The Substrate Noise Coupling For Deep-submicron CMOS Mixed-Signal ICs

With the continued scaling of CMOS processes, the performance of SOC is degraded much more than before by the substrate noise coupling(SNC). At present, the research of the substrate noise coupling is being extensively concerned. With the minimization of the transistor size and increasing integration, substrate noise coupling becomes an unignoreable problem in IC designs, and especially recently, substrate noise coupling begins to enter the field of pure digital circuits from mixed-signal circuits.In this dissertation, the coupling phenomena and modeling of substrate noise are systemically analyzed and studied. The scalable macromodel is developed in this paper. For the reduction of substrate noise coupling, the SNC reducing methods are discussed from the aspects of processes, layout, package and circuits.First, the mechanism of SNC are introduced from the generation, injection, transmission and absorbability of substrate noise coupling. After knowing the SNC mechanism, the mathematic formulas are then deduced. The skin effect is taken into account in these formulas. That assures the accuracy of macromodel.Next, compared with several kinds of the popular modeling methods, the scalable macromodel are developed from an academic model to an applied model. Differing from the traditional macromodel, the scalable macromodel can solve many actual problems including different contact width, different contact geometry shape and multi-contact problem. By using the 2-D device simulator, TMA-MEDICI, the validity of the modeling are confirmed.The researching aim of the work is reducing SNC and therefore the reducing technique of the SNC are presented in the end. Due to the limitations of the simulator and the rapid development of the IC designs, the scalable macromodel needs to be optimized.This paper focus on the substrate noise coupling modelling and the noise reduction method. This dissertation is useful to the engineers of mixed-signal integrated circuit.