Sliding Discrete Fourier Transform (SDFT) is very efficient regarding computational load and it possesses a very fast phase angle detection with excellent harmonic rejection at nominal frequency. However, at off-nominal frequency, SDFT generates errors in both magnitude and phase angle due to spectral leakage. This paper introduces a workaround for Fourier Transform to handle this disability under off-nominal frequency while avoiding variable-rate sampling. Sliding Fourier Transform (SFT) is used as a phase detector for a phase-locked loop whose output frequency is used to drive the SFT. The paper revisits the mathematics of Fourier Transform (FT) in a three-phase setting via a time-domain approach to show a newly proposed filtering technique for the double-frequency oscillation just by summing the FT sine/cosine filter outputs of the three individual phases. Also, the analysis aims to determine and correct the phase and magnitude errors under offnominal frequency operation. The proposed technique (SFT-PLL) is tested in real time on dSPACE DS1202 DSP using voltage vectors that are pregenerated to simulate the most adverse grid conditions. The testing scenarios compare the performance of the SFT-PLL with that of the Decoupled Stationary Reference Frame PLL (dαβPLL). The results prove that SFT-PLL is superior to dαβPLL.
Three-Phase PWM-Switched Autotransformer Voltage-Sag Compensator Based on Phase Angle Analysis
