Measurement And Modeling Of Passive Devices And Development Of IBIS Model Optimization Tool

With the increasing clock frequency,the signal integrity(SI)problem of high-speed digital circuits is becoming more and more serious.The simulation of signal integrity has gradually become an inevitable means to realize high-speed digital circuit design.The simulation is extremely important to ensure the performance and stability of high-speed digital systems.An accurate simulation model is an important foundation for ensuring the accuracy and reliability of the simulation and analysis of signal integrity.This thesis mainly studies the measurement and modeling of the S-parameter model of the passive device and the optimization of the IBIS model of the active device.For the S-parameter model of passive devices,by analyzing various errors in the measurement of S-parameters of passive devices using vector network analyzers(VNA),and comparing various calibration processing methods and calibration techniques,this thesis proposes a method for measuring S-parameters of passive components based on TRM calibration,and designs and manufactures PCB test fixtures and calibration parts suitable for measuring surface-mount passive components of different packages in a two-port series connection.In addition,the quality of the test fixture and calibration parts are investigated through actual tests.Next,the fitting method of the measurement data of the passive device is explained and its S-parameter model is established.First,the measured S-parameters of the passive components are converted to Z-parameters.Compared with other fitting methods,the vector fitting method has the advantages of high fitting accuracy,stable operation,and fast speed.This thesis uses vector fitting method to fit the self-impedance parameters of passive devices to obtain the continuous impedance function of passive devices.Then take the capacitor as an example to compare the fitted measurement data with the factory impedance parameters provided by the capacitor manufacturer to verify the accuracy of the modeling method of first measurement and then fitting.Finally,the fitted Z parameters are inversely converted to obtain S parameters,and then the S-parameter model of the passive device is obtained.For the IBIS model of active devices,this thesis proposes an IBIS model optimization method based on intelligent automatic fitting and accuracy improvement of linear and nonlinear regions.This method first divides the linear region and the nonlinear region by the slope transformation rate of the original data of the model;Then the data in the linear region is fitted to obtain a linear fitting curve,and then uniform sampling is used to obtain more accurate linear region data with fewer data points.For the data in the non-linear region,the radial basis function neural network(RBF-NN)modeling method is used to fit the data,and more accurate data in the non-linear region can be obtained with more data points;Finally,the data of different sources and types are merged and spliced to obtain the optimized IBIS model.Next,use the MATLAB App Designer tool to integrate the corresponding algorithm of the optimization method into a human-computer interaction interface,design and develop an IBIS model optimization tool,as the expression form of the research results of the IBIS model optimization of this subject,and briefly introduced the interface and function of the optimization tool.Finally,taking ACTEL's anti-fuse FPGA chip A40MX04-3PL44 I as an example,the optimized IBIS model was verified in three parts,namely,syntax verification,basic simulation verification,and comparison verification between simulation and actual measurement.Among them,the actual circuit board was produced in the comparison and verification of simulation and actual measurement,and the IBIS model before and after optimization was used for simulation to obtain the time domain output waveform at the test point.Then,they were compared with the real-time output waveforms measured at the actual measurement points using the oscilloscope to verify the feasibility and correctness of the IBIS model optimization method proposed in this thesis.