Characterization and modeling of random vector network analyzer measurement errors

We present a technique for characterizing and modeling random vector network analyzer measurement errors. These errors manifest themselves as random changes or drift of vector network analyzer calibration coefficients. We model a change of calibration coefficients as a set of small electrically lumped perturbations occurring at different locations in the vector network analyzer. We then use measurements of a highly reflective offset termination to determine electrical parameters and locations of these perturbations. Our technique is the first step toward developing a statistical description of random vector network analyzer measurement errors and accounting for them in the measurement uncertainty analysis.