Comparative Study of Chatter Detection Methods for High-Speed Micromilling of Ti6Al4V

Abstract

Chatter detection is necessary to carry out stable machining. It is extremely critical in high speed micromilling (spindle speeds >100,000rpm) where the limited tool stiffness and small fluctuations in the cutting forces can lead to dynamic instability. Chatter can deteriorate the surface finish and is detrimental to the tool as well. Hence, chatter identification is important to obtain stable cutting parameters. This chatter free process parameters can be identified by carrying out experiments at different machining conditions. Chatter detection can be accomplished via in-process real time techniques based on force and displacement signals or off-line machined surface characterization. It may be noted that micromilling process uses a miniature end mill whose natural frequencies corresponding to higher modes can be as high as 10 KHz. Consequently, the sensor bandwidth should be sufficient to detect high chatter frequencies in high speed micromilling. The conventional way of detecting chatter via force sensor may not be applicable in high speed micromilling due to its limited bandwidth. The displacement sensors can also have issues with proper placement and the noise. The chatter surface identification is relatively easy via visual inspection of the surface topography but quantitatively specifying a surface parameter which captures chatter is challenging. To address the issues of chatter identification in high speed micromilling, three different experimental techniques have been studied for chatter characterization. Four different experimental conditions (two stable and two unstable processing conditions based on visual surface inspection of scanning electron micrographs) have been characterized by all three methods and a comparative assessment of different chatter detection techniques has been carried out in high speed micromilling of Ti6Al4V. It has been observed that force sensor is unable to identify the chatter and the displacement sensor is noisy but can be used to determine chatter under certain conditions. However, average power spectral density (APSD) obtained from the surface topography can clearly identify chatter. A critical average power spectral density has been identified for the determination of the onset of chatter.