Multi-Branch Polynomial Model with Embedded Average Power Dependency for 3G RF Power Amplifiers

In this paper, a study of the of multi-branch polynomial model performance for RF power amplifiers operating over a wide input power range is presented. For this purpose, a 100-watts peak power amplifier designed for 3G applications is characterized over a 12 dB average input power range. A multi-branch polynomial model is derived at each average power level. The models´ accuracy in predicting the PA´s output spectrum under average input power mismatch is evaluated. Then, the model´s parameters variation versus the average input power is considered. It is shown that there is no straightforward analytical function that might take into account for the parameters variation with respect to the average input power level. Accordingly, a complexity reduced multi-branch model with embedded average power dependency is proposed. The model coefficients are stored into memory banks, and the adequate coefficients are loaded depending on the average input power level.