Design And Modeling For Microwave LTCC Filter

LTCC(Low Temperature Co-fired Ceramic) technology which has a multi-layer high-density packaging, can be embedded passives, high-density conductive wiring capacity, excellent high frequency performance characteristics, has become the mainstream passive devices integration technology. Filter is not only the most commonly used components in microwave system, but also the most complex and varied structure microwave passive components, which restricts the overall performance and miniaturization of the microwave system. The improvement of LTCC technology spurs the new ideas of the filter design. Meanwhile, how to improve the design efficiency of LTCC components and building a practical model library became the LTCC current research topic. In this paper, taking the advantage of LTCC 3D structure, we designed several modified-form filters in LTCC, and made research of modeling for microwave passive components such as filters in LTCC.This paper first analyzed the filter design method based on coupling matrix. Then combined with electromagnetic simulation, designed three modified-form LTCC multilayer stripline filters, and one stacked cavity filter which can minimize the size of the filter, analyzed the general way of filter optimization. Finally, we produced the LTCC filters, excellent performance of its test results proved the feasibility of proposed LTCC filter structure and design methods in this paper.Meanwhile, artificial neural network(ANN) technology is used for LTCC filter modeling. We established an LTCC filter neural network reverse model which has a electrical parameters for the filter input and external physical dimensions for the filter output. The model used knowledge-based modeling approach, combined with the coupling coefficient matrix, is divided into three sub-structure according to the filter coupling mechanism. Reverse neural network technology is applied to develop filter sub-models, respectively. Finally, we verify the performance of each sub-model and the overall model of LTCC filters. The good results proved the feasibility of appling reverse neural network model to the filter modeling in LTCC.