A small-signal parameter-based metric for nonlinear models of electron devices

Many different nonlinear modeling approaches for electron devices have been proposed in the last few years, and quite often circuit designers suffer from the lack of reliable comparison criteria, on the basis of which the most suitable model for a specific application can be identified. Moreover, similar strategies are needed even by research groups, whose activity is devoted to the model identification and extraction, in order to quantify the degree of accuracy that will be achievable by the modelling approach adopted. In this paper, a new metric for the estimation of large-signal model accuracy is discussed, which is simply based on the comparison between deembedded measurements and model predictions of small-signal Y-parameters versus the bias voltages at the intrinsic device ports.