When the Part Number Is Gone: A Specification Method for Cross-Referencing an Inductive Proximity Sensor

The sensor is discontinued, or it's on a 26-week lead time, and the machine can't wait. Somebody suggests "just find the same size sensor" — an M18 barrel is an M18 barrel, right? That sentence is how a plant ends up with a line that false-triggers every time a forklift passes, or a sensor that quietly stopped seeing its target when the range shrank by half. Cross-referencing an inductive proximity sensor is a specification exercise, not a part-number lookup: eleven parameters have to be captured off the original, roughly half are hard constraints that can't be traded, and the two that most often sink a substitution — the mounting classification and the target-material correction factor — are exactly the two people skip.

Why "the same size sensor" is not the same sensor

Physical interchangeability and functional interchangeability are different questions. A compatible alternative has to thread into the same hole and switch the same load at the same distance off the same target at the same speed in the same electrical and physical environment — and the failure mode when it doesn't is usually silent until the machine runs. Nothing on the new sensor's label warns you that its field now couples into the mounting bracket, or that stainless steel targets read at a fraction of the rated distance. The datasheet knew; nobody asked it.

So before comparing anything, capture everything.

The eleven-parameter capture sheet

Read every one of these off the original sensor's documentation — the datasheet, not the barrel, because the barrel doesn't state half of them: (1) rated sensing distance, (2) mounting classification — shielded/flush versus unshielded/non-flush, (3) output transistor type (PNP or NPN), (4) contact function (normally open or normally closed), (5) supply-voltage range, (6) housing size and thread, (7) connector or cable style and pinout, (8) ingress-protection rating, (9) switching frequency, (10) ambient and environmental limits, and (11) approvals and listings.

Then sort them — because they are not all negotiable, and knowing which is which is the method:

The rest of this article is the evidence behind the two rows that cause the most failed substitutions, plus the electrical rows people assume instead of read. The technical documentation cited throughout comes from Rockwell Automation and Eaton published literature — used here to teach the parameters, not to recommend any manufacturer.

The mounting trap: flush versus non-flush

Eaton's inductive-sensor catalog defines the terms: sensors come "shielded (also known as embeddable or flush mountable) and unshielded (non-embeddable or non-flush mountable)," and what the terms refer to "is the distance to surrounding metal that the device can be mounted." A shielded sensor can be mounted completely surrounded by metal; an unshielded sensor requires a metal-free zone whose size depends on both the sensor's size and its range type. Rockwell's sensor reference manual explains the physics: shielded construction concentrates the electromagnetic field at the front of the sensing face, which is what allows flush mounting in metal without false triggering.

Swap unshielded for shielded and you typically lose range you were counting on; swap shielded for unshielded into a flush-mounted bore and the surrounding metal becomes a permanent target — false trigger, or no trigger, depending on the day. The spacing rules are real enough that Rockwell's manual publishes separate tables for shielded and unshielded devices' clearances to nearby metal — consult your candidate manufacturer's own table rather than treating any one vendor's numbers as universal. For long-range devices, even the bracket enters the specification: Rockwell's extended-range manual instructs that metal mounting hardware sit as far back on the barrel as possible to avoid interference and false triggering, and recommends stainless steel or plastic brackets, especially for 30 mm units.

The correction-factor trap: the datasheet distance is not the distance you get

Rated sensing distance is specified against a standard target — per Rockwell's reference manual, "a mild steel square, one mm thick, with side lengths equal to the diameter of the active face or three times the nominal switching distance, whichever is greater." Your target is almost certainly not that. For anything else, the method is published: (nominal sensing range) × (correction factor) = sensing range. Rockwell's general-guideline factors: mild steel 1.0, stainless steel 0.85, brass 0.50, aluminum 0.45, copper 0.40 — with the manual's own caveat that the specific factors live on each product's specification page.

And here's a live demonstration of why that caveat is the most important sentence in the manual. Eaton's catalog works the same arithmetic on one of its own 18 mm sensors: a 5 mm sensing range on a brass target becomes 2.25 mm (5 mm × 0.45). Read that again next to Rockwell's table: Rockwell's general guideline puts brass at 0.50; Eaton's published size-dependent table puts brass at 0.45 for an 18 mm sensor. Neither is a typo — we verified both against each manufacturer's own document — and that disagreement between two reputable manufacturers on the same material is the single best argument in this article for using your candidate manufacturer's own published factors, never a generic table, and never the outgoing sensor's.

Now the rest of what breaks cross-references. The correction factor is not a property of the target material alone:

It varies by sensor size. Eaton publishes a size-dependent table where, for example, aluminum reads 0.35 on a 4–8 mm sensor but 0.45 at 18 mm, and 400-series stainless runs 0.90 on small barrels but 1.0 at 18 mm and up.

It varies by stainless grade. The same table separates stainless 400 (0.90–1.0) from stainless 300 (0.65–0.85). "Stainless steel" is not one material to an inductive sensor.

