Feedforward PLLs for Motor Position Estimation Using Embedded Magnetic Encoder

The encoder and resolver are commonly used rotor angle position sensors for controlling permanent magnet synchronous motors (PMSMs). However, they require significant axial space. This issue can be avoided with an embedded magnetic encoder (EME), which offers precise rotor angle estimation using a few linear Hall sensors in the motor cavity while occupying less space. Conventional Type-2 [i.e., proportional-integral loop filter (LF)] synchronous reference frame (SRF) phase-locked loop (PLL) can estimate phase and frequency from the measured Hall signals. However, the ramp frequency input during motor speed acceleration/deceleration causes dc phase error in the estimated angle. Type-3 (i.e., proportional-integral-acceleration LF) SRF-PLL can mitigate this issue but faces instability problems. In this article, two small-signal equivalent quasi-Type-1 (i.e., proportional only LF) SRF-PLLs are proposed with linear phase detector (PD)-based forward compensation (FC) terms. The FC terms can be implemented either by using the phase estimation error or through the estimated frequency. As the loop filters are of low order, fast convergence can be achieved by the proposed PLL without sacrificing the stability margin, even though having the same closed-loop system order as that of the conventional Type-3 counterpart. Analytical development of the proposed PLLs is validated with an experimental EME system on a PMSM using various test conditions. Comparative results show that the proposed technique can track the phase faster with zero steady-state error and lower peak overshoot compared with similar other methods from the literature, despite all the methods being tuned using the same phase margin to ensure a fair comparison.