Global extraction of MOSFET parameters using the EKV model: Some properties of the underlying optimization task

We consider global extraction of parameters of the MOSFET structure with the use of the EKV model. The extraction is defined as a task of minimizing the approximation error which is the discrepancy between the measured and model-generated I–V curves. As a result of the minimization, a set of parameter values is obtained where the minimum is observed. These values are treated as the extraction results. In this study we assume certain values of MOSFET parameters, we generate I–V curves, then we treat them as is they were measurements and try to perform extraction. Thus we are able to judge about the quality of results. We study the structure of local minima of the approximation error. We show that the error function has many local minima, and the global minimum appears for true parameter values. We investigate the relation between the approximation error and the error of extracted parameter values. We also show how measurement errors introduce ambiguities in the extraction results.