FANUC torque limit skip: why the recorded position may not be the mechanical contact point
Commanded position, actual position and following error are separated for 30i-A torque limit skip. Two ideal records of the same contact point show why a recorded position difference cannot directly establish a change in workpiece dimensions.
Contents
A coordinate obtained through torque limit skip may include servo following error if it is treated directly as the workpiece contact position. The issue is not just the number of decimal places recorded, but what the recorded position means.
Commanded position can continue changing after mechanical motion stops
According to the 30i/31i/32i-A common user manual B-63944EN/03, §4.22, pages 145–147 (public mirror of the original manual), stopping at mechanical contact and reaching the torque limit to end the block occur at different instants. The CNC may continue accumulating position commands in between. This article uses skip-position variables #5061–#5080 for systems with up to 20 axes; the manual's branch for more than 20 axes uses #100151–#100182. Whether the record is compensated for servo error depends on the function configuration. Do not assume that it always stores an uncompensated command value or always equals the actual contact point.
The same contact point can produce two different records
The following original, ideal one-dimensional model is not a field measurement. Assume a rigid contact point A=100.000 mm, where the mechanism stops in both attempts. By the skip instant, internal command position B has reached 100.080 and 100.140 mm respectively. This article defines following difference E=B−A, positive when the command is ahead.
| Ideal record | Actual position A | Command position B | Defined difference E | B−E |
|---|---|---|---|---|
| First attempt | 100.000 | 100.080 | 0.080 | 100.000 mm |
| Second attempt | 100.000 | 100.140 | 0.140 | 100.000 mm |
The B records differ by 0.060 mm although the contact position is identical. If a recording channel includes this command accumulation, using that difference directly to infer a workpiece change produces a false conclusion. B−E in the table is only an identity explaining compensation, not a program for writing a correction to a FANUC variable.
Why error must not be subtracted again from every record
If the recorded value has already been compensated, subtracting E again applies correction twice. If error and position come from different instants, B−E no longer describes one state. Before comparing records, align variable definitions, compensation state and sampling instant, rather than merely changing display units to micrometers.
This also explains an analytical trap: observing “more stable records after reducing approach speed” shows only that the data in that instance are affected by the motion process. It cannot alone establish a particular positioning repeatability or measurement accuracy.
Position compensation does not compensate contact-force deformation
The model treats the workpiece, fixture and transmission as rigid. Real contact involves elastic deformation, friction, backlash and differences in contact stiffness. Even after the difference between command and feedback is correctly removed, deformation caused by contact force may remain. Interpreting torque limit skip data and calibrating a measuring instrument are therefore separate tasks.
This article explains a function already configured according to the machine builder's design. It supplies no clamping force, torque threshold or trial-contact procedure. The central question is “which instant and coordinate layer does this position represent?” Related resources are in the OPTION section.