Limit Switch vs Inductive Sensor: When to Replace

Limit Switch or Inductive Sensor: The Short Answer

Short version. Replace a mechanical limit switch with an inductive sensor when the mechanics are what fails: the axis cycles faster than your switch's datasheet allows (30 operations per minute on the Omron D4N; the XCKN2121P20 rates its durability at under 60 cycles per minute), the lever bends, the roller packs with swarf, or the switch sits somewhere nobody wants to reach with a wrench. Keep the mechanics in three cases: when the contact has direct opening action per Annex K of IEC 60947-5-1, when a contactor coil hangs directly off the switch in utilization category AC-15, and when switches are wired in a series chain. We sell inductive sensors; we do not carry limit switches. So what follows is not an argument for replacement, it is the set of rules we use ourselves to decide when a replacement makes sense and when we turn it down.

Every number here comes from a manufacturer datasheet or from the text of two standards. The mechanical switches XCKN2121P20 and Omron D4N appear as documentation examples, so there is something to compare against; we do not trade in them. Omron is a brand we carry in sensors, but the D4N specifically is a datasheet example rather than an item of ours. The inductive sensors we link to are in the catalogue.

Not Two Different Worlds, the Same Position Switch

The standard treats both devices as one function. IEC 60947-5-2, clause 2.1.1, defines the non-contact device word for word as "a position switch which is operated without mechanical contact with the moving part". A proximity sensor is not an alternative to a limit switch, it is a limit switch without the touch. The mechanical devices live in the neighbouring part of the same standard: IEC 60947-5-1 is titled "electromechanical control circuit devices".

There is exactly one difference, and everything else follows from it. In the mechanical switch a metal contact does the switching. In the inductive sensor, per clause 2.1.1.1 of that same 60947-5-2, a semiconductor switching element does it, with an oscillator producing the field. From that single difference come all the practical consequences:

• metal wears and carries a rated life in cycles; a semiconductor has neither contacts nor a cycle rating;
• a metal contact carries coil current; the semiconductor output of a two-wire sensor carries 500 mA at most;
• an open metal contact passes nothing; a semiconductor output always passes an off-state current (Ir);
• a metal contact can be forced apart mechanically; a semiconductor cannot, and that is physics, not product quality.

First Decision Point: What Actually Hangs on the Output

Start with the load, not the sensor. If a contactor or starter coil sits directly on the limit switch, an inductive sensor does not drop into that point without an interface relay. The reason is the utilization categories, and both standards spell them out.

IEC 60947-5-1, Table 1 (clause 4.4), assigns mechanical devices AC-15 — control of electromagnetic loads above 72 VA, with AC-14 for loads below 72 VA. IEC 60947-5-2, Table 2, gives electromagnetic loads the category AC-140 — control of small electromagnetic loads with holding (closed) current no greater than 0.2 A, e.g. contactor relays (alongside AC-12, DC-12 and DC-13 in the same table). The difference is not marketing, it is the 0.2 A holding-current boundary.

Look at the datasheet figures. The mechanical XCKN2121P20 carries contact code A300 and in AC-15 switches Ie = 3 A at Ue = 240 V, with a conventional thermal current of 10 A. The Omron D4N gives the same 3 A at 240 V in AC-15, and its A300 table states it plainly: at 240 VAC it makes 30 A and breaks 3 A. The two-wire inductive NBB20-U1-UU runs on 24…230 V AC and is the closest thing to a limit switch we carry, but its rated Ie is 500 mA at 50/60 Hz and its operating current sits in a 5…500 mA window.

The working rule that falls out of those numbers: measure the holding current of the coil your switch drives. Under 0.2 A and a two-wire sensor goes in directly. Above it, put an interface relay between sensor and coil, and allow panel space for it while you are still deciding. The standard already assumes the coil will inrush, incidentally: the AC-140 test in Table 4 is run at six times Ie on make with cos φ 0.3.

Second Decision Point: Three Two-Wire Numbers a Dry Contact Never Had

When you swap a dry contact for a two-wire sensor in an existing circuit, these three datasheet lines wreck the job more often than a wrong gap does. Same NBB20-U1-UU:

• Off-state current Ir ≤ 1.5 mA. An open dry contact passes nothing. A two-wire sensor passes up to 1.5 mA at all times. A sensitive relay or an older controller input may not drop out on that current, and you get a channel that appears stuck for no visible reason.
• Lowest operating current Im = 5 mA. The load has to draw at least 5 mA or the sensor never reaches its operating state. A high-impedance controller input will not provide it.
• Voltage drop Ud ≤ 5 V in DC and ≤ 8 V in AC (typically 2 V and 3 V). In a 24 V circuit, eight volts is a third of the supply, and the sensor takes it, not the load.

