Evaluation of a servo settling algorithm

The aim of this work is to discuss methods of friction identification and provide experimental evaluation of a novel control algorithm that enhances settling after point-to-point motion. This algorithm is called the Nonlinear Integral Action Settling Algorithm or NIASA. As the name suggests, the integral gain is nonlinear, and is based upon a Dahl friction model. The settling resulting from PID + NIASA control is nearly exponential, and governed by a time constant that is specified in the control design. As the NIASA algorithm requires, friction parameters must be identified for the servo under test. Two methods of friction identification (Step Tests and Identification Profile) are contrasted and found to provide comparable results, although the latter can provide advantages. The identified friction parameters are in turn used to perform four sets of control experiments; two PID controllers (standard factory tuning and high performance PID with acceleration feedforward) are tested both with and without NIASA compensation. In the case study with a factory tuned PID controller, servo settling times to within ±3–100 nm, are reduced by between 80.5% and 87.4% when NIASA compensation is added. When the NIASA compensator is added to the high performance PID controller, servo settling time is still reduced by between 50.5% and 73.0%. Although the NIASA compensator was designed to increase settling performance for relatively large point-to-point motions, similar positive results are achieved when the method is applied to smaller step motions that do not leave the pre-rolling friction regime. Frequency domain analyses demonstrated the nonlinear loop-gain of the plant, with a clear distinction between the rolling and pre-rolling friction cases. As expected, the nonlinear loop gain was found to lower the bandwidth for smaller motions. Adding NIASA control was observed to increase the bandwidth for small motions by a factor of 3–6, while having little effect for large motions.

► A novel nonlinear settling control algorithm was investigated.
► The PID control algorithm uses a Dahl friction model for the integral term.
► Two methods of friction model parameter estimation were presented.
► Large step and small steps (near the prerolling friction regime) were studied.
► Settling time was reduced by 51–87% by the algorithm.