Figure 1. Lanbao PDA series product image from the official product page. The image illustrates the product family; confirm the complete model suffix before applying specifications. Source [2].

Separate repeatability from other measurement specifications

Repeatability concerns the spread of successive observations under essentially unchanged conditions. Reproducibility examines changed conditions, such as another operator, another installation, or another day. NIST distinguishes these concepts and recommends stating how precision is expressed, including the relevant statistical measure. 

Resolution is the smallest indicated or distinguishable increment. A display can show fine increments while the underlying readings wander by a larger amount. Bias is a systematic difference from a reference value. Linearity describes departure from a specified relationship across the measurement range. A sensor can repeatedly return the same wrong value, so excellent repeatability cannot remove calibration error.

Read percentage specifications carefully. For an illustrative 30 mm full scale, ±0.1% corresponds to ±0.030 mm, or ±30 μm. That calculation converts a range-based specification into useful engineering units. It does not mean that every measurement has exactly that error, and it does not convert linearity into repeatability.

Ask what the repeatability number actually represents

Two quotations can both state 20 μm and describe different performance. One supplier may report three standard deviations; another may report the entire observed peak-to-peak range. Peak-to-peak results also depend on sample count and outliers. Ask for the calculation method, number of samples, raw readings where available, and whether the number is a limit or a typical result.

Lanbao's dedicated PDA-CR50 series document identifies a 90% reflectivity white card, 23 ±5°C conditions, 24 VDC supply, and a 30-minute warmup. It also mentions a 2 ms sampling period and 100-sample averaging in its test notes. [1] Before comparing a quotation with that document, ask how those settings relate to the quoted response time and repeatability. Do not assume a filtered laboratory result represents the fastest operating mode.

For a procurement comparison, freeze one measurement recipe. Define the target, reference distance, surface orientation, stabilization time, acquisition duration, output interface, and filtering. Request an equivalent test from each candidate supplier. Where equivalent settings are impossible, describe the difference explicitly instead of ranking the bare numbers.

Understand why real surfaces change the result

Many short-range laser displacement sensors use optical triangulation. The receiver observes the position of reflected light, and the processing system maps that position to distance. Surface conditions can therefore influence the measurement even when the physical distance remains constant. The effect must be evaluated for the specific optical design and part.

A polished surface can direct useful light away from the receiver or create strong specular reflections. Dark surfaces may provide less returned light. Texture can change the observed signal as different surface features pass through the spot. At an edge, the illuminated area may include two heights. A larger spot can average local detail that the inspection is intended to resolve.

Mount the sensor so the emitted and received paths remain unobstructed throughout the expected movement. Check pockets, raised edges, and fixtures for optical shadowing. Use the actual range of surface finishes, including acceptable production variation. Testing one attractive sample can hide the failure mode that appears after the next material batch arrives.

Build an inspection error budget before selecting hardware

Start with the drawing tolerance and the consequence of a false accept or false reject. Establish an allowable measurement contribution with the quality team. A fixed measurement-to-tolerance ratio can be a useful internal target, but no single ratio automatically proves suitability for every product or risk level.

Consider an illustrative height tolerance from 9.80 to 10.20 mm, a total width of 0.40 mm. Suppose a pilot test produces a stationary standard deviation of 0.006 mm. Six standard deviations span 0.036 mm, or 9% of the tolerance width. This is a screening calculation for that stationary test, not a complete gauge acceptance result. It excludes repositioning, bias, temperature, and long-term effects.

If independent standard uncertainty contributions are estimated as 6 μm for sensor noise, 8 μm for repositioning, and 4 μm for acquisition, their root-sum-square is approximately 10.8 μm. This calculation is valid only when the estimates share a consistent basis and the independence assumption is reasonable. Correlated effects, uncorrected bias, and specified maximum limits require separate treatment; they should not be inserted blindly into the same formula.

A useful purchasing discussion therefore starts with the application budget. If the fixture dominates variation, a more expensive sensor may add little value. If surface effects dominate, changing geometry or sensing technology may matter more than a smaller catalog repeatability value.