A fourth number sits in the same table: time delay before availability tv ≤ 300 ms. A dry contact reports its state immediately; for the first three tenths of a second the sensor reports nothing definite. If your start-up interlock logic reads that signal, 300 ms belongs in the program.

And the one we see most often in enquiries. The NBB20-U1-UU datasheet carries a flat prohibition: "The sensor is designed for individual operation. A series or parallel connection of sensors is not permissible." Limit switches are wired exactly that way — two travel ends in series, a string of switches along a track. If your circuit is built like that, two-wire sensors will not substitute into it; you need three-wire outputs and the logic in the controller.

The three-wire DC route avoids part of this: Sick IME12-04BPSZC0K drops Ud ≤ 2 V on a 10…30 V DC supply, with supply and output on separate conductors. But then you pull three wires instead of two, and you have to pick PNP or NPN to match the controller input — we covered that separately in the article on wiring a sensor to a PLC.

When to Replace: Mechanics Wear Faster Than the Catalogue Suggests

The rated durability of a limit switch does not mean what it looks like. IEC 60947-5-1, clause C.2.1, defines mechanical durability as the number of no-load operating cycles that will be attained or exceeded by 90 % of all devices tested without repair or replacement of any part. That is not a guarantee for your switch: it is the lower decile of a batch — at least 90 % of devices attain or exceed the figure and the remaining 10 % may fail earlier — and all of it measured with no load on the contacts. Electrical durability (clause C.3.1) uses the same criterion but on load. And per clause C.1.1 both tests are run at the manufacturer's discretion: if a figure is declared it must correspond to those tests, but declaring one is optional. The preferred series is in the same clause: 0.01, 0.03, 0.1, 0.3, 1, 3, 10, 30 or 100 million cycles.

Here is how that lands on a real datasheet. The Omron D4N declares mechanical durability of 15,000,000 operations minimum, dropping to 10,000,000 for fork lever lock versions. Electrical durability in the same document: 500,000 operations minimum at a 3 A resistive load at 250 VAC, and 300,000 at 10 A at 250 VAC. Put those side by side — under working load the life falls by a factor of thirty against no load. The XCKN2121P20 declares 10,000,000 mechanical cycles and 5,000,000 electrical, but the electrical figure is at DC-13, 120 V and a load of 4 W. Four watts. That does not extrapolate to a coil.

Two small lines in the D4N datasheet decide more than the life figure does:

• all durability figures apply at 5…35 °C and 40…70 % relative humidity (note 4). Not a dusty shop floor, not a freezer, not a washdown area. Note 5 adds that above 35 °C you must not pass 3 A at 250 VAC through more than two circuits.
• a contact that has switched a standard load cannot afterwards be used for a smaller one (note 2): the contact surface roughens and contact reliability is lost. Repurposing an old power limit switch as a signal input to a controller is therefore a bad idea even if it is still mechanically sound.

Then there is rate. The D4N allows a maximum of 30 operations per minute and an actuation speed of 1 mm/s to 0.5 m/s. Do the arithmetic: 15,000,000 operations at 30 per minute is 500,000 minutes, roughly 8,300 hours — under a year of continuous running at the rated rate. On a conveyor indexing once a second, a mechanical limit switch is a consumable.

Inductive sensors play in a different league here. NBB2-8GM30-E2 in M8 is rated 0…6000 Hz, NBB4-12GM50-E2-V1 0…2550 Hz, and the IME12-04BPSZC0K 2000 Hz. Against 30 operations per minute, which is 0.5 Hz, that is a factor of four to twelve thousand. The two-wire NBB20-U1-UU is more modest because of its circuit: 0…100 Hz in DC and only 0…30 Hz in AC — still sixty times the mechanical figure.

An honest caveat on those hertz: 60947-5-2 measures switching frequency in clause 8.5 with the target passing at half the rated operating distance, using f = 1/(t1+t2), and takes the DC rated value as the point where t1 or t2 equals 50 µs. That is a laboratory maximum, not a promise of running at that rate in your machine. Even discounted, though, the headroom stays large.

When Not to Replace: Four Cases Where We Say No

We say no here more often than yes, and these are the grounds.

Direct Opening Action

If the old switch has direct opening action, an inductive sensor does not replace it — not a pricier one, not a dual-output one, none. Annex K of IEC 60947-5-1 defines it this way: the break contacts are coupled to the actuator through non-resilient members, and contact separation is achieved as the direct result of actuator movement, "for example not dependent upon springs" (clauses K.2.1 and K.2.2). The standard adds requirements: the utilization category shall be AC-15 or DC-13 (K.4.4), minimum Ui is 250 V, minimum conventional free air thermal current is 2.5 A, every such contact element is marked with a symbol on the outside, and the manufacturer must publish the minimum direct opening travel and force and, specifically for limit switches, the maximum speed and frequency of actuation (K.5.4.1).

