ACM2 – A MOSFET model bridging design and simulation
- Speaker: Deni Germano Alves Neto
- email: denialves77@gmail.com
Abstract
MOSFET models in integrated circuit (IC) design typically fall into two categories: oversimplified textbook models that fail to capture real device behavior and highly complex simulation models requiring hundreds of parameters. Our FSIC 2024 presentation introduced ACM2, a design-oriented model that balances simplicity and accuracy using only five meaningful electrical parameters. This model offers a practical alternative for circuit designers and educators, particularly within the open-source ecosystem. In this talk, we extend ACM2 beyond its initial formulation by introducing a bias-dependent variation of its parameters, improving its accuracy in simulation without losing its design-oriented nature. While the threshold voltage (VT0) and saturation velocity parameter (ζ) remain constant, we allow the slope factor (n), specific current (IS), and DIBL parameter (σ) to vary with the bias point. This refinement enables ACM2 to capture short-channel effects and more accurately match real device behavior across various operating conditions. By incorporating these variations, ACM2 evolves from a pure design model into a unified model that bridges the gap between hands-on design and circuit simulations. Unlike traditional compact models, which obscure transistor physics with empirical fitting, ACM2 maintains physical insight while offering sufficient accuracy for both design and simulation. Implemented in an open-source simulator and validated with the IHP SG13G2 PDK, ACM2 provides a much-needed link between open-source design methodologies and practical IC implementations. This work strengthens the role of ACM2 as a paramount tool for both newcomers and experienced designers.
Downloads
General information
- Repository: https://github.com/ACMmodel/MOSFET_model <<< Slides in here
- Reference paper: D. Germano Alves Neto et al., "Design-Oriented Single-Piece 5-DC-Parameter MOSFET Model," in IEEE Access, vol. 12, pp. 87420-87437, 2024, doi: 10.1109/ACCESS.2024.3417316.