Accuracy Improvement of Back-EMF-Based Position Sensorless Control of PMSM Drives Based on Virtual Current Sampling

The extended back electromotive force (BEMF) of permanent magnet synchronous motors (PMSMs) can be directly calculated according to the current derivative, thereby observing the rotor position. To avoid being limited by the short duration of zero-voltage vector, the twice current samplings can be completed at the beginning and middle of the control cycle. However, the current between two samples is usually approximated by a straight line. As the details of the current change are ignored, the accuracy of position observation decreases. In this letter, by introducing three virtual samplings, more information about the unknown current waveform between the two real samplings is obtained. Without adding new input variables, a set of equations for virtual current sampling and extended BEMF is constructed based only on the output voltage and sampling current. The comparative experiments prove that the virtual sampling current can improve the observation accuracy of rotor position and suppress harmonics of the feedback speed.