Datasheets show it immediately. XCKN2121P20 states "Positive opening: With", with a positive opening minimum force of 10 N and a minimum tripping force of 6 N. The Omron D4N holds a TÜV certificate to EN 60947-5-1 for approved direct opening, file B03 11 39656 061. In an inductive sensor the switching is done by a semiconductor (clause 2.1.1.1); there is physically no mechanical contact to force apart, so there is nothing to certify. Dropping a sensor into a guard interlock or an emergency travel-limit circuit is not a parts swap, it changes the safety function, and that belongs to whoever signs the risk assessment. We do not sell that replacement.

Series Chains

Two limit switches on two travel ends, wired in series, is standard practice and does not work with two-wire sensors because the datasheet forbids it. The outputs have to go to the controller as separate channels.

Load Beyond the Category

A coil holding more than 0.2 A, a heater, a 230 V indicator lamp with a large inrush — all outside AC-140. An interface relay solves it technically, but then the panel gains one more device that also has contacts and also wears. Sometimes leaving the mechanical switch in place is cheaper.

Non-Metallic Target or Too Large a Gap

An inductive sensor sees metal, and only metal. A wooden pallet, a plastic carriage, glass — not its job; that calls for optical sensing. And the gap: a mechanical switch trips on physical contact, which is to say from zero, while an inductive one has an assured operating distance Sa that clause 7.2.1.3.3 caps at 81 % of Sn, before you multiply by the correction factor for your metal. The full calculation procedure and the catalogue tables by size live in a separate article on finding an inductive sensor equivalent; the one thing to carry away here is that you size the gap from Sa, never from the catalogue Sn.

Comparison Table

The Replacement Itself, Step by Step

This is the order we work through with a customer. The first two steps produce ninety percent of the result.

1. Measure what the old switch actually switches. Voltage, type of current, load holding current. That decides the sensor class: two-wire AC/DC, three-wire DC, or "sensor plus interface relay".
2. Look at the target, not at the switch. The roller of the old limit switch presses on a cam or a stop, and that stop is almost never a standard target. Clause 8.3.2.1.1 describes the target as a square of carbon steel Fe 360 to ISO 630, 1 mm thick, rolled finish, with the side equal to the diameter of the circle inscribed on the active surface of the sensing face, or to three times Sn, whichever is greater. A smaller or thinner target cuts the range. The right answer is often to bolt or weld a steel target plate of the correct size to the carriage.
3. Size the gap from Sa for your metal, not from the catalogue Sn, and allow for bracket play. The procedure, formula and correction factors are in the article on finding an equivalent.
4. Isolate the circuit and prove there is no voltage. On cranes, gates and conveyors the limit switch often sits in a 230/240 V circuit. Before you undo a terminal, open the supply disconnect, lock it off against accidental closing and verify with a tester that the switch terminals are dead.
5. Bracket and wiring. The NBB20-U1-UU has a terminal compartment for conductors up to 2.5 mm² at a tightening torque of 1.2 Nm, so the existing cable from the switch usually carries over without rerouting. Housing tightening torque is 1.8 Nm.
6. Do not skip short-circuit protection. XCKN2121P20 calls for a 10 A gG cartridge fuse; the D4N calls for a 10 A gI or gG fuse to IEC269 and warns outright that it is not built in. The NBB20-U1-UU has pulsing short-circuit protection, and on top of that the UL section of its datasheet asks for an external 3 A fuse to UL 248. 60947-5-2 tests sensors at a 100 A prospective short-circuit current (clause 8.3.4.1), so circuit protection is not a formality.
7. Check before handover. Supply, output state in both positions and, for a two-wire device, whether off-state current is holding the relay in. The step-by-step meter procedure is in how to test a proximity sensor with a multimeter.

Our recommendation, not a requirement of any standard: if the limit switch was acting as a travel limit on a crane, a gate or a hoist, do not strip the mechanics out entirely. Leave it as the emergency end stop and add the sensor as the working one. Two independent principles on one axis fail in different ways, and that is the only configuration we recommend without reservations.

One case of its own, which we see on cranes, gates and conveyors: there the limit switch often goes not to a coil but to a discrete input of a variable frequency drive — as a forward or reverse travel limit, or as an external fault that drops the drive. AC-15 is then beside the point: a VFD input is a low-voltage control circuit, and a two-wire AC sensor does not go into one. Check the drive manual for the input type it expects, take a three-wire sensor to match that logic, and reassign the input's function by parameter in the drive menu rather than by rewiring.

