A Research And Application Of Analog-digital Mixed Circuit Simulation Method

Although typically circuits are classified into analog circuits and digital ones, from a perspective of voltages and currents, all the circuits are essentially analog circuits. Despite of this analog nature, with the increase of the integration level in integrated circuit, it costs a huge amount of time to carry out pure analog simulation, although it can achieve high computing accuracy. In the modern methods of circuit simulation, digital function parts are usually extracted and modeled at higher abstract level thus the whole circuit is simulated in a mixed way to reduce the simulation time. This thesis mainly focuses on the optimization of mixed signal circuit simulation.This thesis firstly conducted a literature review of Computer Aided Design, Electronic Design Automation and the mixed circuit simulation. And then, the basic idea of simulation algorithms used in this thesis is presented. Finally, the simulation methods of traditional analog circuits, digital circuits and mixed circuits are described.A unified simulation method of mixed circuit is proposed in this thesis, aiming at the mixed circuit simulation. Analog circuits and digital ones are processed in the same simulation environment to reduce the time of circuit simulation. In circuit simulation, schematic is represented by MOSFET netlist in the beginning, then FROSTY algorithm is applied to extract digital part from the netlist, and corresponding digital circuit models are built. Finally simulation is accomplished with unified algorithm, and one output file is generated. All the above processes are carried out in one program, which avoids the synchronization between analog and digital simulations.The proposed method is verified in this thesis. An oscillator circuit and a 4-bit counter are designed and simulated to verify the simulation performance of the proposed method. Experiment results demonstrate that the method proposed in this thesis can reduce the simulation time effectively without much loss of simulation accuracy.