FANUC G54.2 dynamic fixture offset: why rotary-table center error appears as the angle changes
The 16i/18i-MB dynamic fixture offset definition supports a planar single-table model. Calculating an eccentric workpiece origin at 90° explains how a correct reference-angle setup can still produce position error after the table rotates.
Contents
An eccentrically mounted workpiece can be set correctly at the table's zero-degree position yet show a different position error after rotation to 90°. This cannot be explained simply as needing another G54 adjustment. The origin depends both on the table center and on an eccentricity vector that rotates with the table.
In this edition, G54.2 changes the fixture-offset vector
The FANUC 16i/18i-MB operator's manual B-63534EN/02, §14.14, pages 384–385 (public mirror of the original manual) explains that the control calculates the current offset from the reference fixture offset, reference angle and table angle, then adds it to the workpiece-origin offset. Commands for the associated rotary axis use the block-end angle to calculate the vector. The machine must have the function, and the rotary axis must be correctly associated with the linear axes forming its rotation plane.
Separate center position from eccentricity
Original XY-plane model with one C table: positive rotation is counterclockwise when viewed from +Z. The reference angle is 0°, the table center W=(500,300) mm, and the workpiece origin relative to that center is f₀=(30,40) mm, with no other coordinate transforms. With R(θ) denoting planar rotation, the workpiece origin is O(θ)=W+R(θ)f₀.
| C angle | Rotated eccentricity vector, mm | Workpiece origin, mm |
|---|---|---|
| 0° | (30,40) | (530,340) |
| 90° | (−40,30) | (460,330) |
| 180° | (−30,−40) | (470,260) |
Using the zero-degree origin (530,340) at 90° leaves a difference of (70,10) mm from the correct origin, with distance about 70.711 mm. The table calculates origins only; it does not prescribe the path of rotary and linear axes between rows.
Why correct alignment at zero degrees does not prove the center is correct
Extend the model with an idealized error analysis. Suppose the center used, W′, exceeds the true center by e=(0.10,0) mm, while the reference eccentricity is set to f₀′=f₀−e to align the same workpiece origin at zero degrees. The zero-degree result agrees exactly, but origin error at any angle becomes ΔO=e−R(θ)e.
The error is therefore (0,0) at 0°, (0.10,−0.10) mm at 90°, with magnitude about 0.1414 mm, and (0.20,0) mm at 180°. A measurement at the reference angle hides center error inside the reference vector; rotation breaks that cancellation. This is an original rigid-body model, not a unique diagnosis of actual machine error.
Verify the same datum at different angles
Compare repeatable measurements of the same datum at different angles with the vector changes predicted by the model. Checking only zero degrees cannot distinguish a correct center from two errors that temporarily cancel. Axis-positive directions, reference angle and rotation plane must also be consistent.
Page 389 of this edition additionally explains that a change in offset vector can cause motion. These coordinate calculations must not be treated as an operating path for enabling or disabling G54.2, or as equivalent to continuous TCP tool-tip control. This article gives no center-parameter values to write. See the OPTION section for related function distinctions.