A Simple and Accurate Modeling Method of Channel Thermal Noise Using BSIM4 Noise Models

Abstract

This article presents a method for extracting a channel thermal noise (Sid) using analytical equations and measured noise parameters. Among the four noise parameters considered (R-n, G(opt), B-opt, and NFmin), the equations of R-n and NFmin were used to extract S-id, since the S-id extracted from G(opt) and B-opt was affected more by measurement variations. NTNOI, the noise enhancement factor of the charge-based thermal model of a Berkeley short-channel IGFET model 4 (BSIM4), was modeled as a gate-bias-dependent function to reproduce S-id extracted from the measured noise parameters. Regarding the accuracy of the proposed model, while the error rate was found to vary slightly depending on the device dimensions (finger number), the new model showed only 4% error for r(n), compared to the default model's error of 41% obtained with NTNOI fixed at 1. In addition, the accuracy of G(opt), B-opt, and NFmin is substantially higher than that of the default model. Finally, the modeling technique using the holistic model, another BSIM4 thermal noise model, was briefly discussed.