Detecting variations of small-signal equivalent-circuit model parameters in the Si/SiGe HBT process with ANN

To capture variations in the Si/SiGe HBT process characteristics, we could extract a complete equivalent-circuit model for each device, but this would be a time-consuming process. In this article, we develop an alternative approach based on an artificial neural network (ANN). To keep the complexity of the ANN low, we limit this mapping to the most sensitive elements by utilizing sensitivity analysis on a reference device. The results show that the ANN predicts the model parameters very well. (C) 2004 Wiley Periodicals, Inc.