Chatter detection by periodic sampling of in situ milling signals with spindle speed selection: Experimental validation

This study experimentally validates a chatter detection method for milling based on periodic sampling of in situ time domain signals. The sampling frequency is scanned over a preselected range, and a stability metric is computed from the sampled signal values at each frequency. For stable cuts, the sampled point repeats each tooth passage and a local minimum in the metric occurs at the tooth passing frequency; for unstable cuts, the local minimum occurs at or near the chatter frequency. The locations of the local minima, Fast Fourier Transform (FFT) analysis, and surface roughness measurements are used to confirm cut stability. The method is validated using displacement, acceleration, and sound signals across three tool-workpiece combinations. For spindle speed selection, the chatter frequency identified from an unstable cut and the number of cutting teeth are used to calculate a new spindle speed, which achieves a stable cut with all other conditions unchanged. Finally, a low-latency feasibility study toward future closed loop chatter detection and spindle speed selection is presented, using a hybrid decision framework in which the time domain stability metric detects instability and FFT analysis identifies the chatter frequency for the spindle speed update.