The power quality problem, especially regarding harmonics contamination, has dramatically affected the overall power system stability. In response to this, using an Active Power Filter (APF) is considered one of the compelling methods to overcome harmonics issues. This paper presents the implementation of Shunt APF with an improved adaptive notch filter known as Extended Fryze Adaptive Notch Filter (EFANF) for fundamental signal extraction. The adaptive notch filter has improved the utilization from a single-phase to three-phase application for direct fundamental signal extraction and is designed to cater DC link voltage regulation controllers based on the power loss equation by applying Fryze current control power. This extraction algorithm inherits simple design construction and frequency tracking, eliminating PLL reliance on synchronization. The proposed algorithm also improved the design by eliminating the needs of low pass filter as others time domain algorithms. The algorithm's effectiveness in operation for the Shunt APF is validated through simulation using MATLAB/Simulink and experiment work by integrating the algorithm with DSPACE RS1104. Based on both evaluations, the results obtained show a satisfactory and reasonable agreement in mitigating the harmonics for multi-load conditions. Simulation and experimentally proven harmonics mitigation managed to reduce under 5% following the IEE standard, and the algorithm function within expectation for both steady and transient state conditions. In comparison between the simulation and experimental results, both results show almost similar results in terms of waveform for source voltage, source current, load current, and filter current in both steady state and transient state. In terms of THD values, both simulations and experimental results recorded that the THD values for both conditions are below 5%, where the range is 1.45% to 2.46% for simulation and 2.88% to 3.65% for experimental results. Furthermore, the DC link also tended to be maintained by the algorithm.