Technology / Technical guide
Wind Sensors: Resolution, Repeatability & Hysteresis
A sensor can report many decimal places, repeat the same biased value and respond differently after a changing input. Those are distinct behaviours. Resolution, repeatability and hysteresis describe different limits on what a measurement tells us.
A consistent indication can still differ from the quantity.
Compare the meaning of each metrology term before interpreting it as a general accuracy statement.
01 / Resolution
A detectable change is not complete measurement accuracy.
Resolution concerns the smallest change in a measured quantity that produces a perceptible change in indication.
Resolution concerns the smallest change in a measured quantity that produces a perceptible change in indication. The BIPM vocabulary notes that noise or friction can affect it. Display increments and physical resolving ability are therefore related but not automatically identical.
A screen showing hundredths does not establish that two inputs separated by a hundredth can be distinguished reliably. Conversely, a coarse reporting format can hide detail produced earlier in the chain. The sensor, numerical conversion and display are separate boundaries.
02 / Repeatability
Repeated results require stated comparable conditions.
Repeatability concerns precision under a specified set of repeatability conditions.
Repeatability concerns precision under a specified set of repeatability conditions. Keeping the procedure, system and relevant conditions comparable isolates a particular kind of scatter. Changing the environment or instrument creates a different comparison.
A tightly grouped set of indications does not establish agreement with a reference. A stable bias can remain in every value. This separates the question “how scattered are repeated results?” from “how well does the result represent the intended quantity?”
03 / Bias illustration
Stability and reference agreement can disagree.
Hypothetical indication example: a reference quantity of 10 units repeatedly produces indicated values of 12.0, 12.0 and 12.0 units.
Hypothetical indication example: a reference quantity of 10 units repeatedly produces indicated values of 12.0, 12.0 and 12.0 units. The displayed scatter is zero in that small example, but the indicated difference from the stated reference is 2 units.
The example does not prove zero underlying uncertainty or establish a calibration method. It simply shows why excellent repeatability cannot by itself establish accuracy. The chosen unit is arbitrary and describes no MARSTON wind sensor.
04 / Hysteresis
The route to a value can affect the indication.
Hysteresis describes dependence on prior conditions: approaching a quantity after a rise can differ from approaching it after a fall or a cycle.
Hysteresis describes dependence on prior conditions: approaching a quantity after a rise can differ from approaching it after a fall or a cycle. This is different from unstructured repeat scatter. The relevant input history belongs to the meaning of a hysteresis claim.
Analog Devices describes a thermal-history example in semiconductor sensing. That illustrates history dependence, not a universal wind-sensor magnitude. Similar-looking discrepancies can also involve delay or drift, so history dependence should not be inferred from two unrelated readings alone.
05 / Different effects
One headline percentage can combine unlike quantities.
Drift concerns changes with time or conditions; nonlinearity concerns departure from a defined relationship across a range.
Drift concerns changes with time or conditions; nonlinearity concerns departure from a defined relationship across a range. They do not mean the same thing as resolution or repeatability. A specification that combines contributions needs its stated convention to be understood.
Likewise, a calibration result and a measurement uncertainty statement are broader than a decimal increment. The measurement-chain guide covers the full result, while these terms explain particular behaviours within it.
06 / Interpretation scope
A term needs its stated quantity and conditions.
Apply each concept to the actual reported variable, range and measurement boundary.
Apply each concept to the actual reported variable, range and measurement boundary. The filtering guide explains timing that can otherwise be confused with a history-dependent indication.
No MARSTON resolution, sensor precision, hysteresis magnitude or accuracy is assigned here. The examples explain terms without calibration, heating/cooling cycles, hardware adjustment or measurement procedures.
Continue the review with Wind Measurement Filtering: Bandwidth, Delay & Latency.
Continue the review with Yaw Feedback: Setpoint, Measurement & Closed-Loop Error.
Continue the review with Redundant Sensors: Voting, Agreement & Common Causes.
For separate threshold and event-processing concepts, review Schmitt thresholds and retained state and signal debouncing and event timing; neither establishes a function implemented in a particular sensor.
Continue the measurement-chain terminology review with ADC quantisation and effective bits and ratiometric sensing and reference voltage.
General background: BIPM VIM: Resolution and indication · BIPM VIM: Repeatability as a stated precision condition · Analog Devices: Thermal-history dependence in sensor indications. These references explain industry concepts; they are separate from the applicable MARSTON product and project documents.
Key concepts and conditions.
Read each quantity with its reference, units and stated conditions. A general relation does not establish a model-specific value.
| Concept / quantity | Meaning / relation | Conditions / limits |
|---|---|---|
| Resolution | Smallest perceptible quantity change | Depends on the indication and relevant conditions. |
| Repeatability | Scatter under comparable repeated conditions | Agreement with a reference is separate. |
| Bias | Difference associated with systematic indication | Low scatter does not remove it. |
| Hysteresis | Different indication following different input histories | History and comparison conditions matter. |
| Uncertainty | Characterisation of the measurement result | Not replaced by one decimal or specification. |
Engineering questions.
Do extra decimals prove accuracy?
No. Reporting detail does not establish the whole measurement’s performance.
Does repeatability exclude bias?
No. Repeated values can agree with one another while differing from a reference.
Is hysteresis simply random scatter?
No. It concerns dependence on prior conditions.
Is a delayed reading automatically hysteresis?
No. Timing effects and history dependence need separate interpretation.
Does the 12.0-unit example define a real sensor error?
No. It is an invented illustration of stable bias.
Does this page specify a calibration procedure?
No. It explains terminology and measurement boundaries only.
Technical enquiry
Discuss the technical question.
A sensor can report many decimal places, repeat the same biased value and respond differently after a changing input. Share the model and applicable operating or design information for an engineering discussion.