| The current trend in developing highly linear multi-carrier microwave transmitter with high power efficiency, suitable for wireless communication, pressures power amplifier designers to focus on more sophisticated circuit topologies (class AB, F, Doherty and switching mode family) combined with efficient distortion compensation such as Digital Predistortion, Envelope Elimination and Restoration, LINC, etc.; Digital Predistortion is one of the most promising techniques that can recuperate linearity without worsening the efficiency of a given optimal PA. It has demonstrated the capability in reducing the spectral spreading and how adaptive correction can be achieved using high-speed arithmetic. However, its performance is greatly affected in the adaptation transition time by critical conditions such as starting point of optimization, stability and convergence rate. To address the aforementioned shortcomings this theses proposes efficient algorithms suitable for two different digital predistortion technique structures. The new methods develop an instantaneous characterization of the PA behaviors, which remove the need for complex convergence algorithms in the adaptation update step. In addition, the methods propose new look-up tables' configurations for predistortion purpose using polar representation. The first linearizer is implemented using a DSP environment to linearize an efficient 1 W class AB PA at 1,78 GHz. The second linearizer is implemented under open loop condition, supported by software-instruments connectivity, for a 20 W Class AB PA operating at 1.96 GHz. |