It can invert entirely for extended-range technology. Rockwell's 871TM extended-range manual publishes factors where brass reads 1.3 — better than steel — aluminum reads 1.0, and 1-mm stainless reads 0.45 on a shielded M12, 0.1 on an unshielded M18, and "no detection" on an unshielded M30. The same manual notes stainless sensing distances depend on target size, thickness, and grade, and that extended-range units — unlike conventional ones — lose range on targets under 1 mm thick and generally cannot detect foils.

Add the geometry rules — flat targets preferred, rounded or undersized targets reduce range, the rule-of-thumb target being three times the range or the size of the sensor face, whichever is larger — and one technology caveat: ferrous-selective sensors won't detect brass, aluminum, or copper at all, and nonferrous-selective sensors won't detect steel. This is why "rated Sn" sits in the context-dependent column: two sensors with identical headline distances can behave completely differently on your target. Run the arithmetic for the candidate sensor, your target material, your target size — then require margin.

(One vocabulary note: the standard target and operating-distance vocabulary for proximity switches is defined in IEC 60947-5-2 — current edition 4.0, published 2019 — and ingress-protection codes in IEC 60529. Both are paywalled standards; the correction-factor tables above are the manufacturers' published data, not clauses of the standard.)

The electrical rows people assume instead of read

PNP versus NPN decides which side of the load the output switches — per Rockwell's manual, an NPN (sinking) output takes the load between the output and the positive supply; a PNP (sourcing) output takes it between the output and the negative. Wrong polarity doesn't damage anything dramatic; the input simply never turns on. And you can't reliably infer it from a family name: Eaton's catalog describes four-wire DC families whose auto-configure technology detects a sinking or sourcing connection from the field wiring and switches the sensor accordingly. Read the datasheet, both of them.

NO versus NC are separate catalog numbers in adjacent columns of the same selection table. Miss it and your machine logic inverts.

Supply voltage is not a formality: one manufacturer's own inductive line spans families at 10–30 Vdc, 6–48 Vdc, 15–30 Vdc, and 12–24 Vdc, alongside AC families (one two-wire line runs 20–132 Vac). The candidate's range must contain your actual panel voltage.

Connector and pinout: the same sensing element is commonly offered as a cable, a pigtail, or a connector version, and connector families are standardized by pin count and physical series (4-pin M12 micro, 5-pin micro, mini, M8 pico) with the mating cordset as a separate ordering decision. A cordset change is often acceptable; a pin-count change is not. And verify the pinout against the specific datasheet — wiring color and pin conventions are not universal enough to assume, so we won't print a generic table here.

The fine-print rows: switching performance (transistor outputs switch DC only, with response times ranging from 2 ms down to 30 µs depending on the device — your candidate's switching frequency must exceed the fastest event the machine produces), published repeatability and hysteresis lines (one family publishes hysteresis <15% of rated distance, repeat accuracy <1%, and a <50 ms delay before availability — that last one matters on machines that power-cycle sensors), environmental limits including magnetic-field immunity (one extended-range family permits fields to 1,000 A/m — but only 80 A/m for the 30 mm barrel; relevant near induction heating and welding), IP rating treated as a floor for the real environment (our washdown ratings explainer covers what those codes do and don't promise), and approvals verified against your own panel and jurisdiction requirements.

The seven substitution errors, named

1. Swapping shielded for unshielded (or the reverse) because the barrel size matched. The metal-free-zone requirement changed; the false triggering arrives later.
2. Reading the rated sensing distance as the distance the machine will get. Eaton's own example: a 5 mm device sees brass at 2.25 mm.
3. Assuming a correction factor is a property of the material alone. It shifts with sensor size, differs between manufacturers for the same material, and inverts for extended-range technology — brass at 1.3, unshielded M30 giving no detection on 1 mm stainless.
4. Assuming stainless steel is one material. 400-series and 300-series carry different published factors, and thickness and grade matter.
5. Assuming the target is big enough. Undersized targets reduce range.
6. Assuming PNP/NPN from the family name. Some devices self-configure from field wiring; others are fixed. Verify both datasheets.
7. Treating the cordset as part of the sensor. Connector series and pin count are their own ordering decision.

Document it so the next person trusts it

A cross-reference that lives in one engineer's head gets re-derived — wrong — on third shift. Write it down: for each of the eleven parameters, record the value, the source document (datasheet number and revision, not "the catalog"), and whether the candidate meets the same published specification for that parameter or is merely acceptable for this application and why. That phrasing is deliberate: a compatible alternative meets the same published specification for the parameters that matter in your application — it is never simply "the same as" the original, and the substitution isn't done until it's been tested on the machine, against the real target, at the real speed, before you commit inventory to it.

Two closing thoughts for the planner. If the sensor that died was wired discrete and the machine is due for a controls refresh, an obsolescence event is a natural moment to weigh whether the replacement should speak IO-Link instead — and occasionally whether the sensing technology itself is still the right one, as it increasingly isn't for some photoelectric jobs. And if this method feels like more work than a lookup table — it is. It's also the difference between a substitution you can defend in the morning meeting and one you explain at 3 a.m.