Classify each workpiece before comparing centering machines. Record diameter ranges, material, and required centering tolerance. A 10 mm lens and a 50 mm lens can react differently to the same holding force. Glass may chip at an edge; thin metal parts can flex in a chuck. These are practical screening factors, not universal rules. Verify them with representative samples.

ISO 10110-6:2015 defines how optical centering tolerances are specified; use the drawing’s stated tolerance rather than assuming a machine’s advertised resolution is sufficient. Separate eccentricity from tilt when reviewing the requirement, then check whether the machine can measure and correct both. A common shop-floor mistake is treating indicated runout as the whole tolerance. It isn’t. For measurement records, NIST Technical Note 1297 describes expanded uncertainty as U = kuc; k = 2 provides roughly 95% coverage under suitable conditions. Include that uncertainty when judging whether a result meets specification. This detail is easy to miss. Recheck it. Keep actual workpiece data, since nominal diameter alone may not reveal clamping distortion.

A centering machine can look precise on a brochure, but ISO 230-2 test data gives you a more useful comparison. The standard describes methods for measuring axis positioning accuracy and repeatability. Ask for the actual test report, not just a single accuracy figure. Check the tested axis, travel range, measurement direction, and ambient conditions. A result measured over a short stroke may not represent performance near the ends of your working range.

Compare bidirectional results, since an axis may behave differently when approaching a point from opposite directions. Look for positioning deviation and repeatability across several target positions. Small differences matter. For example, a reported deviation of a few micrometres could affect how consistently a component sits against the machine’s centering stops.

Confirm whether the values are measured before or after compensation, and whether the test setup used calibrated measuring equipment. The chart is not the whole story. ISO 230-2 data describes axis behavior under stated test conditions; it does not guarantee finished-part accuracy in every setup. Ask how temperature, fixture alignment, and workpiece variation are controlled during your own production checks. A missing condition in the report is worth asking about.

Estimate a centering machine from measured cycle time, not the catalog rate. If one part takes 18 seconds, an eight-hour shift has a theoretical ceiling of 1,600 parts. That assumes uninterrupted running. It ignores setup, loading, inspection, tool changes, and rejected parts. For a more useful estimate, subtract planned breaks and changeovers, then apply observed uptime and first-pass yield.

Batch size changes the picture. A 600-part batch may need several minutes of setup before the first piece is produced. Record setup time separately, especially when jobs switch between different diameters or materials. Then calculate:

usable production minutes × 60 ÷ cycle seconds × good-part rate.

Compare that figure with required daily volume, and keep a buffer for normal variation. Small delays matter. A slightly optimistic cycle-time estimate can leave a real order short.

Industry figures offer context, not a machine-specific promise. Deloitte’s 2025 Smart Manufacturing Survey reported production-output improvements of 10–20% among surveyed manufacturers adopting smart manufacturing technologies. Those gains are not a centering-machine benchmark. Track actual cycle times across several shifts, including a difficult batch, and revise the estimate when the data disagree. I would not trust one unusually smooth run. It feels convincing, but it may hide the interruptions that define an ordinary day.

7 Tips for Choosing the Best Centering Machine

A centering machine should prove repeatable performance on your actual parts, not just a polished sample. Request a capability study using the material, geometry, fixture, and operating conditions you expect in production.

NIST’s Engineering Statistics Handbook explains that capability indices are meaningful only when the process is stable and the data meet the method’s assumptions. Cpk of 1.33 is a common acceptance target, not a universal guarantee. Check the control chart, sample coverage, and distance to both specification limits. A strong average can hide drift.

Then examine the measurement system. The AIAG Measurement Systems Analysis Reference Manual, 4th edition, generally considers Gauge R&R below 10% acceptable; 10–30% may need justification, depending on the application. Confirm whether the reported percentage is based on study variation or tolerance. Keep that basis consistent.

Numbers need context. For a practical trial, have different operators repeatedly center parts, remove and reload them, and record readings across the working range. Watch the fixture contact points and the display’s smallest increment.

If measurement noise consumes too much of the tolerance, an impressive Cpk may simply reflect a weak gauge. Ask for raw data and rerun the study after adjustments. Even then, one successful trial is not proof of long-term performance.