Optimization of short channel CMOS LNAs by geometric programming

Geometric programming is a method used to produce globally optimal circuit parameters with high computational efficiency. This method is applied to a short channel (90 nm and 180 nm) CMOS Low Noise Amplifier (LNA) design with common source inductive degeneration to obtain optimal design parameters for minimizing the noise figure. Geometric programming (GP) compatible functions have been determined to calculate the noise figure of short channel CMOS devices by taking into consideration channel length modulation and velocity saturation effects. Optimal design parameters (i.e., channel width and noise figure) from GP optimization are validated by comparing them with simulations obtained from Agilent´s Advanced Design Systems (ADS) software. Furthermore, tradeoff analyses have been performed to examine the influence of various design parameters such as quality factor and drain current on the design optimization. With the continuous downscaling of CMOS technologies, GP optimization offers high performance advantages in the optimal design of short channel CMOS LNAs.