Influence of Different Configurations of Nonideal Calibration Standards on Vector Network Analyzer Performance

To perform system error correction of a microwave vector network analyzer (VNA), several calibration procedures are applied, several of which are derived from the well-known short-open-load-through (SOLT) calibration method where lumped one-port circuits like short circuits, open circuits, and low-reflective loads are used as calibration standards. These derived calibration algorithms only use three calibration standard connections-instead of six for SOLT; hence, the calibration time is considerably reduced. Various configurations of high-and low-reflective standards at test ports 1 and 2 of the VNA are possible. Using inaccurate calibration standard modeling, the VNA calibration is incorrect, and hence, inaccurate S-parameters of a device under test (DUT) are measured. In this paper, it is theoretically and experimentally shown how, due to non-ideal calibration standard modeling, the deviations of measured S-parameters of the DUT from their hardware-defined true values depend on the inaccurate calibration standard modeling, on the configurations of the calibration standards at the test ports, on the reflection values of the calibration standards, and on the S-parameter values of the DUTs themselves.