Optimal Design of CMOS Low Noise Amplifiers via Geometric Programming

An automated and optimal design of CMOS low noise amplifier (LNA) via geometric programming has been presented. All the LNA design specifications are formulated as equality and inequality constrains and the design goal is formulated as an objective function. Some approximations and estimations on the design parameters are made in order to satisfy the formal requirement of the geometric programming. As a result, a LNA design problem is converted to a geometric programming (convex optimization) problem, which can be solved very efficiently, and a global optimal can be found. Trade-off analysis, sensitivity analysis, and robust design can easily be performed via geometric programming, and the design specification verification can also be easily solved as a feasibility problem in geometric programming. An example design of source inductively degenerated, common source, single-ended LNA is presented in this report. And the results calculated from geometric programming and simulations are compared. The comparisons show that the results are pretty close.