Comparison of Motion Control Loops for Industrial Applications

High-performance AC and DC industrial servodrives use standardized motion control algorithms. The algorithms are based on common feedback sensing methods (digital position measurement via encoders or resolvers) and a common assumption that the electromagnetic torque dynamics are substantially faster than the motion control dynamics desired. The vast majority of these motion control algorithms close the motion control loops in one of two ways: 1) an average velocity loop is cascaded with a position loop, or 2) multiple state variable loops are closed in parallel. If the state variable form is properly configured, the command tracking properties are virtually independent of the disturbance rejection properties. However, this controller would require a command for acceleration which is frequently not available. In this case some modified form of cascaded loop controller topology is often used. In this paper the alternative methods for this case are explored and tuning guidelines developed based on both simulation and laboratory results. I. BACKGROUND High-performance AC and DC servodrives are widely used in motion control applications such as machine tools, packaging, printing, web handling, robots, textiles, and food processing. The motion control algorithms are based on the mechatronics assumption of nearly ideal electromagnetic torque control. This assumes ideal field orientation and current regulators of bandwidth considerably beyond the motion control bandwidths desired. Feedback devices, chiefly encoders and resolvers, are employed in these systems to sense motor position and to calculate the sample average motor velocity, albeit with significant average velocity resolution limitations. The vast majority of motion control algorithms employed in industrial applications are of two forms: 1) an average velocity loop is cascaded with a position loop, or 2) state variable loops (ProportionalIntegral-Differential,PID position loops) are closed in parallel. The state variable loops in parallel (PID position) configuration is known to completely separate command tracking tuning from disturbance rejection tuning [6,7,8,11,12]. However, such motion controllers require an acceleration command which is often not provided in industrial servodrive controllers. This causes the tuning to again become cross-coupled and interdependent. The cascaded loop topology is very commonly found in industrial servodrives and has a variety of adjustments to handle this (unwanted) cross-coupling of the tuning process. The cascaded average velocity loop is usually a Type I (integrating) loop which is cascaded with a proportional position loop. In that case, there are two types of average velocity loops that are commonly employed: ProportionalIntegral (PI) and Pseudo-Derivative Feedback (PDF) [1,3,4,5]. Both of these controllers offer different possibilities for handling the tuning cross-coupling. This paper focuses primarily on an alternative to PI and PDF which will be called here PI+ but is also sometimes called PDFF. PI+ will be shown to be a general controller within which PI and PDF are special cases. Further, PI+ will itself be shown to be a special case of the PID position loop. This paper will present simulation and laboratory experiments for the PI+ controller. The goals of this paper are to provide a quantitative analysis of PI, PDF, and PI+, and to compare the methods to the PID position controller for the case where no acceleration command is available. II. VELOCITY LOOP ALTERNATIVES In cascaded motion control systems position profile generators provide a position command and also an average velocity command to support so-called “velocity feedforward.” The average velocity loop output feeds a cascaded current regulated field oriented drive which creates electromagnetic torque in the motor. Position is fed back from a position sensor. Velocity is not measured. Currently most systems calculate sample average velocity from the difference of the two most recent positions divided by sample time, T, as shown in Fig. 1. PCMD PFB Current reg. field oriented AC servo drive VPCMD TQCMD