Accuracy improvements in two-port noise parameter extraction method

A simple method for the determination of the two-port noise parameters is presented. It is a weighted least squares regression, where the weights are a function of the measurement uncertainties. This method fits the best minimum noise figure and noise equivalent resistance, which are the most sensitive noise parameters. A more rapid convergence and a better accuracy are demonstrated with this new method by using a reduced number of reflection coefficient states than with the method of M. Mitama and H. Katoh (1979). In order to prove that it is possible to reduce the error effect on the noise parameter extraction by choosing the right fitting algorithm, simulation software that allows the test and comparison of the different methods was developed. Experimental results are presented for a 0.25- mu m HEMT (high-electron-mobility transistor).<>