An Automated Methodology for the Measurement and Non-Linear Model Parameter Extraction of GaAs MESFETs

Abstract

This paper presents an automated methodology for the measurement and non-linear modeling of GaAs MESFETS. The FET model extraction is based on the optimization of bias-dependent representations of DC-IV and linear equivalent circuit element values to small signal measurements of DC-IV and s-parameter data acquired over a wide range of operating bias, Vgs and Vds. The data acquisition is fully automated through the use of ANACAT(TM) and is collected and managed in an open ASCII format. The single set of measurement data serves as the input to the extraction of several popular bias-dependent FET models using Xtract(TM), a fully interactive FET model extraction tool developed by EEsof. The Xtract model file output may be used directly by Touchstone(TM), Libra(TM), or Microwave Spice(TM) to simulate the large-signal performance of the device in addition to serving as a master linear model for replicating linear small signal s-parameters at any operating point within the device's useful range. As an example, the paper presents the model extraction for a half-micron TriQuint GaAs MESFET. The example discusses the fully automated FET measurement and the subsequent extraction of three bias-dependent models. Both linear and non-linear model simulations are compared with actual measurements of the same device.