Published:2026-10-19 00:00:00
Updated:
Linear actuators and worktables · Accuracy inspection
Turn precision into a repeatable acceptance requirement
Purchasing needs compliance with a drawing, commissioning needs stable settling, and inspection needs traceable results. Bringing parameter names, measurement conditions and decision methods together gives all three teams a shared basis for discussion.
01 / Definitions
First define what accurate means
Returning to the same place, reaching the specified coordinate and travelling along a straight path are different requirements. Naming the requirement correctly is the first step toward a useful acceptance test.
| Parameter | Question it answers | Define for inspection |
|---|---|---|
| Resolution | How small an increment can the control or measurement system distinguish? | Command resolution, feedback resolution or actual minimum incremental motion: these are not automatically equivalent. |
| Repeatability | How tightly do repeated arrivals at one target cluster? | Approach direction, repeat count, load, settling criterion and statistical method. |
| Positioning error and accuracy | How far is measured position from the target, and how is performance evaluated over the test range? | Target positions, unidirectional or bidirectional evaluation, compensation state and calculation definition. |
| Reversal difference | Do opposite approaches produce the same mean position? | Record both approach directions; the difference is not necessarily all mechanical backlash. |
| Travel geometry | How does the path or orientation relate to a datum or another axis? | Define straightness, travel parallelism, pitch/yaw/roll and squareness separately, with their datums and methods. |
A 1 μm command increment does not by itself prove reliable 1 μm mechanical motion or positioning error below 1 μm over the full travel.
02 / Single-target example
Tight repeats can still miss the target
Assume a target at 100.000 mm. Approach from each direction five times. Errors below are measured position minus target position in μm: +9 μm means an actual reading of 100.009 mm.
| Repeat | Positive approach / μm | Negative approach / μm |
|---|---|---|
| 1 | +8 | +2 |
| 2 | +9 | +3 |
| 3 | +10 | +4 |
| 4 | +9 | +3 |
| 5 | +9 | +3 |
| Sample mean | +9 | +3 |
| Sample range: maximum − minimum | 2 | 2 |
ei = xmeasured,i − xtarget ē = (Σei)/n Rsample = max(ei) − min(ei)
These formulas describe this small sample. Its observed range is not a complete repeatability statistic defined by a particular standard.The direction means differ by 6 μm. Combining all ten observations gives an 8 μm sample range. Reporting only a 2 μm variation over five repeats hides the direction difference and target offset. Nor can that 2 μm range simply be relabelled as ±1 μm positioning repeatability.
Formal acceptance must follow the agreed sampling, statistics and decision method. Five repeats are used here for illustration, not prescribed for every product.
03 / Travel and direction
Distribute targets across the working range
Extend the teaching example to five positions within a 200 mm stroke. Each position and approach direction again has five readings, with observed extrema 1 μm above and below its mean.
| Target / mm | Positive mean / μm | Negative mean / μm |
|---|---|---|
| 10 | +1 | −3 |
| 50 | +4 | 0 |
| 100 | +9 | +3 |
| 150 | +13 | +7 |
| 190 | +17 | +11 |
Each cluster remains narrow, while its mean changes with position. In a real test, such a trend warrants checking scale factors, screw lead errors, thermal changes, measurement layout and compensation. Curve shape alone cannot identify the cause.
Cover the agreed working range and leave room to approach from both directions. Add targets in a critical process region where useful, but distinguish local results from travel-wide evaluation. Rehoming reproducibility, loaded positioning and dynamic trajectories may require separate tests.
04 / Reading HZMotion data
Keep the model, parameter name and column heading together
C/P identifies configurations within specific HZMotion families. Matching letters across families do not guarantee matching values, and a screw grade is not an assembled-axis acceptance result. The following examples come from the HZMotion manual; all values are in mm.
| Manual reference | Model or configuration | Listed values | Interpretation |
|---|---|---|---|
| TKA/TKB, section 1.2.5; PDF page 46 | TKA, C / P | Positioning repeatability: C ±0.01; P ±0.005. | These are repeatability values, not a full-travel absolute-position error specification. |
| HZM worktables, section 2.1.2; PDF page 300 | HZM200-200, P grade | Positioning accuracy 0.012; repeatability ±0.003; travel parallelism 0.015. | The P accuracy column states full travel. Retain all three separate requirements. |
| Same worktable accuracy table | C / P column headings | C accuracy is headed /300; P is headed full travel. | Retain the stated evaluation length; do not invent linear conversions or add a ± sign. |
For the P-grade HZM200-200, repeatability of ±0.003 mm must not become a blanket machine accuracy claim of ±0.003 mm. Similarly, do not rewrite the listed 0.012 mm positioning accuracy as ±0.006 mm without a definition supporting that conversion.
