Sensorless Control of a Hybrid Stepper Motor

Electrical drives are widely used in today’s society. They can be found in bothhousehold products and in the industry. One application where electrical drivesare used is in robots for mowing lawns. In the studied robots the motors in theelectrical drives used for propulsion are Brush Less Direct Current motors, BLDCmotors.The BLDC-motor has its maximum torque at high speeds and thereforea gearbox is needed. The gearbox is space consuming, add costs and consists ofmechanical parts that wear during use. Of interest is therefore to investigate ifthere are other electrical drives which can be used for propulsion.A motor who has its maximum torque at low speeds is the Stepper motor, andtherefore it is of interest to investigate if a stepper motor could replace the BLDCmotor.A drawback with the stepper motor is that it always consumes maximumcurrent and therefore a current controller is beneficial. Together with currentcontrol, speed control is needed to make the robot run at desired speed. To beable to perform an accurate current and speed control feedback from the motor isneeded. Information about the rotor angle and velocity can be used for the speedcontrol and the load angle can be used for the current control since the current isproportional to the load torque.To estimate the rotor angle and velocity a model has been developed. Themodel is based on fundamental electrical and mechanical equations and neglectsthe current and position dependence of the inductance and flux linkage. To completethe model three motor parameters, the maximum detent torque Tdm, themaximum flux linkage m and the friction constant B was determined. Parameterdetermination was done by linear regression and by using an Extended KalmanFilter, EKF. The result of the parameter determination were Tdm = 0.2152 Nm, m = -0.002854 Vs/rad and B = 0.01186 Nms/rad.The model is used in an EKF to estimate the rotor angle and angular velocity.The result of the implemented EKF seems promising. When making the rotortake a step in velocity from 3.927 rad/s to 7.85 rad/s the EKF estimates the stateswith only a small bias: 0.02 rad for the angle, 0.3 rad/s for the velocity, 0.005 Afor phase a current and 0.0004 A for phase b current.To estimate the load angle the Sliding Discrete Fourier Transform is used. Theexpected relation between the load torque and load angle is sinusoidal. The loadangle is calculated from data where the external load is between 0-2.5 Nm. Inthat area the load angle shows the expected sinusoidal appearance and the loadangle is in the area between 0.1 and 0.45 rad. At 3 Nm the rotor stalls and it isshown that the load angle varies between 0 and 2 rad when the rotor is stalled