A high-order model looking beyond the first-order harmonic superposition assumption

Abstract

In this paper, a hybrid high-order behavioural model is proposed to mimic the response of strongly nonlinear unmatched RF transistors. In this model, a high-order Multi-Harmonic Volterra (MHV) behavioural model is used to predict the DC and fundamental frequency components of the output signal, while higher harmonic components are predicted by the Poly-Harmonic Distortion (PHD) model. The added coefficients of the MHV model augment the first-order expansion (harmonic superposition) of the PHD model to improve the model accuracy where it is needed. The hybrid MHV-PHD model improves the DC drain current prediction by 5dB and fundamental frequency output-power by 2dB in terms of Normalized Mean Squared Error (NMSE), while improving overall time-domain prediction by 1dB.