Fast Evaluation of Analog Circuits Using Linear Programming

Although analog circuits are usually only a small part of an overall mixed-signal design, the time it takes to design these circuits is significantly longer. On the digital side, design automation and optimization tools have been around for a long time and are well established. These tools free up the designer to focus on more innovative tasks and allow for greater design productivity. The analog side, however, has yet to reach this point. There is much ongoing research in this area and many different approaches to try to mitigate the complexity of designing analog circuits. We propose an algorithm that utilizes the fast and efficient execution of linear programming to provide a solution for single objective optimization of small circuits. We show the results of this algorithm on three sample circuits with the objectives of maximizing gain and bandwidth. The results from our proposed optimization algorithm were within 10 percent of the maximum gain/bandwidth derived using traditional manual optimization methods.