Efficient large-signal FET parameter extraction using harmonics

An accurate and truly nonlinear large-signal parameter extraction approach is presented, using not only DC bias and fundamental frequency but also higher harmonic responses. The harmonic balance method for nonlinear circuit simulation, adjoint analysis for nonlinear circuit sensitivity calculation, and state-of-the-art optimization methods are applied. Improvements to the convergence of the optimization process are discussed. An automatic weight assignment algorithm enhancing parameter extraction is given. Numerical results demonstrate that the method is both theoretically and computationally feasible, i.e. it can uniquely and efficiently determined the parameters of the nonlinear elements of the GaAs MESFET model under actual large-signal operating conditions. Numerical results also show that under multibias, multipower inputs, and multifrequency excitations, spectrum measurements can effectively reflect the nonlinearities of the model and improve model reliability when they are used in nonlinear large-signal-model parameter extraction.<
>