Substrate Noise Coupling in Mixed-Signal Integrated Circuits: Compact Modeling and Grounding Strategies

Abstract: Integration of digital and analog circuits on the same chip is the result of the microelectronic industry's strive for low-cost, small, hand-held products. However, these mixed-signal circuits can experience interference problems; digital circuits inject noise into the substrate which is transmitted throughout the chip and received by sensitive analog circuits. This substrate noise can therefore degrade the performance of the chip. When performing noise coupling analysis, accurate substrate models are needed. Previous compact models were either based on two dimensional simulations, which is not sufficient since the substrate problem is inherently three dimensional, or required extraction of empirical parameters, which makes the models less predictable. This thesis presents accurate compact substrate models which can predict the noise coupling of integrated circuits. A physics-based modeling approach has been employed to yield scalable and predictive three dimensional models. Such models for uniformly doped substrates have been considered in detail since most models presented in the research literature are for epitaxial substrate types. Furthermore, general models for multi-layer substrates and arbitrary aggressor and victim geometries are presented as well. The substrate models have been utilized for investigating the efficiency of several substrate biasing methods, such as guard bands, guard rings, and distributed grounds. It was concluded that distributed grounding was the most effective. The performance of an active decoupling circuit has also been studied applying our substrate models. It was shown that dc grounding is equally good as active decoupling, for all reasonable values of the substrate resistivity.