system knows a fault occurred.

With a PNP sensor the same earth fault simply shorts the output to 0 V. The load cannot energise. The input reads permanently off, which is noticed quickly and is the safer of the two failure states.

This is the practical reason IEC-aligned machine builders standardised on PNP, and it is why every Accent inductive proximity switch is offered in PNP as the default DC output, with NPN available where an existing installation demands it.

Reading the wiring before you connect

Three-wire DC sensors follow a colour convention that is consistent across reputable manufacturers:

• Brown — positive supply, +10 to +30 V DC on most units
• Blue — negative supply, 0 V
• Black — the switching output
• White — present on four-wire units: either a second, complementary output or the normally-closed contact

On an M12 connector the pin allocation is pin 1 brown, pin 2 white, pin 3 blue, pin 4 black. If you are staring at an unmarked sensor, a multimeter between brown and black with the supply connected will read close to supply voltage on an operated PNP unit and close to zero on an operated NPN unit.

Can you convert one to the other?

Not reliably, and not by rewiring. What you can do is adapt the input side:

1. Change the PLC input card common. Most modern digital input modules let you choose sourcing or sinking by which rail you tie the group common to. This is the correct fix and costs nothing.
2. Use an interposing relay. A 24 V DC relay coil driven by the sensor gives you a dry contact that suits any input. It adds a few milliseconds of delay and a wear item, so avoid it on high-speed counting duties — see our note on proximity switches in high-speed RPM counting.
3. Fit a switching amplifier or control unit. Where the sensor is remote, in a hazardous area, or must drive an AC load, a standard control unit decouples the sensor from the load entirely.

What you must not do is add a pull-up or pull-down resistor and hope. That works on a bench and fails at temperature, because the output transistor’s leakage changes with junction temperature while your resistor does not.

NO, NC and why it is a separate question

PNP and NPN describe how the output switches. Normally open (NO) and normally closed (NC) describe when. A PNP NO sensor puts +24 V on the output when a target is present. A PNP NC sensor puts +24 V on the output when no target is present and removes it on detection.

For anything with a safety or presence-monitoring function, prefer NC. A broken cable or a failed sensor then produces the same signal as a missing target, so the fault is detected rather than masked. For genuinely fail-safe detection in hazardous areas the correct answer is a NAMUR proximity switch, which reports cable break and short circuit as distinct current levels — explained in our guide to how NAMUR sensors work.

Choosing for a real machine

Work through these in order:

1. Look at the PLC first. Read the input card wiring diagram and see which rail the group common connects to. That single fact decides PNP or NPN, and everything else follows.
2. Match the existing plant. If a line already runs on NPN, adding one PNP sensor creates a spares problem out of all proportion to the saving. Consistency beats theoretical preference.
3. Confirm the load current. A typical DC three-wire output handles 100–200 mA. Contactor coils and lamp loads can exceed that; check the datasheet rather than assuming.
4. Decide NO or NC on safety grounds, not on which one makes the ladder logic shorter.
5. Check the supply. If the panel runs on 110 or 230 V AC and there is no 24 V rail, a two-wire AC sensor may be simpler than adding a power supply — but read our comparison of two, three and four-wire sensors before committing, because two-wire units bring leakage current problems of their own.

Frequently asked questions

Will an NPN sensor be damaged if I wire it to a sinking PLC input?

Usually not. The input simply never turns on, because both the sensor and the input are trying to pull the same line to 0 V and nothing sources current into it. The sensor is undamaged and works correctly once the input common is moved to +24 V. Damage only occurs if you connect the output directly across the supply.

How do I identify PNP or NPN on an unmarked sensor?

Power the sensor from 24 V DC with nothing connected to the black wire. Measure between black and blue (0 V) with a multimeter, then present a target. A PNP unit will swing to near +24 V. An NPN unit will stay near 0 V; measure between brown and black instead and you will see the voltage collapse when it operates.

Is PNP faster than NPN?

No. Switching speed is set by the oscillator and trigger stages, not the output transistor. Both types from the same product family carry the same rated switching frequency. Speed differences you see in practice come from the sensor size and sensing principle, not the output polarity.

Which should I specify for a new machine in India?

PNP, unless you are matching existing equipment. Indian machine building follows IEC practice, panel 0 V is normally earthed, and PNP gives the safer earth-fault behaviour. Availability of spares is also better.

Can one sensor have both PNP and NPN outputs?

Some four-wire units provide complementary NO and NC outputs of the same polarity, which is not the same thing. Genuine dual-polarity outputs exist but are uncommon and cost more than simply specifying the right sensor. For most applications, choosing correctly at order stage is cheaper.