Tian-Wei Huang

The implementation of time-domain diakoptics in the FDTD method

The time-domain diakoptics is implemented in the finite-difference time-domain (FDTD) method with two types of connecting interfaces: i) directional interface (TLM-type), and ii) total-field interface (FDTD-type). The FDTD-type interface provides a more efficient way to realize time-domain diakoptics than TLM, especially for device optimization problems. To emulate the TLM-type interface in FDTD, two novel algorithms are developed in this paper. One is to implement an ultra-wideband absorbing boundary on the excitation plane during excitation. The other is to separate directional waves without the knowledge of incident waves. For a large circuit with cascaded modules, sequential and parallel algorithms are provided to connect them. With these connecting algorithms, time-domain diakoptics is one candidate method to realize modular and parallel computation in FDTD simulations. The validity of these algorithms is confirmed by comparison with simulated results from Microwave SPICE. >

Efficient modes extraction and numerically exact matched sources for a homogeneous waveguide cross-section in a FDTD simulation

Using the orthogonality of modes, an efficient real-time modes extraction is available for the finite-difference time-domain (FDTD) simulation of a waveguide. This paper demonstrates that, even at one cell beyond the discontinuity, S-parameters still can be accurately extracted. Combined with numerically exact matched sources, the computational volume for S-parameter computation can be reduced to a minimum.<<ETX>>

Microwave structure characterization by a combination of FDTD and system identification methods

Microwave structure is characterized by application of the system identification (SI) technique to the finite-difference-time-domain (FDTD) algorithm. The parameters of a deterministic autoregressive moving-average model (ARMA) are computed recursively such that the model output matches the FDTD simulation. The ARMA model parameter convergence is rapid, and provides savings in the computation time.<<ETX>>

The FDTD diakoptics method

A finite-difference-time-domain (FDTD) diakoptics method is developed. Sequential and parallel algorithms are provided to connect cascaded segments and to realize the modular computation of large circuits. When the large circuit is modified, only a few segments need to be recalculated, while the mutual interaction is preserved. This means that the repetitive computation of large structures is avoided and the modular computation of large circuits is realized. The proposed methods are applied to a one-dimensional case to confirm their validity by comparison with the results from microwave SPICE (MWSPICE).<<ETX>>

The influence of metallization thickness on the characteristics of cascaded junction discontinuities of shielded coplanar type transmission line

A full-wave analysis based on the mode-matching technique is applied to analyze cascaded junction discontinuities of coplanar-type transmission lines, coplanar waveguide (CPW) and finline. Results for a CPW-finline transition, a shielded CPW gap and a symmetric notch incorporating the finite metallization thickness effect are presented. The influence of metallization thickness on the coupling effect exhibited by cascaded junction discontinuities is also presented and discussed. >

Fast sequential FDTD diakoptics method using the system identification technique

The sequential finite-difference time-domain (FDTD) diakoptics method is accelerated by a system identification technique. The time-domain convolution in diakoptics is replaced by a difference equation with a small number of coefficients. Hence, the memory size is also reduced. The validity of this method is confirmed by comparison with simulated results from Microwave SPICE. The accuracy of this fast algorithm is also demonstrated by the analysis of cascaded circuit modules.<<ETX>>

The FDTD wide-band absorbing boundary conditions for the excitation plane

A numerically exact absorbing boundary condition (ABC) for the finite-difference time-domain (FDTD) method is developed. This ABC is independent of the angle, or the dispersive properties of an incident wave. Applying this ABC on the excitation plane, the reflected wave can be absorbed during the excitation. Hence, the size of the computational volume can be reduced to a minimum. One-dimensional and two-dimensional cases are used to confirm the validity of this algorithm.<<ETX>>

Application of system identification technique to FDTD and FDTD Diakoptics method

The system identification technique is applied to the FDTD Diakopties method. The computational time and computer storage are significantly reduced. This technique is successfully used in the FDTD Diakoptics simulation of active microwave circuits.

Time Domain Characterization of Active Microwave Circuits — The FDTD Diakoptics Method

A finite-difference time-domain (FDTD) Diakoptics method is developed in this paper. A two overlapped cells boundary is proposed for implementation of this method. The numerical dispersion error of FDTD in finding the impulse response is circumvented by a deconvolution method with a band-limited excitation. The transient responses of several linear and nonlinear active circuits are computed and compared with simulated results from Microwave SPICE (MWSPICE).

Efficient FDTD wideband matched sources for a homogeneous waveguide cross-section

A finite difference time domain (FDTD) wideband matched source, for waveguide problems, is developed by the FDTD diakoptics method. The efficiency of this matched source is obtained by using the modal amplitude, instead of the 2-D transverse field, to represent the incident and reflected waves. The performance of this matched source is demonstrated by the S-parameter calculation of an inductive iris and a back short. Comparisons with other matched sources with first-order absorbing boundary condition (ABC) and superabsorbing conditions are also provided.