What Sensor Hysteresis Means in Practice

Sensor hysteresis means the result depends on how an input was reached. A measurement sensor can report different values on an increasing and a decreasing input cycle. A switching sensor uses separate operate and reset thresholds. Understand which meaning applies before judging a specification—or treating a retained ON state as a fault.

Why can the same input produce a different result?

Because a sensor can respond to the path taken to reach an input, as well as to the input itself. The practical meaning depends on whether you are reading a measurement or watching an ON/OFF output.

For a measurement sensor

The same pressure, force or position can give different readings during an increasing and a decreasing input cycle. This directional difference is usually an unwanted contribution to measurement error.

For a switching sensor

The output changes at one threshold and resets at another. This intentional separation can prevent repeated ON/OFF transitions when the input fluctuates near a switching boundary.

A device with a measured process value and a separate switching output can have both: measurement hysteresis in the value and a configured hysteresis band for the output. Changing the output band does not correct the measurement channel.

The underlying cause is not always the same. A measurement chain can show direction-dependent behavior from its sensing element, friction or mechanical backlash. A switching circuit can deliberately retain its last state between two thresholds. The word alone does not identify a fault.

Why does a proximity sensor switch on and reset at different distances?

It uses separate operate and reset boundaries. For an axial target moving toward a proximity sensor, the face-to-target gap decreases. On withdrawal, the gap increases. OMRON calls the difference between these operating and reset distances hysteresis, or differential travel.

At the same gap, the previous state matters

Illustrative example: assume a normally open sensor turns ON as a target approaches through an 8.0 mm gap, and turns OFF as it withdraws through an 8.8 mm gap. These are assumed values, not an xsz sensor specification.

Approach: gap decreasesWithdraw: gap increases

At 8.4 mm, the blue approach path is OFF; the dashed withdrawal path is ON. This is an axial-distance example, not a timing plot or a lateral pass-by detection curve.

Starting far away and moving only to 8.4 mm leaves the output OFF: the target has not reached the operate point. Moving inside 8.0 mm first, then withdrawing to 8.4 mm, leaves it ON: the target has not reached the reset point.

There is no contradiction. Within the band, the output retains its previous state. This describes a normal powered operating cycle; do not infer the device's power-up state from the drawing. A normally closed output reverses the electrical ON/OFF logic, while keeping the distinction between detection and release.

A wider band trades easier stability for more reset travel

After switching, a small reversal that stays inside the band does not immediately switch the output back. That helps resist chatter. But the process must still cross the other boundary to produce the next transition. A wider band can be inconvenient when travel is limited or parts have very short clear gaps.

Hysteresis cannot guarantee stable detection if target motion or signal fluctuations cross both thresholds. For a lateral target passing across the face, use the relevant sensing-area curve rather than converting an axial hysteresis distance directly into a minimum part spacing.

How do you calculate hysteresis without comparing the wrong percentages?

First calculate the difference in physical units. Then divide by the reference quantity stated in the specification. A percentage without its denominator is not enough to compare sensors.

Switching distance: identify the reference distance

H = |reset distance − operate distance|

For the example: 8.8 − 8.0 = 0.8 mm.

Pepperl+Fuchs describes inductive switching hysteresis relative to the effective operating distance, s_r, at stated reference conditions. If that distance is 8.0 mm, the example gives:

0.8 mm ÷ 8.0 mm × 100 = 10%

Dividing the same gap by a 10 mm reference gives 8%. That arithmetic is valid only for a specification using that reference; it does not make the sensor more precise. Do not substitute a nominal range, housing diameter or convenient test distance for the declared denominator.

Measurement output: compare matched points on both paths

For a measurement sensor, compare the readings at the same input after approaching it from below and above. A full-cycle evaluation looks for the largest matched-point separation over the specified test range, not simply the difference between the two endpoint readings.

Increasing inputDecreasing input

Conceptual measurement loop, with separation exaggerated. Compare vertically at equal input. Neither the curve shape nor the position of the largest gap is universal.

Illustrative 4–20 mA calculation: suppose a controlled up/down test gives a largest matched-point output difference of 0.08 mA. If hysteresis is specified as a percentage of output span, the span is 20 − 4 = 16 mA:

0.08 mA ÷ 16 mA × 100 = 0.5% of span

Using 20 mA as the denominator would incorrectly give 0.4%. If only one input point was checked, report the difference at that point; it does not establish the maximum over the range. Follow the stated cycle, settling time and test-point method before comparing a measured result with a datasheet limit.

Is hysteresis the same as repeatability, deadband or response time?

No. Hysteresis describes a directional or state-dependent difference. The other terms describe different features of the result, even when they appear together in a troublesome application.

Separate the question before choosing the test.
Term	What to compare	Practical distinction
Hysteresis	Opposite input paths, or operate versus reset thresholds.	The same input can lead to a different reading or state.
Repeatability	Repeated equivalent events under the same stated conditions.	A boundary can repeat closely while remaining far from the other boundary.
Deadband	The interval over which an input change produces no specified output change.	Some switch manuals use this for the operate/reset band. Confirm the manufacturer's definition.
Response time	Input change and the corresponding output change in time.	Delay can shift a moving target's recorded position without changing the static hysteresis.
Drift	The result at the same reference input over a stated time interval.	A time-dependent shift is not, by itself, evidence of a direction-dependent gap.

