A technique for the fast and efficient modeling of complex microwave modules is investigated. In order to reduce simulation and optimisation time of spatially extended transmission line structures, a library of basic microwave components together with an approach for the decomposition and cascaded simulation of extended microwave circuits is proposed. To verify the design method, simple test structures as well as constituents of challenging low temperature co-fired ceramics (LTCC) modules from an ongoing R&D project are analysed and evaluated. The promising results encourage further efforts to exploit the discussed method for automated design approaches.
I. INTRODUCTIONThe complexity and functionality of microwave circuits and modular packaging techniques is growing continuously. Low temperature co-fired ceramics (LTCC) are a well-known technology for compact and cost-effective hybrid-integration of microwave modules up to high GHz frequencies [1]. In the public R&D project KERAMIS (ceramic microwave circuits for satellite communications), several components for Kaband multimedia satellite applications (17-22 GHz) have been developed by the industrial-academic project consortium [2]. Among these components is a switch matrix module that consists of six LTCC layers and incorporates truly threedimensional microwave circuitry, which is quite challenging from a simulation and optimisation point of view [3], [4].In particular for extended three-dimensional transmission line structures, electromagnetic field simulation can be very time-consuming. Considering a finite-difference time-domain (FDTD) solver, a rough estimation can be made: Given the simulation of a simple transmission line as a reference, the simulation of exactly the same, but twice as long transmission line requires that the computational domain, i.e. the number of calculated field vectors, as well as the signal propagation time are doubled. Assuming the same precision for both simulations, the second run will need twice the working memory and at least four times the computation time of the first run. As can be concluded from Fig. 1, this leads to a nonlinear increase of the computational effort for transmission line structures of increasing length and gets even worse if the computational domain is expanded in more than one dimension or if resonant structures are considered.On the other hand, microwave circuits may consist of a limited number of different though repeatedly used transmission line geometries and transition elements. Moreover, results of prior design stages or appropriate circuitry from other projects are usually available and could be incorporated in advanced