Hybrid analog-sampled adaptive active noise control

Abstract

Feedforward adaptive controllers are widely used for active noise control. For good broadband performance, the secondary path delay must be shorter than that of the primary path. In this work, a hybrid analog-sampled adaptive feedforward controller is developed to eliminate the additional delay associated with the sampling. An adaptive finite-impulse-response system is used for the sampled controller, and a state filtered adaptive linear combiner is used for the analog controller. Two approximations are used for the linear combiner. The first minimizes the added electrical noise by having individual state filters, all driven by the reference signal, while the second gives a superior bandwidth by driving each component filter with the output of the previous one. A sampled version of the normalized projection algorithm is developed for the analog controller. The different error paths used to filter the reference signal for the controllers’ adaptation are emphasized. Finally, the performance of the proposed controller is assessed. The results highlight the capabilities of the suggested controller and the different state filters, especially their ability to achieve significant attenuation levels close to the causality limit and when long delays of the reference signal are required.