Optimization of RF Low Noise Amplifier Design Using Analytical Model and Genetic Computation

A. Papadimitriou, M. Bucher (Tech. Univ. of Crete, Greece)

This paper introduces a methodology for optimum low noise amplifier (LNA) design, using the combination of analytical model and genetic algorithms. The optimum operation is defined by a figure of merit (FoM) that considers both the power efficiency and the RF performance of the device. This FoM suggests that the optimum region of operation lies in the moderate inversion. This knowledge can be used as a strong starting point for the design and optimization procedures. Initially, the LNA component values are extracted using the analytical model. The model does not fully take into consideration the parasitic behavior of the components in a real design, thus it produces an approximation of the optimum design. The final LNA design is generated through the use of a genetic algorithm that uses the aforementioned approximation as an input, aiming in faster convergence. To demonstrate the effectiveness of the methodology a 5 GHz common source LNA with inductive degeneration has been designed using a standard 90nm CMOS process.