Characterization of polynomial device model for distortion contribution analysis

Abstract

Nonlinear distortion in high-frequency electronic circuit of modern telecommunication systems poses significant challenges to meet the requirements of the performance of the circuits and systems. In this thesis multi-dimensional polynomial modeling is used to characterize the effect of nonlinear distortion within intrinsic device (intrinsic MOSFET model) for distortion contribution analysis. A polynomial based approach is used to extract nonlinear input and output capacitances and drain current of intrinsic device from large signal Harmonic Balance simulation data to represent device nonlinear behavior.

In this methodology, intrinsic model is replaced with polynomial device model including I-V and Q-V sources. These sources as functions of terminal voltages replicates the nonlinear behavior of nonlinear input and output capacitances and 2-dimensional drain current. The proposed method employs various polynomial fitting techniques such as spectral convolution, least squares optimization for extraction of the nonlinear coefficients dealing with difficult ill-conditioned matrices because of correlated control voltages.

Although the polynomial model accurately fits the nonlinear device characteristics, the coefficients are frequency independent but remain dependent on bias voltage and signal amplitude, so that the model must be refitted for every bias condition. The proposed approach is validated by using a polynomially modelled stacked power amplifier, which shows that nonlinear sources of the nonlinear device can be captured with this technique. The outcomes of this research can be used to improve the nonlinear distortion analysis techniques and integrate it into commercial circuit simulators.