End-launcher repeatability in broadband methods for characterization of the propagation constant of transmission lines using two-port measurements.

Abstract

This work presents an analysis of the influence of connector repeatability in three different methods for estimating the propagation constant of transmission lines from two-port measurements. For this purpose, the repeatability of 16 transitions using 1.85 mm coaxial-to-microstrip end-launcher connectors has been tested. It has shown that using the same pair of connectors instead of the whole set significantly reduces the standard deviation of the transition S-parameters that affects the final estimation of the propagation constant, and especially the attenuation constant. In addition, the hypothesis that the measured data have a normal probability distribution has been validated by performing an Anderson-Darling test on the estimated S-parameters of the transition. The obtained standard deviation has been included in a sensitivity analysis, generating S-parameters from normal distribution and performing a Monte Carlo simulation. The objective is to study the standard deviation of the propagation constant obtained using the proposed methods when there are errors related to connector repeatability. In this case, unlike random errors of the analyzer, it has been found that all the compared strategies for the estimation of the propagation constant (traces, eigenvalues, and determinants) work in the same way concerning launcher repeatability errors. Furthermore, it has been seen that the propagation constant obtained also follows a normal distribution. Finally, to validate the presented theory, methods have been applied to several measurements of two lines in the 0.01- to 67-GHz frequency range, using the same kit and different combinations of different connectors. Results show that higher accuracy is obtained when using the same pair of connectors, considerably reducing attenuation constant ripple, which assesses the suitability of the proposed error analysis.