Efficient Broadband Simulation Method And Modeling Technology Applied To RF Passive Devices And Antennas

With the rapid development of VLSI technology,the involved models are becoming more and more complex,and the requirements for the accuracy and speed of simulation are getting higher and higher.In computational electromagnetics,there are two common ways to solve electromagnetic field problems,analytical method and numerical method.Although the result of the analytical method is very accurate,it has great limitations.It cannot be calculated for complex devices and large-scale integrated circuits.Therefore,numerical methods have a dominant position.The numerical method is based on Maxwell's equations,which are in time domain form and frequency domain form.In the process of solving,the result of frequency domain solution is more accurate than that of time domain solution.This paper is based on the method of moments in the frequency domain to solve electromagnetic problems.The key to the method of moments is matrix vector multiplication,but when calculating large-scale targets,the number of unknowns increases and the calculation efficiency is very low.Therefore,in order to improve the computational efficiency,the multi-layer Green's function interpolation method is adopted.Its core content is to use the same-layer interpolation technology and the interpolation technology from the low-level to the high-level to interpolate the Green's function,thereby reducing the time complexity from to.Although the time complexity is reduced,the overall time spent in solving the wide frequency band is still very much.Therefore,this paper proposes two adaptive frequency scanning algorithms,which use less results to interpolate or fit the entire frequency band.Frequency response.First,based on the Lagrange matrix rational interpolation algorithm,the first round uses less sampled data for interpolation,and then judges whether the convergence error is reached,and if the standard is reached,the result is returned.On the contrary,the dichotomy algorithm is used to increase the sampling points and proceed to the second round interpolation,and then judge whether the convergence error requirements are met,so that the sampling data is continuously increased to iterate until the error requirements are met.Then,a new adaptive frequency sweep algorithm based on the pole-zero model is proposed.This method uses a rational function to approximate the calculated frequency domain response.It uses an iterative method to continuously replace the initial poles with the calculated improved poles.It quickly and accurately captures all resonance points on the RF scattering curve,combined with the pole-zero model,Obtain the function containing the finite resonance point,and then fit the frequency response.And realize the self-adaptive fitting algorithm through dichotomy algorithm.The adaptive frequency scanning algorithm uses less sampled data to obtain the response of the entire frequency band,which improves the simulation efficiency.The robustness of the above algorithm is verified through examples.Finally,the algorithm proposed above has realized adaptive frequency sweeping,and the simulation results can be obtained quickly.The equivalent circuit model of the simulation results has better guiding significance to the designer.Therefore,the frequency response obtained by the simulation of the target device using the above-mentioned adaptive frequency scanning algorithm,and then an accurate circuit model can be generated through the genetic algorithm and the Powell method,The generated circuit model can better characterize the frequency domain characteristics.At the same time,the circuit model can also verify the accuracy of the above algorithm.And through examples to prove the accuracy and effectiveness.