Motion Platforms: Resolution Is Not Positioning Accuracy

A motion controller that accepts tiny steps does not necessarily place a tool accurately at those positions. Command resolution, repeatability, and accuracy describe different properties. Confusing them can make a promising prototype look more precise on paper than it is on the workbench.

This distinction matters for small fabrication machines, inspection stages, and desktop robotics. Use a measurement plan tied to the intended operation rather than treating a motor specification as a complete system specification. NIST Technical Note 1297 provides a foundation for expressing measurement uncertainty; the practical test plan below is an illustrative engineering approach, not a calibration certificate.

Define the three quantities

Resolution is the smallest change a system can represent, command, or distinguish in the relevant context. A controller may command a one-micrometre increment even when mechanical friction prevents that increment from producing a corresponding physical movement.

Repeatability describes how closely repeated positioning attempts agree under stated conditions. A stage can repeatedly return to the wrong position. Accuracy concerns agreement with the intended or reference position, taking the measurement context into account.

State the conditions with every claim: axis, travel position, direction of approach, load, speed, temperature, and measurement instrument. “Accurate to five micrometres” without those conditions leaves too much room for different interpretations.

Start with a reference you can trust

Choose a measurement method appropriate to the expected errors and travel. The reference instrument must have adequate range and a known uncertainty contribution. A display with many decimal places is not automatically a suitable reference.

Mount it so alignment and contact forces do not dominate the result. If the probe approaches at an angle, it may measure a projection rather than the axis motion you intended. Keep the mounting arrangement fixed during a comparison and record any adjustment.

Let the machine and instrument reach a stable operating condition. A cold test and a warmed-up test answer different questions. If normal operation includes warm-up drift, measure that drift rather than removing it from the report without explanation.

Measure repeated approaches from one direction

Choose a target within the working travel. Move away, approach from the same direction at the same speed, and record the reference reading. Repeat enough times to see variation; ten trials can be a useful small diagnostic, not a universal qualification standard.

For an invented set of results, suppose the commanded target is 10.000 mm and readings cluster between 10.018 and 10.022 mm. The four-micrometre observed range suggests close agreement among those trials, while the roughly twenty-micrometre offset suggests a positioning bias.

Do not describe that as “four-micrometre accuracy.” It is a limited repeatability observation at one position under one test setup. The reference uncertainty and systematic effects still matter.

Reverse the approach direction

Repeat the same target approaches from the other direction. Suppose those readings now cluster near 9.988 mm. The difference between directional clusters points toward reversal effects such as backlash, compliance, or friction; further diagnosis is needed to distinguish them.

Test several positions across the axis. An error that changes with travel may require a different explanation from a nearly constant offset. A lead-screw scale error, for example, should not be treated as the same problem as direction-dependent lost motion.

Include the actual working load. A stage carrying a light indicator may behave differently when supporting the tool, fixture, cable drag, or acceleration used in production. A no-load demonstration is not evidence of loaded performance.

Test small commands honestly

Near a stable position, command several small increments and observe whether the reference registers a corresponding change. Repeat in both directions. Record delayed movement, uneven increments, and hysteresis rather than rounding them into the commanded sequence.

Microstepping can improve motion smoothness and command granularity without guaranteeing that every microstep produces an equally accurate physical displacement. Treat the complete electromechanical system as the object under test.

Avoid using the controller’s own reported position as the independent measurement if it merely counts commanded steps. That would verify arithmetic inside the controller, not actual tool position.

Build a useful uncertainty and decision record

List relevant contributions: reference calibration, resolution, mounting alignment, thermal drift, repeatability, and load effects. Do not simply add decimal places or claim an uncertainty smaller than the evidence supports. For formal acceptance, use a qualified metrology process appropriate to the application.

Define the allowable positioning error before tuning compensation. If a process tolerates 0.1 mm, a costly micrometre-level optimization may not help its actual failure rate. If a process requires tighter control, explain how acceptance accounts for measurement uncertainty near the limit.

Record raw readings before calculating summary statistics. Preserving the sequence lets reviewers see warm-up drift or alternating directional errors that a single average would conceal.

Conclusion

Judge a motion platform by measured behavior under stated conditions. Command resolution is a starting specification; bidirectional accuracy, repeatability, load response, and uncertainty determine whether the machine can perform the job.