Optimal vector-network-analyzer calibration algorithm developed by NIST staff.

NIST and PTB staff have completed StatistiCAL[TM], a software algorithm that combines a decade of experience in applying orthogonal distance regression and other iterative techniques to solving vector network analyzer (VNA) calibrations with an easy-to-use user interface. The new software features a robust algorithm capable of finding solutions even with poor initial estimates, greatly increasing its ease of use and broadening its applicability. The algorithm provides uncertainties of the estimated para-meters that account for both random and systematic errors in the VNA measurements, calibration model, and standards. The uncertainties in the solution are represented by a covariance matrix that relates errors in both the VNA calibration and measurements of the device under test. In addition, the algorithm determines coverage factors based on the different numbers of degrees of freedom associated with various parts of the solution.

Researchers not only have shown that the new algorithm outperforms NIST's popular MultiCal[R] software in the presence of measurement noise, but they have performed experiments demonstrating the accuracy of the uncertainty estimates generated by the algorithm. Software implementing this method can be downloaded at www.boulder.nist.gov/dylan/.

CONTACT: Dylan Williams, [phone]; [e-mail].