Systematic errors of noise parameter determination caused by imperfect source impedance measurement

We present a rigorous two-step analysis of systematic errors in the four-noise parameter determination of a two-port network using the cold-source technique. This analysis is based on an original model that accounts for residual errors in the source impedance measurement. The method employs two linear fractional transforms to decompose the errors into relevant factor sets affecting the parameters in different ways. Analyses performed for a low-noise pseudomorphic high-electron mobility transistor (PHEMT) and a microwave amplifier show that the noise parameters of the low-noise PHENIT are highly vulnerable to such errors.