SED navigation bar go to SED home page go to SED publications page go to NIST home page SED Home Page SED Contacts SED Projects SED Products and Publications Search SED Pages

contents     previous     next

3.2.4 Vector Network Analyzer Performance

Jolene Splett, Hari Iyer
Statistical Engineering Division, ITL

Robert M. Judish
Radio Frequency Technology Division, EEEL

Vector automatic network analyzers (VANAs) are widely used in the microwave industry to characterize electromagnetic properties of microwave components. VANAs are capable of generating complex-valued measurements (reflection coefficients) at hundreds of frequencies, so it is difficult to verify the precision and accuracy of VANAs used in industry and provide traceability to NIST.

Traditional approaches for analyzing these data examine the magnitudes and phases of the reflection coefficient separately. Our initial work has been focused on updating current VANA uncertainties across broad frequency ranges. Existing data was used to build a linear mixed model with variance and covariance components and corresponding random effects.

The existing database of past calibrations is known to contain discrepant measurements, so we are also working to develop procedures for identifying outlying measurements and/or calibrations. Outliers can occur in one or more of the random effects so new statistical and graphical tools appear to be necessary to correctly identify them. Methods are also needed to combine the uncertainty information across the entire frequency range. New techniques need to be developed for computing robust uncertainty estimates for key characteristics of VANAs that will replace the traditional uncertainty estimates.

Our ultimate goal is to extend these methods so that the complex-valued reflection coefficient is the basis of statistical analysis. In fact, each test device has a characteristic ``signature'' curve when the reflection coefficient is plotted in the complex plane as a function of frequency. We intend to develop methods for robust estimation of these signature curves along with corresponding uncertainties. This will lead to improved characterizations of test devices and improved uncertainty estimates.

Plots displaying signature curves in the complex plane for two similar, but distinct devices are shown respectively in the top and bottom graphs.


Figure 10: Measured values of reflection coefficient for increasing frequencies from 2 to 18 GHz for two ``similar'' devices. Each graph shows three repeated reflection coefficient signatures corresponding to three reconnections of the device.

contents     previous     next

Date created: 7/20/2001
Last updated: 7/20/2001
Please email comments on this WWW page to