Predistortion of Charge Trapping Memory Effects in GaN based RF Power Amplifiers with Artificial Neural Networks

Energy efficiency and bandwidth of RF power amplifiers (PAs) is always a challenge for future radios. Static and dynamic nonlinear effects are responsible for degrading before mentioned parameters. Static effects can be characterized by measuring the PA output with a defined input signal. Fast changing dynamic nonlinearities like thermal memory and charge trapping effects with different time constants are more difficult to determine but have a strong influence on the transfer characteristic of the PA and require a dedicated and costly feedback path for fast adaption of the digital predistortion (DPD). This work shows a method to measure the trapping condition in the transistor to enhance the predistortion performance with artificial neural networks (ANN) and reduce complexity of the feedback path.