Research On Numerical Method Of Passive Intermodulation

As the transmit power is increasing and receiver with high sensitivity are used more widely than before, the generation of passive intermodulation (PIM) which seriously affect satellite performance are hindering the development of wireless satellite communication system. However, the phenomenon of PIM are affected by so many factors that no reliable models exist to predict this phenomenon. Up to now, experimental test is the most common way to research PIM and to numerically analyze the PIM is still in start stage. In this paper, we use Time-Domain Finite Integration Technology (TDFIT) analyze the phenomenon of PIM caused by metal-to-metal (MM) contacts.Firstly, according to the physical mechanisms of MM contact and the field-circuit coupling thought used in analyzing electromagnetics problem, we propose a reasonable calculation model for PIM by using distributive lump elements. In the meantime, we detail in solving the loading problem of lump elements and also provide solution to extract voltage for them. In order to solve the nonlinear equation that appears during the process of analyzing PIM, we investigate various kinds of newton method and use typical functions to test their actual calculation efficiency. Based on the test result and the special need of solving the nonlinear equation, we propose an efficient and robust solution for solving the nonlinear equation.In order to analyze the phenomenon of PIM happening in waveguide structure, we develop the technology for waveguide port excitation. By analyzing both the structure of waveguide port in TDFIT and the field distribution of waveguide, we propose a mode-modulation method for waveguide port excitation. In this method, we use finite element method to obtain the steady field distribution of waveguide first, then by modulating the field of wave guide port, we can effectively excite the waveguide. Finally, we use an example test our solutions which prove the validity of this method.In this paper, we also use the GPU to accelerate the TDFIT in order to reduce the simulation time and achieve a more efficient analysis. Firstly, we learn the programming model of GPU and summary several basic strategies used for GPU parallel computing by investigating the architecture of GPU. Then, by analyzing the serial process, we propose a basic solution for parallelization of TDFIT. Finally, we optimize the memory throughput by utilizing the share memory, which further improve the computation performance.