FANUC interpolated straightness compensation: why position-dependent corrections cannot be replaced by a zero offset
Within the 0i-F Plus straightness-compensation scope, an original cumulative correction curve compares a constant zero offset, interval interpolation and incremental changes. The analytical curve is kept separate from raw controller compensation data.
Contents
As a long-travel axis moves, geometric deviation in another direction may vary with position. If zeroing at one location looks locally correct but the stroke ends retain deviations in different directions, one constant zero offset cannot describe the problem.
The moving and compensation axes have different roles
The FANUC 0i-F Plus descriptions manual B-64692EN/01, §9.8–9.9, pages 250–251 (public mirror of the original manual) describes outputting correction on a compensation axis according to another moving axis's position. Reference positions must be established for both; the same axis cannot serve both roles here. The interpolated type distributes interval compensation as pulse output. Availability of this and accompanying compensation functions depends on the actual configuration.
For example, X position can be the independent variable for a correction in Y. This differs from correcting X itself based on X position and from applying a constant translation to the whole workpiece coordinate system.
Understand interval distribution through a cumulative correction curve
The following original geometric model is not a FANUC parameter table. Desired cumulative Y correction is c(0)=0, c(100)=+8 μm and c(200)=−4 μm, with X in mm. Assume ideal linear distribution within both intervals:
For 0≤x≤100, c(x)=8×x/100; for 100≤x≤200, c(x)=8−12×(x−100)/100. Results are in μm.
| X position/mm | 25 | 50 | 75 | 125 | 150 | 175 |
|---|---|---|---|---|---|---|
| Cumulative correction c/μm | 2 | 4 | 6 | 5 | 2 | −1 |
The second interval's change is −4−8=−12 μm, not its cumulative endpoint value of −4 μm. Incorrectly using −4 as the additional change starting at 100 mm gives endpoint +8−4=+4 μm, differing by 8 μm from the target −4.
Why changing one zero value cannot solve it
A uniform Y shift of +8 μm agrees with the target at X=100 mm but adds 8 μm too much at X=0 and 12 μm too much at X=200. A constant offset has only one constant; a position-dependent curve has different interval trends. They cannot replace each other.
Likewise, this ideal cumulative-displacement curve cannot be copied into controller compensation data. Actual table-entry definitions, detection units, multipliers, reference points and other active compensation all affect conversion from measurement to data. Pulse quantization also means actual output cannot be identified directly with a mathematically infinitely continuous curve.
Verify residual error over the entire stroke
Separate measured geometric error, desired additional cumulative correction and correction changes over individual intervals, with consistent sign conventions. After compensation, check whether residual errors at several positions change as expected. Zero at a single point does not establish improvement over the entire travel. This article supplies no universal parameter-entry or existing-compensation-stacking procedure. Further accuracy-control resources are in the OPTION section.