Vibration detection and backlash suppression in machine tools

Mechanical backlash is a common trouble in all servo mechanisms that is created by mechanical defects i.e. wear or looseness. Backlash imposes nonlinearity on a linear control system and may be contributed to a limit cycle. Considering mechanical subsystem of a servo axis as an elastic two-mass system, modeling the backlash with a dead zone and choosing reduced order control system for the speed and current controller, the frequency bandwidth of vibration is estimated. The estimated frequency is supported by the experimental vibration analysis of six different machine tools. To eliminate unwanted vibration, a backlash detection and suppression approach is proposed in this paper. In the case of vibration, load sensor is disregarded and the regulator conducts the setting speed of interpolator to the drive. Then stability is satisfied with the cost of some tracking error. When stability is detected load sensor is considered and the regulator output converges to the speed provided by the position control. To examine the approach, dynamic model of a real servo axis is obtained and it is validated by experimental data. Then the proposed modified control system is simulated.