An RF power amplifier modelled with a simple fifth order polynomial extracted from basic characterizations

In this paper, a model describing the nonlinear behavior of an RF power amplifier is proposed. This model is based on two polynoms extracted from measured AM/AM and AM/PM characteristics with a complex stimulus excitation, like WCDMA, richer than a CW signal. Instantaneous power variation of WCDMA produces peak amplitudes which generate different AM/AM and AM/PM responses depending on their time history. Indeed, the accuracy of existing model based on CW excitation are insufficient to describe such non linear behaviour in new wireless applications. Then the two performance degradation factors: amplitude (AM/AM), and phase (AM/PM) polynoms are used to predict ACPR. The same signal excitation for characterization and simulation is guaranteed by the connected-solutions bench (Agilent) permitting real comparisons. For the experiments, we use a STMicroelectronics PPA (pre power amplifier), with 18dB gain, +6dBm of maximum power output and less than -50dBc of ACPR. The bandwidth corresponds to 1920-1980MHz for WCDMA applications. Our model can be extended to other signals and can be easily implemented into a simulator used by RF designers. The proposed method is less expensive and less time consuming than a complete electrical and dynamic characterization.