Research On Electromagnetic Simulation And Planar Near Field Scanning Algorithm Based On Chip Package Model

In recent years,with the continuous innovation and progress in information technology,automation,and intelligence,the requirements for chip computing power in areas such as autonomous driving,the Internet of Things,5G communications,and high-performance computing have become higher and higher,resulting in the complexity of semiconductor chips,R&D and manufacturing costs have also risen significantly,bringing many technical challenges.In a complex multi-physics environment,accurate modeling of chips and chip packages,as well as accurate analysis of interconnection transmission lines and electrical characteristics of chip package structures are crucial.In addition,it is also necessary to fully consider the electromagnetic interference problem that affects the normal operation of the entire integrated circuit,including the electromagnetic interference problem within the integrated circuit system and between systems.Therefore,it is of great significance to study the electromagnetic compatibility of chip packaging and the near-field scanning of integrated circuits.Based on the high-frequency electromagnetic theory of chip packaging,this thesis discusses the general method of constructing typical chip packaging,and effectively and accurately builds the physical model of QFP chip packaging.Based on this model,a full-wave electromagnetic simulation is completed using the ANSYS electromagnetic simulation platform,which verifies that the built QFP chip package model has better signal transmission quality.At the same time,the specific process of using this method to construct a typical BGA chip packaging model is described,and it is further demonstrated that the chip packaging model construction method discussed has strong applicability.Aiming at the problem that the near-field scanning technology faces the problem that the scanning time is too long in the electromagnetic interference diagnosis,which seriously affects the reconstruction efficiency of the equivalent radiation source model,this thesis proposes a method for near-field scanning The two-stage plane adaptive sampling algorithm,which further optimizes the near-field scanning sampling process,significantly reduces the number of near-field samples required to solve the radiation source model through uniform and non-uniform dualstage sampling methods,and can be based on the near-field Fluctuation controls the density of sampling,achieving the effective balance of nearfield sampling efficiency and accuracy.In addition,since the accuracy of the near-field scanning data affects the reconstruction accuracy of the radiation source equivalent reconstruction algorithm,combined with the performance of the two-stage planar adaptive sampling algorithm,a high-precision interpolation method based on the generalized regression neural network model is proposed.Through Compared with the actual simulation,it is verified that the interpolation method further improves the prediction accuracy of near-field data than the BP neural network interpolation method used in the two-stage planar adaptive sampling algorithm.