Statistical analysis of accuracy in noise figure measurements

There is a growing trend to integrate noise figure measurement capabilities in microwave instruments originally devoted to other general purpose measurements as Vector Network Analyzers or Spectrum Analyzers. This opens up new possibilities to take into account vector corrections (at least in the case of VNA), or even to consider other techniques (such as cold-source) for computing the noise figure, different from the classical Y-factor approach used by most noise figure meters. In order to evaluate the suitability of these new options it is important to correctly analyze the benefits in terms of final accuracy that they can provide. However, errors in noise figure measurements have very different sources and nature, which sometimes do not allow straight forward conclusions on measurement accuracy. In this paper, a consistent evaluation of systematic errors and measurement uncertainty associated to different methodologies is presented. For that, the recommendations of the IEC-ISO Guide to the Expression of Uncertainty in Measurement (GUM) are followed. Some representative examples are given to illustrate the interest of the analysis.