High Dimensional Noise and Power Optimization of a Three-Stage Amplifier with Feedforward Reversed Nested Miller Compensation

Abstract

The design and optimization of a three-stage operational transconductance amplifier are discussed in this paper. Reversed nested-Miller compensation with feedforward transconductance is used to satisfy the stability requirements. A systematic gm/ID-based procedure that uses pre-computed lookup tables is used to take all the circuit degrees of freedom into account and solve a high-dimensional non-linear optimization problem. Exact symbolic expressions are obtained by a symbolic solver for both the objectives and constraints. Two optimization problems are formulated: A single-objective optimization problem for minimizing the noise and a multi-objective optimization problem for minimizing both noise and power consumption. The proposed optimization flow is very fast and takes less than 20 s on a standard computer. Moreover, the design space is visualized using one million design points to illustrate the design trade-offs and the multi-objective Pareto-optimal front. The synthesis results are compared with Cadence Spectre simulation results showing very good agreement.