Optimal Load Angle Learning Algorithm for Sensorless Closed-Loop Stepping Motor Control

Abstract

Stepping motors are well suited for open-loop positioning tasks at low-power. The rotor position of the machine is simply controlled by the user. Every time the user sends a next pulse, the stepping motor driver excites the correct stator phases to rotate the rotor over a pre-defined discrete angular position. In this way, counting the step command pulses enables open-loop positioning. However, when the motor is overloaded or stuck, the relation between the expected rotor position based on the number of step command pulses and the actual rotor position is lost. To avoid this, the bulk of the widely used full-step open-loop stepping motor drive algorithms are driven at maximum current. This non-optimal way of control leads to low efficiency. To use stepping motors more optimally, closed-loop control is needed. A previously described sensorless load angle estimation algorithm, solely based on voltage and current measurements, is used to provide sensorless feedback. A closed-loop load angle controller adapts the current level to reach the setpoint load angle to obtain the optimal torque/current ratio. The difficulty is that the optimal load angle depends on the mechanical dynamics. To avoid the requirement of knowledge of the mechanical parameters, a practical learning algorithm to determine the optimal load angle is presented in this paper. Measurements validate the proposed approach.