Dynamic Load Angle Estimation Through an Angular Domain GDFT Filter for Optimal Energetic BLDC Control

Abstract

These days, BLDC motors are increasingly used to drive speed-controlled applications (e.g., drones, air compressors, etc.) thanks to their high power density, good torque inertia ratio, and their simple square wave commutation principle. However, with growing attention towards energy usage, a more efficient sinusoidal current wave is favored over the classical square wave profile. Nevertheless, changing the waveform to absent the silence phase implicates that simple control principles (e.g., hallsensors or back-EMF tracing) are no longer functional, and a more advanced sensorless solution is required. Earlier, an innovative sensorless load-angle-based controller was suggested with the potential to overcome all restrictions concerning stateof-the-art sensorless control. One of the proposed estimator's fundamental building blocks is a sliding discrete Fourier filter (SDFT), requiring both a signal with a fixed base frequency and a full signal period of samples before updating completely. Both requirements severely limit the application field. This paper presents an enhanced load angle estimator that can estimate the load angle 66% faster on average and without quality deterioration, regardless of the rotor speed. Both speed and accuracy of the estimated load angle are validated based on BLDC-measurements, confirming the potential and revealing possible further estimation lag decreases if a limited estimation quality regression is allowed.