Sensorless control of brushless DC motors

Abstract

This thesis presents novel sensorless control techniques applied to six-stepped controlled brushless permanent magnet motors (BPMM). Such motors are becoming appealing for many industrial applications and consumer appliances. This is due to their simple speed and torque control, compact size, high efficiency and reliability. It is essential that the rotor position is known in order for BPMMS to operate. The commutation instances take place at 300 electrical for torque optimisation. The position is detected by physical sensors such as Hall Effect sensors, encoders or resolvers. Sensors increase the motor size, cost, complexity, require maintenance in a confined space, and complicate machine commissioning. Further, the operational speed range is limited by the capability of the physical sensors used. The torque is also limited by the accuracy of the sensors' installation position. Various techniques have been developed to control the motor in sensorless mode to overcome the drawbacks of using physical sensors. The traditional approaches of sensorless control utilise the zero crossing point detection of back emf. The zero crossing point is shifted by various means and techniques to provide correct commutation instances. The zero crossing point method is adopted extensively for being inexpensive and for its simplicity. Such a method is unreliable under overloading, or may provide false commutation instances pulse width modulation electrical noise. This thesis proposes a novel approach called back emf mapping. The approach is developed to provide a high degree of reliability and robustness especially under overloading and electrical noise surrounding. The method computes the actual back emf value directly corresponding to commutation instances. The back emf mapping algorithm is suitable for the six-step control of BPMMs as it relies on the back emf of the floating phase. The method has a wide speed range and simple to implement. Simulation and experimental tests have been conducted to verify the approach. The results indicate that the back emf mapping commutation signals closely follow the physical sensors signals. Moreover, the back emf mapping has the advantage of directly identifying the rotor position commutation angle eliminating any additional circuitry or phase shift algorithms. This thesis identifies specific conditions where the back emf mapping approach may reduce the accuracy and robustness. Virtual back emf mapping projection is another approach that is investigated. The approach utilises the back emf mapping; however, it is assisted with a virtual built back emf slope for commutation. Extensive simulation is successfully conducted showing a high degree of accuracy within the method proposed. The performance is compared with the zero crossing point detection and examined against the sensor trigger signals. Under overloading conditions, the back emf mapping projection is found robust, whereas the zero crossing point method fails to sustain the motor operation. This thesis recommends investigations that could be conducted with the proposed sensorless control methods. These future directions include; integrating start-up algorithms, phase advance control, electrical noise effect on accuracy, using pulse width modulation for speed control and applying speed estimation in a closed loop speed control.

Date of Award	2012
Original language	English