Modeling Of Substrate Coupling Noise In Lightly-doped Mixed-signal ICs

As the widely application of System on Chip (SoC), substrate coupling noise hasbecome one of the most important issues in mixed-signal ICs’ design. The sensitiveanalog/RF circuits are surrounded by a great lot of digital circuits which are fabricatedwith each other on the same die. The parasitical current which are generated fromdigital circuits will be injected to the substrate in various ways, thus they can transmit tothe sensitive modules through substrate, which can deteriorate performance ofanalog/RF circuits, even cause failure of the whole chips.In this dissertation, the mechanisms of injection, transmission and picking up ofthe coupling substrate noise are studied and explicated. A3D circuit level model oflightly-doped substrate which is applicable in analyzing mixed-signal IC is proposed inthis dissertation. This model is a network with distributing resistance and capacitance,and is simplified according to the complexity of simulation. The accuracy of the modelis verified by four-probe method in surveying resistivity, the coupling noise from thedistributing network of power supply and ground is also taken into account, meanwhile,the path of the model is established. The noise from digital switch is modeled which isequivalent to a current pulse source with noise injection to substrate. These models arethen applied to some certain circuits in SPECTRE simulation, to analyze the influencethe coupling noise in substrate brings about.The ultimate aim of modeling substrate is to eliminate and suppress such noise inthe process of circuit design. In this dissertation, several isolated structures in SMIC0.18μm technology are modeled and simulated, respectively, including the single guardring isolation, double guard rings isolation and deep N-well isolation. An active noisesuppression amplifier is proposed, which can improve the effect of suppression andrange for guard ring isolation. At last, the affectivities of these suppression technologiesare compared and quantitatively analyzed to draw a general conclusion.