[OpenArm Troubleshooting] Repeatability Worse After Fast Recalibration (Intermediate)
OpenArm repeatability degradation after fast recalibration: warm-state validation loops, drift trend checks, and accept/reject criteria.
A frustrating OpenArm calibration problem appears when a quick recalibration seems to fix homing or startup checks but actual repeatability becomes worse. Motions still run, yet return-to-point accuracy and task consistency quietly degrade.
How are you diagnosing OpenArm repeatability that gets worse after a fast recalibration?
Please share how you compare before-versus-after accuracy, which checks reveal a calibration that is merely acceptable versus truly trustworthy, and how you decide when to redo the full procedure instead of moving on.
If you reply, include one exact repeatability symptom and one exact validation routine that proved the recalibration was not good enough.
Intermediate context: We got fooled once: fast recalibration cleared startup checks, but return-to-pin repeatability degraded from +/-0.08 mm to +/-0.31 mm. Operators only noticed when insertion yield dropped on the second hour.
Intermediate context: The revealing check was a 30-cycle A-B-A fixture loop logged at warm operating temperature. Cold test looked acceptable, but warm drift grew steadily, which explained why quick validation missed it.
Intermediate context: Now we accept recalibration only if 30-cycle sigma stays below 0.12 mm on two fixtures and no monotonic drift trend appears. If not, we stop tuning and rerun the full procedure end-to-end.
Intermediate context follow-up: Follow-up question: what minimum cycle count and warm-state duration do you require before accepting recalibration as production-ready?