Electronic Thesis and Dissertation Repository

Degree

Master of Engineering Science

Program

Electrical and Computer Engineering

Supervisor

Lyndon J. Brown

Abstract

Brushless DC motor speed is controlled by synchronizing the stator coil current with rotor position in order to acquire an accurate alignment of stator rotating field with rotor permanent-magnet field for efficient transfer of energy. In order to accomplish this goal, a motor shaft is instantly tracked by using rotating rotor position sensors such as Hall effect sensors, optical encoders or resolvers etc. Adding sensors to detect rotor position affects the overall reliability and mechanical robustness of the system. Therefore, a whole new trend of replacing position sensors with sensorless rotor position estimation techniques have a promising demand.

Among the sensorless approaches, Back-EMF measurement and high frequency signal injection is the most common. Back-EMF is an electromotive force, directly proportional to the speed of rotor revolutions per second, the greater the speed motor acquires the greater the Back-EMF amplitude appears against the motion of rotation. However, the detected Back-EMF is zero at start-up and does not provide motor speed information at this instant. There-fore, Back-EMF based techniques are highly unfavourable for low speed application specially near zero. On the other hand, signal injection techniques are comparatively developed for low or near zero motor speed applications and they also can estimate the on-line motor parameters exploiting the identification theory on phase voltages and currents signals.

The signal injection approach requires expensive additional hardware to inject high frequency signal. Since, motors are typically driven with pulse width modulation techniques, high frequency signals are naturally already present which can be used to detect position. This thesis presents rotor position estimation by measuring the voltage and current signals and also proposes an equivalent permanent-magnet synchronous motor model by fitting thedata to a position dependent circuit model.

Share

COinS