These examples explain the catalog, not the final technical agreement. The table does not specify every test condition in the same location. Confirm load, mounting, approach directions, temperature, compensation and evaluation method. Two-axis tables also need separate squareness requirements.
05 / Inspection methods
Match the instrument and datum to the parameter
| Object | Possible measurement approach | Control |
|---|---|---|
| Axis positioning and repeatability | A laser interferometer, independent linear measurement system or suitable length-comparison arrangement. | Select suitable range and uncertainty; align the measurement line and control Abbe offset, thermal state, environmental correction and compensation settings. |
| Local repeated arrivals | A stable displacement sensor or comparator at one target. | Check range, resolution and uncertainty. One-point results cannot replace travel-wide positioning measurements. |
| Travel parallelism and straightness | An indicator with a suitable reference artefact, or an applicable optical method. | Define the reference plane/line, artefact quality, support and measurement location; distinguish table-surface variation from path error. |
| Angular motion and squareness | An appropriate angular system, or a square and indicator under an agreed method. | Identify the reference axis, span and mounting arrangement; single-axis positioning data are not a substitute. |
If the measurement line is offset from the process point, angular motion can create Abbe error. For small angles, one component is approximately e ≈ hθ. With h = 100 mm and θ = 20 μrad, the magnitude is about 2 μm. This is an independent geometry example; direction and sign depend on the layout.
A small following error at the motor encoder therefore does not prove equivalent accuracy at the tool tip or workpiece point. Use the multi-axis Abbe error and squareness guide to help define the inspection location.
Displayed digits are not measurement uncertainty. Calibration, fixtures, alignment, temperature and repeatability all contribute; review uncertainty together with the limit and acceptance decision rule.
06 / Acceptance record
Write a requirement another engineer can reproduce
- Fix the conditions before testingRecord model, serial number, mounting datum, fastening, load and centre of gravity. Agree warm-up, temperature range, feedback and compensation version.
- Keep the raw observationsRetain every target, reading, direction, speed, settling time and anomaly, not only the final headline result.
- Apply the agreed decision methodCalculate the specified parameters and retain raw curves, uncertainty and decision rules. Handle near-limit results as agreed rather than deciding through rounding alone.
| Field | Complete before acceptance |
|---|---|
| Machine and axis | Machine: ____; axis: ____; model/travel/configuration: ____; drawing revision: ____. |
| Parameter and limit | Full name: ____; direction(s): ____; evaluation range: ____; limit and units: ____. |
| Method and sampling | Method/standard edition: ____; targets: ____; repeats per direction: ____; approach distance: ____. |
| Duty | Load/centre of gravity: ____; mounting: ____; warm-up: ____; speed/settling: ____; temperature log: ____. |
| Measurement system | Instrument/ID/calibration: ____; datum and line: ____; environmental correction: ____; uncertainty: ____. |
| Controls and decision | Feedback: ____; compensation version/state: ____; raw-data file: ____; decision rule: ____; signatures: ____. |
A practical purchase requirement identifies the model and stroke, agreed load, mounting and thermal state, and the method edition for bidirectional positioning and repeatability. Its attachment then specifies targets, cycles, compensation, limits and decision rules. This is much more reproducible than a request for high precision.
If ISO 230-2 is selected, confirm applicability to the equipment and contract, and use the agreed edition. It addresses positioning accuracy and repeatability testing of numerically controlled machine-tool axes. The examples here are not a conforming test report.
07 / Questions
Accuracy and inspection questions
Does good repeatability imply accurate absolute position?
Not necessarily. Repeated arrivals can cluster tightly while missing the target. Check target error, approach-direction differences and multiple positions across the agreed range.
Can C and P grades be compared directly across families?
Not by the letters alone. Check family, model, stroke, parameter definitions and test conditions; one letter does not automatically mean one value.
Is maximum minus minimum over ten readings the standard repeatability value?
It is the observed sample range. Formal repeatability depends on the chosen method, statistical definition and direction combination; do not substitute one for another.
Does position compensation remove the need for mechanical checks?
No. It may improve repeatable systematic deviations, but cannot automatically remove random variation, backlash, load deformation or thermal drift. Retain before/after settings and verify the required conditions.
Can an indicator at one point certify the whole stroke?
It can support a suitable local check, but one position cannot represent the full travel. Select a measurement system and targets appropriate to the acceptance task.