Beginning with version 2020, the N1500A computes the uncertainty of the material parameters extraction for Transmission Line and Free Space Method - Opt. 001.
Uncertainty is an integral part of measurements in this imperfect world, but not an undefineable part. The uncertainty of probe measurements results can be estimated with sensitivity numbers (or data) . Sensitivity numbers represent the slope of the model that relates permittivity to reflection coefficient. Recall that the network analyzer measures the reflection coefficient of the MUT, not its permittivity. The software, through a model, converts the reflection coefficient to permittivity.
Since the actual measurement is a reflection coefficient measurement, the question of measurement uncertainty is this: what effect does a reflection coefficient measurement error have on permittivity?
Sensitivity numbers can show a relative, qualitative indication of uncertainty. For example, if in one instance the sensitivity number is 50 and in a second instance it is 25, the measurement uncertainties of the first instance are twice (50/25 = 2) those of the second.
More precisely, error sources can be divided into two categories: network analyzer error sources and dielectric error sources. Examples of network analyzer error sources are noise (about 0.0006) and the fixed load/probe directivity contribution (0.05 to 0.15, for a LOAD/AIR/SHORT calibration, depending on frequency). Examples of dielectric error sources are probe model accuracy (3% to 5%) and uncertainty due to the accuracy of the permittivity characterization of calibration or reference standards.
The sensitivity numbers are useful in determining the measurement error contributions from both categories.