Research On The Technique Of Co-simulation And Optimization Extraction Of Device Model Parameters For Circuit Design

Circuit simulation is indispensable to circuit design. Parts in circuits for simulation are represented by models (equivalent circuits), so the accuracy of models and model parameters plays the key role in circuit simulation. With higher density of parts per chip and more complex circuit structure, the runtime needed in circuit simulation becomes formidable. And part models provided by circuit simulation software, such as PSpice, can not meet the demands of circuit design. Thus, according to the requirement of circuit simulation development, three problems are proposed to be resolved, including how to improve extraction accuracy of model parameters, how to model new devices and how to solve the problem relevant to large-scale circuits'low simulation efficiency. This thesis analyzes and researches these three problems. And the applicable results are obtained.Because the extraction accuracy of the model parameters decides circuit simulation accuracy, this thesis proposes the co-extraction method based on physical analysis and global optimization to improve extraction accuracy and efficiency. With the increase in component categories used in the circuits, it is impossible for the model libraries in common circuit simulation software to contain all the parts or circuit blocks. In order to solve this problem, this thesis researches how to use ABM model, one kind of Macro Model, to build devices and circuit blocks which do not exist in the present PSpice model library. In this thesis, the method for creating subcircuit model is introduced. And this thesis also summarizes the method for modeling circuits used by Simulink model. Because more complicated device models and larger circuit scale have obviously increased the cost and the run time of circuit simulation, this thesis proposes the co-simulation method based on SLPS tool. This method can fully combines high simulation accuracy of PSpice with high simulation efficiency of Simulink. Thus, under the premise that simulation accuracy is ensured, the co-simulation method can greatly improve simulation efficiency.