A Robust Extraction Technique for Second Order PHD Based Behavioral Models

Measurement based black box behavioral models for non-linear devices are increasingly popular nowadays. One important candidate is the X-parameter model, which is based on polyharmonic distortion modeling (PHD). The X-parameter model only utilizes linear terms of the PHD based approximation, hence, devices operating in a highly non-linear regime cannot be modeled accurately. However, quadratic PHD (QPHD) terms can be utilized to increase the model prediction accuracy, but they are generally hard to extract. In this paper a technique is presented, which allows to extract QPHD terms by utilizing load-pull measurements. With this technique a model with a significantly reduced model error was generated. This was demonstrated by comparing the proposed model with a state of the art X-parameter model.