The Research On Key Technologies For Sensorless Control Of Brushless DC Motors

Permanent magnet brushless DC motors (BLDCM) have been widely used due to their simple structure, high efficiency, ease of control, and low maintenance. When using three-phase six-step 120 degree commutation for tranditional BLDCM, position sensors are necessary to sense rotor position information for proper commutation. However, the drawbacks of rotor position sensors in the cost, volume and reliability limit the application of BLDCM. The research on sensorless BLDCM drives has important theoretical significance and practical value. In this thesis, a systematic and in-depth theoretical analysis, simulation and experimental study on several key technologies such as rotor position identification methods, PWM control strategies, motor start-up control algorithms, estimation error compensation, and torque ripple minimization schemes are performed for sensorless BLDCM drives.The method based on back electromotive force (EMF) of BLDCM is one of the major approaches for rotor position estimation because of the advantages of simple and reliable detection. Based on the analysis of traditional zero-crossing of back EMF detection methods requiring a virtual mid-point of motor windings, an improved phase back EMF sensing rotor position method is proposed. The unexcited phase back EMF signal is indirectly obtained from line voltage calculation instead of traditional hardware sensing method. This method is simple and reliable to implement without the need to sense or reconstruct the motor mid-point and depth filter circuits.A novel scheme of zero-crossing of line back EMF sensing rotor position is presented based on the analysis of back EMF waveforms of BLDCM. By qualitative waveform analysis and quantitative calculation, the conclusion that the zero-crossing of line back EMF is actual commutation point is detailedly proved. For arbitrary flat top width of the trapezoidal wave or sine wave, as long as three phase back EMFs are symmetrical, the proposed method is applied to the rotor position estimation. On this basis, in order to reduce dependence on motor parameters, a simplified rotor position emstimation method of line back EMF is also proposed. This method is only dependent on the stator resistance of motor parameters and has some theoretical and engineering application value because of unrequired phase delay by 30°or the last commutation information.Eight kinds of PWM strategies for BLDCM are analyzed and compared. Through the analysis to the diode freewheeling currents in the unexcited phase on non-commutation period, PWM strategies are obtained corresponding to above rotor position identification methods.How to control motor start-up is another difficult problem for sensorless BLDCM drives. Based on principle of stator core saturation, a new start-up control method is proposed. This method employs 12 voltage vectors injecting the stator sides, and the changes of the stator inductance are obtained by detecting the peak stator current of each phase. Therefore, the 30°range of rotor initial position is determined. And then the rotor position is accurately located in the next commutation position through the injection of particular voltage vector. On this basis, motor starting can be completed through the principle of raising voltage raising frequency. Once the zero-crossing of back EMF is detected, operating mode can be switched to sensorless control mode. Studies have shown that this method make the motor start reliably and is easy to implemente practically. The impact of the motor parameters variation, voltage and current sampling error, phase delay from filter circuits on rotor position identification error is detailedly analyzed.The method of fitting a linear function for open loop phase delay compensation based on a least square method is proposed, and a close loop rotor position error correction method is also presented according to the symmetry of back EMF waveforms. Simulation and experimental results show that above methods have practical value. Torque ripple is an inherent shortcoming of BLDCM drives. In sensorless control, torque ripple may be further increased due to rotor position identification errors. The impact of the phase back EMF flat top width, rotor position estimation error, non-commutation phase current distortion on torque ripple is detailedly analyzed. Afterward, a stator current predictive control method is proposed to suppress torque ripple of BLDCM.A system experimental test bed based on a dsPIC30F6010 DSP processor is established for proposed sensorless control schemes of BLDCM. Sensorless control experimental results are compared with test results of position sensor control under the same conditions. Experimental results reveal the effectiveness of the proposed schemes.