Measurement of loudspeaker parameters by inverse nonlinear control

Abstract

Nonlinear loudspeaker models have been developed to predict and explain the generation of nonlinear distortions in the reproduced sound at large amplitudes. These models require additional nonlinear parameters such as the force factor, stiffness of the mechanical suspension, and voice-coil inductance depending on the instantaneous excursion of the voice-coil. These parameters have to be measured dynamically since static approaches have shown to produce inaccuracies in describing the dynamic behavior of the loudspeaker due to hystereses and creep effects. The identification of the optimal model parameters can be accomplished by using an adaptive nonlinear system provided with the known input signal and an output signal measured on the speaker. An inverse nonlinear identification by using a nonlinear controller connected to the loudspeaker input has advantages over direct loudspeaker modeling in respect to robustness, convergence speed, and stability. Results of practical loudspeaker measurement are presented and discussed in respect to quality assessment of loudspeakers in the large-signal domain.