BLADES: an artificial intelligence approach to analog circuit design

An expert-systems-based automated design approach for analog circuits is presented. The approach uses both formal and intuitive knowledge in the design process. A prototype design environment, BLADES, which uses a divide and conquer solution strategy, has been successfully implemented and is currently capable of designing a wide range of subcircuit functional blocks as well as a limited class of integrated bipolar operational amplifiers. BLADES is believed to be the first successful design expert system in the analog design domain. It uses different levels of abstraction depending on the complexity of the design task under consideration. The importance of the abstraction level lies in the fact that once design primitives are defined, the problem of extracting the knowledge (design rules) become less complex. Two design examples are given to demonstrate the viability and versatility of the knowledge-based design technique as an analog design tool. None of the circuits designed and tested using BLADES were unstable. >