Comparative Embedded Frequency Estimation Based on Digital PLL and Extended Kalman Filtering under Standardized Grid Conditions

This article presents an experimental comparison between a digital phase-locked loop (PLL) and an extended Kalman filter (EKF) implemented as frequency estimators on identical embedded hardware platforms. Both algorithms were deployed on STM32-based microcontrollers and evaluated in real time using a calibrated electrical power standard as reference source. The experimental campaign followed test conditions inspired by IEC/IEEE 60255-118-1, including off-nominal steady-state frequencies, voltage amplitude variations, controlled harmonic distortion, and dynamic frequency steps. A systematic tuning procedure was adopted to ensure comparable steady-state noise levels for the two estimators, enabling a fair assessment of their metrological and implementation-related characteristics. The comparison jointly considers steady-state accuracy, statistical dispersion, dynamic response, and computational cost on the embedded target. The presented results provide quantitative insight into the practical tradeoffs associated with PLL- and EKF-based frequency estimation under realistic grid conditions.