Good reference parts realistically map the planned application area
A reference set should contain typical materials, sizes, characteristics, machines and lot sizes. Only simple or particularly well run parts provide a distorted picture of the later accuracy.
Calibration is done on a part of the data set; testing is done for separate cases. Deviations and outliers per part class are reported, not just a single average value.
Key points
Calibration is explained in such a way that inputs, procedures, effects and limits in the CNC calculation remain comprehensible.
- Select representative: Take typical and difficult cases from the real field of application.
- Separate data: Do not perform calibration and independent testing with the same cases.
- Explanation of deviation: Evaluate errors by cause, part class and time type.