Validate stationary and moving measurements separately

Use a traceable reference or an appropriate calibrated displacement stage to check known positions. The reference uncertainty must be suitable for the decision being made. Record ambient conditions and let the system stabilize according to the manufacturer. Preserve the settings and raw data so the result can be repeated after installation.

1. Measure one fixed reference repeatedly to establish short-term noise.

2. Remove and replace the same part to expose loading and fixture variation.

3. Measure representative parts across the required range and surface finishes.

4. Repeat at production speed with the real trigger and acquisition path.

5. Repeat after warmup, temperature changes, and normal cleaning or maintenance.

6. Compare the results with agreed limits and investigate failures before release.

These quantities are proposed starting points, not standard-mandated sample sizes. Increase coverage according to process diversity, confidence requirements, and the cost of an incorrect decision. A short trial can identify a problem without proving long-term reliability.

Balance averaging against line speed

Averaging can reduce random noise, but it can also delay the result and smooth a feature that should be detected. For independent random samples, the standard deviation of an average decreases with the square root of the sample count. Correlated noise, slow drift, and surface-dependent errors do not necessarily follow that improvement.

In an illustrative moving inspection, a 5 mm feature traveling at 0.5 m/s remains under a point measurement location for about 10 ms. A nominal 40 ms response setting is longer than that observation window. The feature may be attenuated or mixed with its surroundings. A faster setting may be necessary, followed by a fresh repeatability test at that setting.

Account for trigger delay, controller scan time, communication updates, and any averaging performed in the PLC. The sensor display may look stable while the PLC reads stale or differently filtered data. Evaluate the value actually used for the production decision.

Check integration and acceptance rules

For an analog interface, verify scaling, input resolution, grounding, cable routing, and the loop load. For a digital interface, verify units, register interpretation, update rate, invalid-data handling, and timeouts. Neither interface automatically guarantees a better inspection result; the complete implementation determines performance.

Set an explicit decision rule near specification limits. A measurement close to a limit may require a guard band, remeasurement, or an alternative inspection method, depending on the uncertainty and agreed quality policy. Do not hide uncertain classifications by repeatedly measuring until a passing value appears.

Procurement can use the Lanbao PDA product family as a candidate for evaluation, while requesting the exact model datasheet, output configuration, and application test. [2] Release the sensor only with a recorded baseline, mounting drawing, settings file, verification procedure, and replacement instructions. Those records help preserve the validated performance after maintenance.

References

[1] Lanbao PDA CR50 high precision series datasheet and test conditions

[2] Lanbao PDA laser displacement sensor product family and image source

Frequently Asked Questions

Is repeatability the same as accuracy

No. Repeatability describes agreement between repeated readings under defined conditions. Accuracy concerns agreement with the quantity being measured. A stable offset can produce excellent repeatability while the reported distance remains unsuitable for an absolute inspection decision.

What repeatability is good enough for my inspection

Choose a requirement from the part tolerance, uncertainty budget, and consequences of incorrect classification. Validate the complete installed system. A catalog value alone cannot establish an appropriate acceptance limit for an unknown process.

Can a sensor with 10 micrometre resolution inspect a 10 micrometre tolerance

Resolution alone does not support that conclusion. Noise, bias, linearity, thermal effects, fixture variation, and acquisition behavior may be larger than the displayed increment. The tolerance requires a suitable measurement capability study.

Why are readings less stable on shiny metal

Reflective surfaces can change the light reaching the receiver as angle or position changes. Test the actual finish and geometry. A different mounting angle, optical arrangement, or sensor technology may be needed after application evaluation.

Does averaging always improve repeatability

Averaging often reduces uncorrelated random variation. It does not reliably remove drift or systematic error, and it can reduce the visibility of short features on moving parts. Compare settings at the required production speed.

How often should the sensor be checked

Set the interval from process risk, observed drift, environment, and maintenance history. Check after installation changes or sensor replacement. A reference check at defined production intervals helps reveal changes before they affect a large batch.

What should be sent to a supplier for a useful trial

Provide representative parts, drawings, tolerances, surface variation, mounting constraints, line speed, controller details, and the desired decision rule. Request raw measurements and complete settings so the trial can be reproduced on the machine.