A PNP proximity sensor sources current to the load (it switches the +24 V rail), while an NPN proximity sensor sinks current to 0 V (it switches the negative rail) [S1][S2]. The output type does not, by itself, set the nominal sensing range; an M12 inductive unit specified at 4 mm will read 4 mm whether you order it as PNP NO, NPN NO, PNP NC, or NPN NC [S2][S5].
What does change with output type is wiring, PLC input-card compatibility, and the behaviour under a wiring fault. In a typical 24 V DC three-wire circuit, the brown wire is +V, blue is 0 V, and black is the signal line, and the sensor internally uses either a PNP or an NPN transistor to tie that signal line to +V or to 0 V when target metal enters the field [S2][S3]. Because the difference lives in the BJT doping stack (P-N-P versus N-P-N), a single vendor can ship the same housing in either polarity without changing the coil, oscillator, or trigger threshold [S5].
What "sensing range" actually means on a proximity sensor datasheet
Nominal sensing range (Sn) is the distance at which a standard target (mild steel, 1 mm thick, square, side = sensor face diameter or 3 × Sn, whichever is greater) is detected under no-load, 23 °C reference conditions [S2]. The rated Sn is the same for PNP and NPN builds of the same housing, because the output transistor sits downstream of the detector stage.
Real-world operating distance (Sr) is 0.81 × Sn, and the "assured" on/off window is 0 ≤ Sr ≤ Sr_max, a 10 % hysteresis band between switch-on and switch-off that prevents chattering on a vibrating target [S2]. Temperature drift, target material (the so-called "reduction factor", roughly 1.0 for mild steel, 0.4-0.5 for stainless steel, 0.3 for aluminium and copper), and flush versus non-flush mounting shift Sr by 10-50 % [S2]. None of those parameters are output-type dependent.
Typical industrial ranges stack as: M5 inductive 0.8-1.5 mm, M8 1-3 mm, M12 2-6 mm (standard 4 mm), M18 5-12 mm, M30 10-20 mm; capacitive sensors, used for non-metallic targets, run 1-30 mm depending on face size and target dielectric constant [S2]. Both NPN and PNP bodies are available across that full range.
Output-type selection matrix: how to pick PNP vs NPN
PNP (sourcing) is the default in Europe and North America because most modern PLC input cards are "sinking" type, that is, they expect the field device to deliver +24 V into the terminal [S2][S4]. The signal-line voltage is active HIGH: when the target enters range, the black wire rises to roughly +V (typically within 1-2 V of supply) [S3][S5].
NPN (sinking) is the default in legacy Japanese equipment and many Asian-built machines, and pairs with PLC input cards configured as "sourcing" [S2][S4]. The signal-line voltage is active LOW: when the target enters range, the black wire is pulled to 0 V, and current flows from the PLC, through the sensor, to ground [S3][S6].
Ground-fault safety is the third axis. If a PNP signal wire is pinched to chassis, the input sees a hard 24 V regardless of target state, a fail-as-on condition. If an NPN signal wire is pinched to chassis, the input sees 0 V regardless of target state, a fail-as-off condition [S2][S4]. In a machine where "sensor shows target" must default to a safe stop, NPN is the conservative pick; in a machine where a missed target must halt motion, PNP is. The next two sub-sections lay out the decision rule numerically.
Side-by-side comparison: PNP vs NPN on the criteria that matter
Switching rail: PNP closes the +24 V side of the circuit, NPN closes the 0 V side [S1][S2]. Signal state on detect: PNP = HIGH (≈ +V), NPN = LOW (≈ 0 V) [S3][S6]. Current direction: PNP flows out of the sensor into the load, NPN flows from the load into the sensor and out to ground [S2][S5]. Required PLC input card: PNP → sinking input module, NPN → sourcing input module [S2][S4]. Regional default: PNP in Europe and North America, NPN historically in Asia [S2][S4]. Ground-fault behaviour: PNP wire-to-chassis reads as "target present", NPN wire-to-chassis reads as "target absent" [S2][S4].
