A Statistical VNA Calibration Method Applying Multiple Reflection Standards

In this paper, a statistical calibration method for the two-port vector network analyzer (VNA) is proposed, where an arbitrary known two-port device and multiple reflections are used as standards. According to our recently proposed theory, the T-matrices of error boxes are constructed by measuring the known two-port standard and only three wave ratios are defined to represent the calibration coefficients. Then, based on the measurements of multiple reflection standards, the three calibration coefficients are initially solved through the ordinary linear regression analysis. After that, with the uncertainties from the connection repeatability introduced to the measurements of multiple reflection standards, the calibration coefficients are secondarily modified by using the weighted linear regression analysis. By this means, the calibration result is obtained under the impact of the connection repeatability, and then used for the error correction of S-parameter measurements. Finally, experiments are performed to verify the proposed method.