Sensorless Accuracy Improvement Strategy for Electrolytic Capacitorless Drives Based on Frequency Adaptive Notch Filter

The capacitance value of the DC-link in the electrolytic capacitorless drives is small, which leads to the generation of rich harmonics in the DC bus voltage and motor phase current. In the medium to high speed region of motor operation, sensorless position estimation methods based on fundamental mathematical models are usually adopted. However, harmonics in the electrolytic capacitorless drives will lead to increased position estimation errors in sensorless control. A harmonic suppression method based on frequency adaptive notch filter (FANF) is proposed to reduce estimation errors of the sensorless. Firstly, a sliding mode observer (SMO) designed to estimate the extended back electromotive force (BEMF) is based on the proposed novel fast sliding mode reaching law. Secondly, the order of various harmonics in the electrolytic capacitorless drives is derived according to the relationship between extended BEMF and DC bus voltage. Finally, FANF with unchanged phase and amplitude at the center frequency is used in the front end of the phase-locked loop (PLL) to preprocess the estimated BEMF, with the aim of eliminating harmonics. The method proposed in this article is validated through an experimental platform based on the electrolytic capacitorless drives.