Ltcc Microwave Stack Filter Design Method Based On Artificial Neural Network Research

The low temperature co-fired ceramics (LTCC) technology have advantages over other traditional packing technologies for its excellent high frequency characteristics, low resistance metallization. The impetus in the LTCC industry for manufacturability driven by fast design to meet time to market, coupled with ever-increasing LTCC devices complexities and operating frequencies, demands powerful computer aided design methodologies. Modeling and optimization still remains a major bottleneck, as it does in microwave circuits, for efficient design of LTCC microwave devices and circuits. The primary research contents and results are presented as follows:1. An efficient multiple training data generation radial basis function networks optimization technique was presented, aimed at fast design of LTCC devices, by directing the bulk of intensive computing to radial basis function networks. The advantage of this technique was demonstrated through practical design example of a LTCC multi-layer microwave low-pass filter with compact size and excellent frequency property.2. High quality factor (Q) inductors have been a goal for years in RF and microwave fields. In this thesis, a high Q (57) 40nH inductor was designed. The inductor is integrated in the filter without counteraction of magnetic flux between the spiral lines, which is primary reason of decreasing quality factor of the spiral inductors. A configuration of triple-capacitors was designed, with capacitance 7.36pF, 1.78pF and 7.36pF respectively.3. As an example, an Inverse-Chebyshev filter with cut-off frequency at 400MHz was designed through multiple training data generation radial basis function networks optimization technique. The filter have size of 2.0mm×1.2mm×1.0mm and achieve 20dB attenuation between 700MHz and 1600MHz.