Accuracy of s-parameter measurements under discussion

In all the years I’ve been dealing with Vector Network Analyzer I used a “Calibration Kit” to remove system errors. A “Verification Kit” was not necessary and as long cables were not moving a lot measurement results should be fine.

But for s-parameter measurements there are also “Random Errors” involved which are of great influence of total accuracy. Parameters are for the VNA: non-linearity, drift, noise floor and trace noise. Also Isolation (cross-talk) could be of any influence but practically this is hardly none.

Components with major influence on total accuracy are Test port cable stability and Connection repeatability. But now the question is, what’s the influence of a cable, a connector on the of accuracy of a s-parameter measurement? The challenge is how to determine these “Random Errors” to get full insight in measurement accuracy.

To find out I used Maury’s “Insight s-parameter calibration and measurement software”. With this software it’s simple to determine the influence on accuracy of every component in my measurement setup. With the help of the “Insight software” I characterized the instrument on “Noise” and “Drift” of the VNA. Also I characterized the cables I used and connector. More information about the “Insight Calibration and Measurement” Software van be found via this link.

As this software is compatible with most VNA’s in the industry I received from Maury’s technology partner “Copper Mountain” a S8085 – USB VNA with a bandwidth of 9KHz to 8.5GHz and a T911 Calibration kit. However, I’m really hoping I get an opportunity to compare results with VNA’s of Maury’s other technology partners: Keysight Technologies and Rohde&Schwarz. Furthermore I’ve done tests only to 8.5GHz wo what will be the results on much higher frequencies. For more info about the VNA I used for these tests.


I compared 3 different cables over the complete bandwidth of the VNA and done a simple VSWR Measurement. I used a standard SMA-, Maury StabilityPlus- and a Maury StabilityBench Cable.

As DUT I used the “Load” of the Calibration kit. The image below shows the results of the VSWR measured on the different cables.


Although results are quite similar it’s clearly visible the uncertainty (shown as a shadow in the graph) is different. to compare results the StabilityPlus cable is used as reference, as this cable seems to have the best performance. The graph here below gives a better view on the total error on a s-paremeter measurement with three different types of cable.


So let’s take a look at the different components used for this measurement. First of course the VNA I used. On the graph here below you can see there is some influence on accuracy but it doesn’t explain deviation.


So it must be the connector and/or the cable which causes the largest part of inaccuracy. So let’s check out.



So checking out the results we can conclude, Choosing the right cable of doing the measurement is most important! If the importance of this choice is neglected it may results in wrong assumptions. Result will not be as repeatable and accurate as expected, even be useless.

Secondly, only a “System Calibration” is not enough. “Random errors” should be determined for having full insight in total accuracy of your s-parameter measurements. If the accuracy of a measurement is not known also the value of a measurement is not known.

I hope you enjoyed this “Insight” of uncertainties on s-parameter measurements. Please share your comments or questions.


Best regards,
Dirk Faber, Hitech BV