Reducing the effects of random errors in six-port network analysers

The performance of six-port network analyzer is degraded by the presence of noise and random error mechanisms. Experimental results demonstrate that the effects of random errors can be significantly reduced (10*) by using a nonlinear Kalman filter. An attractive property of the nonlinear Kalman filter is the ability to deduce minimum variance estimates for a given parameter(s) from a collection of noisy observations. In the six-port environment this property is exploited, reducing the degrading effects of random errors that undermine the integrity of calibration and measurement.<>