Accuracy of Microwave Transistor $f_{\rm T}$ and $f_{\rm MAX}$ Extractions

Abstract

We present a complete methodology to evaluate the accuracy of the microwave transistor figures-of-merit f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> (current gain cutoff frequency) and f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MAX</sub> (maximum oscillation frequency). These figures-of-merit are usually extracted from calibrated S-parameter measurements affected by residual calibration and measurement uncertainties. Thus, the uncertainties associated with f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> and f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MAX</sub> can be evaluated only after an accurate computation of the S-parameters uncertainties. This is done with the aid of two recently released software tools. In the uncertainty propagation, the standard de-embedding techniques are assumed to be error free, but still contribute to the final uncertainty budget via their measurement uncertainty. We also present an analysis of how different interpolation/extrapolation methodologies affect uncertainty. In addition, an overview of the possible causes of errors and suggestions on how to avoid them are given. With the continued rise of reported f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> /f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MAX</sub> values, this work has become necessary in order to evaluate the accuracy of these figures-of-merit both by adding confidence intervals to their values and by identifying possible extraction errors.