On the automatic real-time detection of chatter onset in turning

Abstract. A low-cost device has been implemented in order to detect the onset of chatter in turning operations. This involves a microcontroller, which receives an acceleration signal from the workpiece being turned, converts it to digital form, processes it in real time, identifies specific features and monitors them also in real time. As a first step, Fast Fourier Transform of the acceleration signal for a reference tool path that involves no chatter identifies high energy peaks at particular frequencies, which depend on the eigen-frequencies of the dynamic system machine tool – cutting tool – workpiece. Then, at the onset of chatter a specific pattern of additional high energy peaks appears in the spectrum 0-2.5 kHz. Signal processing and comparison with the reference take less than 200 ms. Experiments on a CNC lathe proved that the approach is effective and transferable beyond longitudinal turning.