Pneumatic Cylinders and T-Slots: A Different Choice

If the limit switch sits on a pneumatic cylinder, the comparison is not with an inductive sensor but with a magnetic slot sensor. And there is a trap: a reed is also a contact device. The datasheet for Sick RZT7-03ZRS-KP0 states a Reed output, 400 Hz, typical sensitivity 3 mT, 5…30 V AC/DC supply, current to 500 mA — and B10D of 10,000,000 switching cycles, with a footnote in that same datasheet: "only for devices containing electro-mechanical components". The reed has a wear figure in cycles because it has something to wear. The semiconductor MZT7-03VPS-KP0 and MZT8-03VPS-KP0 with PNP outputs carry no such figure, for the same reason inductive sensors do not. Our choice between them is simple: reed when you need to switch an AC circuit on two wires, PNP when the signal goes to a controller. Everything we carry in this class is grouped under cylinder sensors.

Frequently Asked Questions

Can an inductive sensor replace a limit switch in a safety circuit?

No, if the old switch had direct opening action. Annex K of IEC 60947-5-1 requires contact separation to be the direct result of actuator movement through non-resilient members and not dependent upon springs, whereas an inductive sensor switches through a semiconductor and has no mechanical contact at all. That is a change to the safety function rather than a parts swap, and the decision sits with whoever owns the risk assessment for the machine. A neighbouring topic is safety light curtains, OSSD and safe distance.

Will an inductive sensor drive a contactor coil?

It depends on the coil's holding current. IEC 60947-5-2 places non-contact devices in category AC-140, bounded at 0.2 A holding current, and the two-wire NBB20-U1-UU is rated Ie 500 mA. A mechanical switch in AC-15 switches 3 A at 240 V, six times more. If the coil holds more than 0.2 A, you need an interface relay between the sensor and the coil.

Why did the relay stop dropping out after I fitted a sensor?

Almost always the two-wire sensor's off-state current (Ir). The NBB20-U1-UU is rated at up to 1.5 mA off-state current, and a sensitive relay can hold in on that. A dry contact never had the problem. The fix is either a three-wire output or a shunt resistor sized for that specific relay.

How long does a limit switch really last compared with a sensor?

The rated mechanical durability figure is the lower decile of a batch: at least 90 % of devices attain or exceed it and the remaining 10 % may fail earlier, and it is measured with no load (IEC 60947-5-1, clause C.2.1). For the Omron D4N that is 15 million operations, but under a working load of 3 A at 250 VAC the same datasheet gives 500 thousand — thirty times less — and those figures hold at 5…35 °C and 40…70 % humidity. At the rated 30 operations per minute, 15 million cycles are consumed in roughly 8,300 hours. A sensor has no contacts, so there is no cycle-based wear figure in its datasheet either; what you get instead is MTTFd — 540 years for the NBB20-U1-UU with a Mission Time of 20 years, 1735 years for the IME12-04BPSZC0K. MTTFd is failure statistics, not a promised service life.

What if the target is not metal, or the gap is more than 20 mm?

An inductive sensor is the wrong tool: it responds to metal only, and the assured zone is capped at 81 % of the rated distance by clause 7.2.1.3.3, before the material factor is applied. Non-metallic targets call for optical sensors, and large gaps for optical through-beam arrangements. We compared the principles against each other in the article on types of proximity sensor.

What to Do Next

If the replacement looks workable after these rules, send us five things: the voltage and load current at the old switch, the actual gap from sensor to target, the target material, the controller input type, and a photo of the mounting. We will check it against the datasheets, calculate the zone for your metal and propose a catalogue item. If nothing fits your case, or the swap touches a safety function, we will say so plainly. Send those five items through the contacts page — email and phone numbers are there.

We do not sell limit switches and we do not take on selecting them. What we do have: inductive sensors, over 60 items, proximity sensors including magnetic ones for cylinders, and the full catalogue of industrial sensors. If you already know the old switch's mounting, start from the thread: M8, M12, M18 and M30 are the four common barrels, and all of them sit in the inductive sensor section. By brand: Sick, Pepperl+Fuchs, Omron, IFM. The current status of a specific part number is confirmed by a manager on request — write or call us.

Related material in this cluster: how an inductive sensor works and how to select one, replacing an inductive sensor, wiring a sensor to a PLC and the NPN versus PNP difference, testing a sensor with a multimeter.

Sources. Telemecanique/Schneider XCKN2121P20 datasheet (document XCKN2121P20_EN, 30 Jul 2024); Omron D4N datasheet (document C68I-E-02); IEC 60947-5-1, Annex C (durability, clauses C.1.1–C.3.1) and Annex K (direct opening action, clauses K.2.1, K.2.2, K.4.3.1.2, K.4.4, K.5.2.7, K.5.4.1); IEC 60947-5-2 (Tables 2, 4 and 5, clauses 2.1.1, 2.1.1.1, 7.2.1.3.3, 8.3.2.1.1, 8.3.4.1, 8.5); Pepperl+Fuchs NBB20-U1-UU, NBB2-8GM30-E2, NBB4-12GM50-E2-V1 and Sick IME12-04BPSZC0K, RZT7-03ZRS-KP0 datasheets.