Evaluation of Tuner-Based Noise-Parameter Extraction Methods for Very Low Noise Amplifiers

Abstract

This paper compares the performance of source-tuner noise-parameter extraction methods used to measure noise parameters of low-noise amplifiers that have very low (1 dB) noise figures. The methods discussed are known as the Cold method and the modified Y -factor method (or Hot-Cold method). The paper describes equations used in the extraction algorithms. In a Monte Carlo analysis by randomly adding various sources of uncertainties to ?measurements,? created with a computer simulation, performances of the noise parameter extraction methods are compared. It is shown that the iterative Cold method and the direct Cold method are the best at extracting Rn and ?opt noise parameters in terms of lowest standard deviation and close proximity of the extracted mean values to the true values. The simplified Cold method, used in a number of commercial systems, has largest systematic offsets in extracted noise parameters while being the quickest to perform. The modified Y-factor method is the slowest to perform due to additional time required for hot measurements. This method is marginally the most accurate to extract F min. These conclusions are also supported with measurement results. This study assembles in one place necessary theoretical background information to serve as a reference for those who are working in the field of noise parameter extraction using tuner-based methods.