For example, repeated switch-on positions might fall between 7.99 and 8.01 mm, while reset positions fall between 8.79 and 8.81 mm. Each direction has a 0.02 mm observed spread, yet the centers are 0.80 mm apart. Combining both directions into one dataset would obscure the distinction.

A switching deadband is also different from a near-field dead zone, where a target may be too close for valid detection. And when a reading changes during a constant-load hold, investigate time-dependent behavior such as creep separately from the up/down comparison.

Filtering and output delays affect timing. They may suppress short disturbances, but they do not remove an underlying measurement hysteresis loop or establish that a switching target has crossed the reset boundary.

How can you tell useful hysteresis from a real sensing problem?

Record both directions under controlled conditions, then compare sensor behavior with the machine's event record. One LED observation, one direction of motion or one PLC bit cannot separate hysteresis from geometry, timing and wiring effects.

Define the measured quantity.Use face-to-target gap for an axial proximity test, the actual reference input for a measurement sensor, or a clearly defined process value. Identify output logic, settings and which signal is being recorded.
Keep the physical setup stable.Use a suitable reference, fixed target and rigid fixture. Hold supply, temperature, filtering and sampling conditions as constant as the method requires. Isolate machinery before changing mounts or entering a hazardous area.
Run a complete, repeatable cycle.For a switch, start clearly released, approach through operation, then withdraw through reset. For measurement, traverse the specified increasing and decreasing input range and record matched points after the defined settling period.
Repeat before interpreting.Keep separate records for the two directions. Report both the directional gap and the within-direction spread. Include the reference instrument's resolution and uncertainty when judging small differences.
Then check production conditions.If the slow-cycle result is stable but fast-cycle records change, examine sensor response, input filtering and controller capture. If both boundaries move after a fixture or target change, investigate that change rather than assuming the hysteresis setting is responsible.

Why an output can stay ON without being faulty

Illustrative scenario: a positioning flag moves inside the example's 8.0 mm operate distance, then retracts only to 8.5 mm. The controller expects a new ON edge on the next stroke, but none appears.

The first question is whether the sensor ever reset. With the assumed 8.8 mm reset point, an 8.5 mm withdrawal remains inside the hysteresis band. The expected state is still ON, so another inward stroke does not create a new rising edge.

What this changes: check that the return path reaches a reliably released condition. A supported threshold adjustment or different geometry may be appropriate, but reducing a PLC timer cannot create a missing physical reset. Any revised setup must still provide reliable detection and release across the actual operating variation.

This example explains one mechanism; it does not diagnose every stuck output. If the target is clearly outside the validated release condition and the sensor still does not reset, check configuration, other detectable material and the electrical output path.

Should you reduce hysteresis or choose a different sensor?

For bidirectional measurement, lower specified hysteresis generally makes the error budget easier to meet. For ON/OFF control, the useful band is one that resists expected disturbances while still allowing the process to operate and reset reliably. “As small as possible” is not a universal selection rule.

When measurement accuracy is the priority

Check whether hysteresis is a separate limit or already included in an accuracy specification. A real example is the WIKA SA-11 datasheet, PE 81.80, March 2026, page 3: its accuracy footnote includes hysteresis along with non-linearity, zero offset and end-value deviation. That inclusion is a reason not to add the same hysteresis term again automatically; it is not a separate hysteresis rating for an xsz sensor.

A consistent one-direction measurement procedure may reduce the variation encountered in that procedure, but it does not remove hysteresis or establish performance after reversals. Zero/span adjustment alone cannot make two different outputs at the same input become one correct bidirectional response.

When switching reliability is the priority

Compare both thresholds with the real target travel and the expected disturbances. If the sensor supports adjustable hysteresis, use its documented parameter limits and retest both transitions. Do not assume that teaching sensitivity also provides independent control of the band.

Before comparing two products, make these four items explicit:

Meaning: measurement error, switching differential, or configured output band.
Reference: engineering units and the denominator behind any percentage.
Limit: typical value, maximum specification, adjustable setting, or one observed test result.
Conditions: input cycle, target, mounting, temperature, supply and applicable output settings.

The practical takeaway: record how the input arrived, measure both directions, and keep the absolute gap beside its percentage. For a switch, successful detection includes a reliable reset. For a measurement sensor, a good one-way calibration does not prove a good return path.

Sources and method references

OMRON — Proximity Sensors: Explanation of Terms. Differential travel, output logic, response time and approach geometry; includes the manufacturer's distance diagrams.
Pepperl+Fuchs — Inductive Sensors: Output Logic. Switching hysteresis referenced to effective operating distance under stated conditions.
NI — Sensor Terminology and NIST — Quantifying Uncertainties from a Gauge Study. Direction-dependent readings and full-range matched-point comparison.
Novotechnik — Technical Reference, section 8.2. A position-transducer example of direction-dependent mechanical behavior and a manufacturer hysteresis plot.
Texas Instruments — Comparator Chatter, SSZT122. How threshold separation and filtering affect chatter and timing.
WIKA — SA-11 Datasheet, PE 81.80, March 2026, page 3. A documented example of hysteresis included within an accuracy specification.

Numerical examples and simplified drawings are illustrative, not measured product performance. Manufacturer references describe their stated products or methods; they do not establish xsz sensor model specifications.