Raphael Chatzipantelis, Loukas Chevas, Nikolaos Makris and Matthias Bucher
Noise Propagation and Statistic Variability in MOSFETs Using Probability Density Functions
Fluctuation and Noise Letters (Accepted Paper)
DOI: 10.1142/S0219477525400255
1) School of Electrical and Computer Engineering, Technical University of Crete, Chania 73100, Greece
2) Foundation for Research and Technology Hellas, Heraklion 70013, Greece,
Abstract: Probability density functions using stochastic methods are shown to be an effective tool in the context of MOSFET noise and variability modeling. These methods are employed here in the context of the charge-based EKV MOSFET model. As an example, a Gaussian noise density function applied at the gate or the source of a MOSFET causes a corresponding drain current noise density function, which may be expressed analytically as a function of inversion coefficient only. The same expression may be used to model drain current variability due to MOSFET parameters such as threshold voltage. Furthermore, the method is extended to variations of quantities such as transconductance and transconductance-to-current ratio. The method shows promise in variability modeling of MOSFETs and may complement traditional approaches.
FIG: Comparing the traditional ”small-signal” transconductance method with the stochastic PDF method for 𝑖𝑓, derived from the charge-based model, where in both cases the same noise (or variability) at the gate is applied (𝑉𝐺= 87mV, 𝜎𝑉𝐺=10mV), centered at 𝑖𝑓=2, showing slightly different mean and ±3𝜎 values, while the tail distributions differ significantly.
Acknowledgements: The authors gratefully acknowledge Dr. Predrag Habas from EM Microelectronic S.A. for valuable discussions and wafers for noise and statistical analysis. This work was partly funded by the European Union, and by Greek National funds, under the topic DIGITAL- Chips-2024-SG-CCC-1 - Competence Centers, Project No 101217803 - HCCC.
