Resonant circuit model evaluation using reflected S-parameter data

Nonlinear regression is used to fit S-parameter resonance data to a full-circuit model that includes coupling factors and self impedances. This model fits the data better than the simpler Q-circle model that can be derived from the full-circuit model, but a systematic pattern in the residuals persists. This pattern indicates a discrepancy between the full-circuit model and the observed data. By looking at parameter estimates calculated from subsets of the original data, it is demonstrated that the cause of this discrepancy also could introduce significant errors in the model's estimated parameter values.<>