Research On Large Signal Frequency Domain Behavioral Modeling For RF Power Amplifier

Microwave power amplifier has become the dominant device as the nonlinear distortion generated by amplifier plays a key role in influencing the efficiency of transmitter/receiver. This dominance has resulted in more requirements for RF power components, for example, higher power and efficiency of transmitter/receiver and smaller in size. However, Empirical cut-and-try methods in wireless infrastructure power devices design are time-consuming and require an experienced touch. As a result, experts turn to or even become fascinated with one rapidly developing technology, Computer-aid Design.Fundamental to the Computer-aid Design for nonlinear devices is building adequate nonlinear large-signal RF models. If these models exist, nonlinear devices design and simulation can save numerous iterations and simplify design.The introduction of the notion of large signal scatter functions for modeling nonlinear RF devices exploits a novel way for researchers to solve the existing problems of modeling from new perspectives.The thesis begins with an introduction of present methods of modeling RF power component as well as their advantages and limitations, and then proposes a new approach to modeling RF power component based on SVMs by identifying large signal scatter functions and accurately characterizing the nonlinear properties of RF power component. Findings prove the effectiveness of the approach and the characteristic of requiring fewer data compared with the existing ANN-based method and the resulted advantage of less test time.Based on nonlinear large-signal network analyzer measurement data, we describe a novel method to analyze RF modeling and testify its effectiveness with various experiments using AM-AM, AM-PM, harmonic distortion analysis and so on.We design a RF power amplifier behavior modeling software platform to realize two purposes: one is to identify scattering functions and then to acquire AM-AM, AM-PM, harmonic distortion characters and time domain current and voltage waveforms; the other is to calculate PHD model by using measurement data and to analyze nonlinear characters of power amplifier such as AM-AM, AM-PM, harmonic distortion characters and Large-signal reflection coefficient by comparing the PHD model with classical S parameter model.The research of RF large signal behavior and frequency modeling is a hot issue at present and many scholars all over the world are devoted themselves to exploring it in different directions. With further research and experiments, we are sure that more researchers will discover its advantages and values.