Leakage current and voltage drop are the two second-order numbers. A typical PNP output drops 1.5-3 V at 100-200 mA load, an NPN drops a comparable 1-2.5 V, and both show 0.1-0.5 mA off-state leakage, which matters only when the load is a high-impedance PLC input [S2][S3]. Switching speed for both polarities is 0.5-5 kHz, again polarity-independent because the driver stage is symmetrical [S2].
Power-supply polarity is the one field-fatal gotcha. NPN and PNP sensors are not interchangeable on the same wiring loom: feeding +24 V to a sensor whose NPN output transistor expects the load on the high side will not work, and a reversed brown/blue pair on a 4-wire DC sensor can destroy the regulator [S2][S6]. A complete design rule of thumb: match output type to the PLC input-card polarity, then verify the sensor's brown/blue pinout against the same PLC's field-wiring diagram before energising.
Wiring, IO-Link, and when output type stops mattering
Three-wire hookup is identical in pattern: brown to +24 V, blue to 0 V, black to the PLC input terminal, with the load (input card optocoupler) wired so that current flows through it on detect [S2][S3]. Four-wire versions add a second output (typically one NO and one NC) or a complementary pair; the second transistor is the opposite polarity of the first, so a single housing can present both PNP and NPN to two different loads [S2][S6].
Push-pull outputs (sometimes labelled "PP" or "PNP/NPN selectable") drive the signal line actively to either rail on detect, which lets one device work with both sinking and sourcing PLC inputs and removes the polarity decision entirely at the cost of a slightly higher off-state leakage (~1 mA) [S2]. IO-Link takes the decision out of human hands altogether: the digital protocol sits on a 3-wire cable and the master module handles sourcing/sinking translation, so the field device no longer needs a fixed PNP or NPN personality [S2].
For a primer on the inductive-detection physics that is identical under both output types, see the proximity sensor fundamentals reference. When the same body ships in P12, P18, and P30 form factors, the proximity probe form factor guide covers which housing suits which Sn.
Who should choose PNP, who should choose NPN, and who should not be choosing
Choose PNP if the PLC is European or North American (Siemens S7-1200, Allen-Bradley CompactLogix, Schneider M340), if the cabinet standard is CE/UL with PNP sourcing as the documented default, or if a fail-as-on fault mode is acceptable [S2][S4]. Choose NPN if the machine is Japanese-built (many Mitsubishi FX, Omron CP1E, legacy Keyence KV), if the panel spec already calls NPN, or if a fail-as-off fault mode is required by the risk assessment [S2][S4].
Do not choose based on a single stock-number preference; mismatched polarity is the most common cause of "sensor never turns on" field calls, and the fix is a re-wire, not a firmware change [S2][S6]. For long-range metal detection in stainless-steel-rich panels, remember the reduction factor is independent of output type: a 4 mm Sn inductive PNP and a 4 mm Sn inductive NPN will both drop to roughly 1.6-2.0 mm on a 304 stainless target, so re-spec the housing before you re-spec the wiring.
A common field failure worth naming: an NPN proximity sensor wired to a PNP-only input card will read "always off", while a PNP sensor on an NPN-only input card will read "always on", and in both cases the sensor itself is healthy, the miswire is the fault [S2][S6]. For shops chasing that exact symptom on an existing line, false triggering from adjacent brackets is a separate but easily-confused failure mode; confirm the wiring polarity first before chasing mounting geometry.
Selection rule, one line, and what to verify before energising
Match the output type to the PLC input-card polarity, confirm +V on brown and 0 V on blue, then verify the signal line swings to the expected rail (≈ +V for PNP, ≈ 0 V for NPN) when a steel target crosses the face [S2][S3][S6]. Sensing range, hysteresis, and reduction factor are not on the decision tree, because those numbers are fixed by the housing and target material, not by the output transistor.
Two trackable signals: (1) the IEC 61131-2 input-card type, sinking or sourcing, must be confirmed from the PLC datasheet, not assumed from the sensor's home market; (2) on 4-wire complementary sensors, the second output's polarity must be re-verified after any harness repair, because crossed pairs on brown/blue remain the dominant field-failure cause [S2][S6].
The underlying component specifications are covered under dry type transformer.