[OpenArm Troubleshooting] Base Mount Shift: Calibration Reference Mismatch (Intermediate)

OpenArm troubleshooting for base mount shift and stale calibration references: fixture drift symptoms, geometric checks, and safe recalibration decisions.

A subtle OpenArm precision problem appears after the robot base or mounting plate shifts slightly. The arm still powers up and can home, but task-space references, fixtures, or known poses no longer line up with what the old calibration assumed.

How are you diagnosing OpenArm cases where the base mount moved and the calibration reference no longer matches reality?

Please share how you detect mount shift, what checks distinguish a base-reference problem from encoder or tool issues, and how you decide whether partial recovery is safe or a full recalibration is required.

If you reply, include one exact symptom after the mount change and one exact geometric or fixture-based check that exposed it.

4 comments

  • Liam Chen

    Intermediate context: In our cell the first symptom was a clean home but a hard-stop fixture touch in +Y was off by about 1.8 mm. A 3-point fixture touch check exposed it in under 2 minutes, while joint repeatability stayed under 0.05 mm, so encoders were likely fine and the base frame was not.

  • Marta Ruiz

    Intermediate context: We separated base shift from tool drift by swapping end effectors and rerunning the same known-pose cube-corner check. TCP behavior stayed consistent but every waypoint was translated by almost the same vector (about 1.6 mm, mostly +X), which pointed to mount shift rather than tool or wrist calibration.

  • Noah Patel

    Intermediate context: Our rule: partial recovery only if a rigid-frame fit on 6 fiducials leaves less than 0.2 mm residual and less than 0.15 degrees yaw error. If not, do full recalibration plus collision-model revalidation. We restored trust with 30 dry cycles on two fixtures before releasing back to production.

  • Mason Cole

    Intermediate context follow-up: Follow-up question: after your frame correction, how much residual error remained on your worst fixture point (mm), and did that value stay stable after a full warmup shift?