Multi-line TRL Calibration of Vector Network Analyzers: Classical versus Tensor Implementation

A comprehensive comparison between two multi-line thru-reflect-line calibration techniques is provided.

One of the methods is based on the classical combination of N measurements following the Gauss-Markov theorem.

The other one has been developed along this paper in a complete and detailed way using the recently published idea of considering each of the N measurements as part of a tensor.

The measurement of a calibration kit that covers from 500 MHz to 20.5 GHz has been carried out for both methods and the results discussed.

From the analysis of the results, it seems that the tensor approach could show some advantages in the (not uncommon) cases where a strictly white and Gaussian noise is not the only perturbation of the measurements. On the other hand, the decomposition of the tensor requires a higher computational effort, which would better be reduced.