A new SQP based space-mapping algorithm for on-chip spiral inductor optimization

Computer aided design (CAD) and optimization of on-chip spiral inductors is of immense interest to RFIC designers. Traditional optimization methods rely on approximate/fast "coarse models" of spiral inductors, eg. equivalent circuits, empirical equations, etc. Naturally, the accuracy/validity of optimized results from such methods depends highly on the quality of the coarse models used. Further, the inability of such models to account for full-wave electromagnetic (EM) effects, limits their application to low-frequency circuit designs and/or necessitates additional efforts to formulate an acceptable optimization objective. The pressing need for first-pass success of CAD tools in terms of inductor design mandates consideration of full-wave EM effects within the CAD framework. Space-mapping optimization combines the speed of circuit models and the accuracy of EM simulations. In this paper, we propose a new space-mapping optimization algorithm, which exploits the sequential quadratic programming (SQP) technique for coarse model optimization. Starting with the user-specifications of the spiral inductor, the proposed algorithm yields an acceptable spiral inductor design/layout for RFIC design. The algorithm is illustrated via an on-chip spiral inductor optimization example.